Integrative Multimodal Metabolomics to Early Predict Cognitive Decline Among Amyloid Positive Community-Dwelling Older Adults

Alzheimer's disease is strongly linked to metabolic abnormalities. We aimed to distinguish amyloid-positive people who progressed to cognitive decline from those who remained cognitively intact. We performed untargeted metabolomics of blood samples from amyloid-positive individuals, before any sign of cognitive decline, to distinguish individuals who progressed to cognitive decline from those who remained cognitively intact. A plasma-derived metabolite signature was developed from Supercritical Fluid chromatography coupled with high-resolution mass spectrometry (SFC-HRMS) and nuclear magnetic resonance (NMR) metabolomics. The 2 metabolomics data sets were analyzed by Data Integration Analysis for Biomarker discovery using Latent approaches for Omics studies (DIABLO), to identify a minimum set of metabolites that could describe cognitive decline status. NMR or SFC-HRMS data alone cannot predict cognitive decline. However, among the 320 metabolites identified, a statistical method that integrated the 2 data sets enabled the identification of a minimal signature of 9 metabolites (3-hydroxybutyrate, citrate, succinate, acetone, methionine, glucose, serine, sphingomyelin d18:1/C26:0 and triglyceride C48:3) with a statistically significant ability to predict cognitive decline more than 3 years before decline. This metabolic fingerprint obtained during this exploratory study may help to predict amyloid-positive individuals who will develop cognitive decline. Due to the high prevalence of brain amyloid-positivity in older adults, identifying adults who will have cognitive decline will enable the development of personalized and early interventions.

Supervised contrastive learning enhances graph convolutional networks for predicting neurodevelopmental deficits in very preterm infants using brain structural connectome

Very preterm (VPT) infants (born at less than 32 weeks gestational age) are at high risk for various adverse neurodevelopmental deficits. Unfortunately, most of these deficits cannot be accurately diagnosed until the age of 2-5 years old. Given the benefits of early interventions, accurate diagnosis and prediction soon after birth are urgently needed for VPT infants. Previous studies have applied deep learning models to learn the brain structural connectome (SC) to predict neurodevelopmental deficits in the preterm population. However, none of these models are specifically designed for graph-structured data, and thus may potentially miss certain topological information conveyed in the brain SC. In this study, we aim to develop deep learning models to learn the SC acquired at term-equivalent age for early prediction of neurodevelopmental deficits at 2 years corrected age in VPT infants. We directly treated the brain SC as a graph, and applied graph convolutional network (GCN) models to capture complex topological information of the SC. In addition, we applied the supervised contrastive learning (SCL) technique to mitigate the effects of the data scarcity problem, and enable robust training of GCN models. We hypothesize that SCL will enhance GCN models for early prediction of neurodevelopmental deficits in VPT infants using the SC. We used a regional prospective cohort of ∼280 VPT infants who underwent MRI examinations at term-equivalent age from the Cincinnati Infant Neurodevelopment Early Prediction Study (CINEPS). These VPT infants completed neurodevelopmental assessment at 2 years corrected age to evaluate cognition, language, and motor skills. Using the SCL technique, the GCN model achieved mean areas under the receiver operating characteristic curve (AUCs) in the range of 0.72∼0.75 for predicting three neurodevelopmental deficits, outperforming several competing models. Our results support our hypothesis that the SCL technique is able to enhance the GCN model in our prediction tasks.

Early prediction of gestational diabetes mellitus using maternal demographic and clinical risk factors

OBJECTIVE

To build and validate an early risk prediction model for gestational diabetes mellitus (GDM) based on first-trimester electronic medical records including maternal demographic and clinical risk factors.

METHODS

To develop and validate a GDM prediction model, two datasets were used in this retrospective study. One included data of 14,015 pregnant women from Máxima Medical Center (MMC) in the Netherlands. The other was from an open-source database nuMoM2b including data of 10,038 nulliparous pregnant women, collected in the USA. Widely used maternal demographic and clinical risk factors were considered for modeling. A GDM prediction model based on elastic net logistic regression was trained from a subset of the MMC data. Internal validation was performed on the remaining MMC data to evaluate the model performance. For external validation, the prediction model was tested on an external test set from the nuMoM2b dataset.

RESULTS

An area under the receiver-operating-characteristic curve (AUC) of 0.81 was achieved for early prediction of GDM on the MMC test data, comparable to the performance reported in previous studies. While the performance markedly decreased to an AUC of 0.69 when testing the MMC-based model on the external nuMoM2b test data, close to the performance trained and tested on the nuMoM2b dataset only (AUC = 0.70).

VAP risk index: Early prediction and hospital phenotyping of ventilator-associated pneumonia using machine learning

BACKGROUND

Ventilator-associated pneumonia (VAP) is a leading cause of morbidity and mortality in intensive care units (ICUs). Early identification of patients at risk of VAP enables early intervention, which in turn improves patient outcomes. We developed a predictive model for individualized risk assessment utilizing machine learning to identify patients at risk of developing VAP.

METHODS

The Philips eRI dataset, a multi-institution electronic medical record (EMR), was used for model development. For adult (≥18y) patients, we propose a set of criteria using indications of the start of a new antibiotic treatment temporally contiguous to a microbiological test to mark suspected infection events, of which those with a positive culture are labeled as presumed VAP if 1) the event occurs at least 48 h after intubation, and 2) there are no indications of community-acquired pneumonia (CAP) or other hospital-acquired infections (HAI) in the patient charts. The resulting VAP and no-VAP (control) cases were then used to build an ensemble of decision trees to predict the risk of VAP in the next 24 h using data on patients' demographics, vitals, labs, and ventilator settings.

RESULTS

The resulting model predicts the development of VAP 24 h in advance with an AUC of 76 % and AUPRC of 75 %. Additionally, we group hospitals that are similar in healthcare processes into distinct clusters and characterize VAP prediction for the identified hospital clusters. We show inter-hospital (teaching status and healthcare processes) and cohort-specific (age groups, gender, early vs late VAP, ICU mortality status) differences in VAP prediction and associated symptomologies.

CONCLUSIONS

Our proposed VAP criteria use clinical actions to mark incidences of presumed VAP infection, which enables the development of models for early detection of these events. We curated a patient cohort using these criteria and used it to build a model for predicting impending VAP events prior to clinical suspicions. We present a clustering approach for tailoring the VAP prediction model for different hospital types based on their EMR data characteristics. The model provides an instantaneous risk score that allows early interventions and confirmatory diagnostic actions.

Genetic and Environmental Influences on Structural and Diffusion-Based Alzheimer's Disease Neuroimaging Signatures Across Midlife and Early Old Age

BACKGROUND

Composite scores of magnetic resonance imaging-derived metrics in brain regions associated with Alzheimer's disease (AD), commonly termed AD signatures, have been developed to distinguish early AD-related atrophy from normal age-associated changes. Diffusion-based gray matter signatures may be more sensitive to early AD-related changes compared with thickness/volume-based signatures, demonstrating their potential clinical utility. The timing of early (i.e., midlife) changes in AD signatures from different modalities and whether diffusion- and thickness/volume-based signatures each capture unique AD-related phenotypic or genetic information remains unknown.

METHODS

Our validated thickness/volume signature, our novel mean diffusivity (MD) signature, and a magnetic resonance imaging-derived measure of brain age were used in biometrical analyses to examine genetic and environmental influences on the measures as well as phenotypic and genetic relationships between measures over 12 years. Participants were 736 men from 3 waves of the Vietnam Era Twin Study of Aging (VETSA) (baseline/wave 1: mean age [years] = 56.1, SD = 2.6, range = 51.1-60.2). Subsequent waves occurred at approximately 5.7-year intervals.

RESULTS

MD and thickness/volume signatures were highly heritable (56%-72%). Baseline MD signatures predicted thickness/volume signatures over a decade later, but baseline thickness/volume signatures showed a significantly weaker relationship with future MD signatures. AD signatures and brain age were correlated, but each measure captured unique phenotypic and genetic variance.

CONCLUSIONS

Cortical MD and thickness/volume AD signatures are heritable, and each signature captures unique variance that is also not explained by brain age. Moreover, results are in line with changes in MD emerging before changes in cortical thickness, underscoring the utility of MD as a very early predictor of AD risk.

Establishment of the early prediction models of low-birth-weight reveals influential genetic and environmental factors: a prospective cohort study

BACKGROUND

Low birth weight (LBW) is a leading cause of neonatal morbidity and mortality, and increases various disease risks across life stages. Prediction models of LBW have been developed before, but have limitations including small sample sizes, absence of genetic factors and no stratification of neonate into preterm and term birth groups. In this study, we challenged the development of early prediction models of LBW based on environmental and genetic factors in preterm and term birth groups, and clarified influential variables for LBW prediction.

METHODS

We selected 22,711 neonates, their 21,581 mothers and 8,593 fathers from the Tohoku Medical Megabank Project Birth and Three-Generation cohort study. To establish early prediction models of LBW for preterm birth and term birth groups, we trained AI-based models using genetic and environmental factors of lifestyles. We then clarified influential environmental and genetic factors for predicting LBW in the term and preterm groups.

RESULTS

We identified 2,327 (10.22%) LBW neonates consisting of 1,077 preterm births and 1,248 term births. Our early prediction models archived the area under curve 0.96 and 0.95 for term LBW and preterm LBW models, respectively. We revealed that environmental factors regarding eating habits and genetic features related to fetal growth were influential for predicting LBW in the term LBW model. On the other hand, we identified that genomic features related to toll-like receptor regulations and infection reactions are influential genetic factors for prediction in the preterm LBW model.

CONCLUSIONS

We developed precise early prediction models of LBW based on lifestyle factors in the term birth group and genetic factors in the preterm birth group. Because of its accuracy and generalisability, our prediction model could contribute to risk assessment of LBW in the early stage of pregnancy and control LBW risk in the term birth group. Our prediction model could also contribute to precise prediction of LBW based on genetic factors in the preterm birth group. We then identified parental genetic and maternal environmental factors during pregnancy influencing LBW prediction, which are major targets for understanding the LBW to address serious burdens on newborns' health throughout life.

Predicting sepsis using deep learning across international sites: a retrospective development and validation study

Background

When sepsis is detected, organ damage may have progressed to irreversible stages, leading to poor prognosis. The use of machine learning for predicting sepsis early has shown promise, however international validations are missing.

Methods

This was a retrospective, observational, multi-centre cohort study. We developed and externally validated a deep learning system for the prediction of sepsis in the intensive care unit (ICU). Our analysis represents the first international, multi-centre in-ICU cohort study for sepsis prediction using deep learning to our knowledge. Our dataset contains 136,478 unique ICU admissions, representing a refined and harmonised subset of four large ICU databases comprising data collected from ICUs in the US, the Netherlands, and Switzerland between 2001 and 2016. Using the international consensus definition Sepsis-3, we derived hourly-resolved sepsis annotations, amounting to 25,694 (18.8%) patient stays with sepsis. We compared our approach to clinical baselines as well as machine learning baselines and performed an extensive internal and external statistical validation within and across databases, reporting area under the receiver-operating-characteristic curve (AUC).

Findings

Averaged over sites, our model was able to predict sepsis with an AUC of 0.846 (95% confidence interval [CI], 0.841-0.852) on a held-out validation cohort internal to each site, and an AUC of 0.761 (95% CI, 0.746-0.770) when validating externally across sites. Given access to a small fine-tuning set (10% per site), the transfer to target sites was improved to an AUC of 0.807 (95% CI, 0.801-0.813). Our model raised 1.4 false alerts per true alert and detected 80% of the septic patients 3.7 h (95% CI, 3.0-4.3) prior to the onset of sepsis, opening a vital window for intervention.

Interpretation

By monitoring clinical and laboratory measurements in a retrospective simulation of a real-time prediction scenario, a deep learning system for the detection of sepsis generalised to previously unseen ICU cohorts, internationally.

Funding

This study was funded by the Personalized Health and Related Technologies (PHRT) strategic focus area of the ETH domain.

Early prediction of treatment outcome for lenvatinib using 18F-FDG PET/CT in patients with unresectable or advanced thyroid carcinoma refractory to radioiodine treatment: a prospective, multicentre, non-randomised study

BACKGROUND

Lenvatinib is widely used to treat unresectable and advanced thyroid carcinomas. We aimed to determine whether ¹⁸F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) performed 1 week after lenvatinib treatment initiation could predict treatment outcomes.

RESULTS

This was a prospective, nonrandomised, multicentre study. Patients with pathologically confirmed differentiated thyroid carcinoma (DTC) and lesions refractory to radioiodine treatment were eligible for inclusion. Patients were treated with 24 mg lenvatinib as the initial dose and underwent PET/CT examination 1 week after treatment initiation. Contrast-enhanced CT was scheduled at least 4 weeks later as the gold standard for evaluation. The primary endpoint was to evaluate the discrimination power of maximum standardised uptake value (SUVmax) obtained by PET/CT compared to that obtained by contrast-enhanced CT. Evaluation was performed using the area under the receiver operating characteristic (ROC-AUC) curve. Twenty-one patients were included in this analysis. Receiver operating characteristic (ROC) curve analysis yielded an AUC of 0.714 for SUVmax after 1 week of lenvatinib treatment. The best cut-off value for the treatment response for SUVmax was 15.211. The sensitivity and specificity of this cut-off value were 0.583 and 0.857, respectively. The median progression-free survival was 26.3 months in patients with an under-cut-off value and 19.7 months in patients with an over-cut-off value (P = 0.078).

CONCLUSIONS

The therapeutic effects of lenvatinib were detected earlier than those of CT because of decreased FDG uptake on PET/CT. PET/CT examination 1 week after the initiation of lenvatinib treatment may predict treatment outcomes in patients with DTC.

TRIAL REGISTRATION

This trial was registered in the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (number UMIN000022592) on 6 June, 2016.

Early prediction of remaining useful life for lithium-ion batteries based on CEEMDAN-transformer-DNN hybrid model

A reliable and safe energy storage system utilizing lithium-ion batteries relies on the early prediction of remaining useful life (RUL). Despite this, accurate capacity prediction can be challenging if little historical capacity data is available due to the capacity regeneration and the complexity of capacity degradation over multiple time scales. In this study, data decomposition, transformers, and deep neural networks (DNNs) are combined to develop a model of RUL prediction for lithium-ion batteries. Complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is used for battery capacity sequential data to account for the capacity regeneration effect. The transformer networks are leveraged to predict each component of capacity regeneration thus improving the model's ability to handle long sequences while reducing the amount of data. The global degradation trend is predicted using a deep neural network. We validated the early prediction performance of the model using two publicly available battery datasets. Results show that the prediction model only uses 25%-30% data to achieve high accuracy. In the two public data sets, the RMSE errors were 0.0208 and 0.0337, respectively. A high level of accuracy is achieved with the model proposed in this study, which is based on fewer capacity data.

Optimal machine learning methods for prediction of high-flow nasal cannula outcomes using image features from electrical impedance tomography

BACKGROUND

High-flow nasal cannula (HNFC) is able to provide ventilation support for patients with hypoxic respiratory failure. Early prediction of HFNC outcome is warranted, since failure of HFNC might delay intubation and increase mortality rate. Existing methods require a relatively long period to identify the failure (approximately 12 h) and electrical impedance tomography (EIT) may help identify the patient's respiratory drive during HFNC.

OBJECTIVES

This study aimed to investigate a proper machine-learning model to predict HFNC outcomes promptly by EIT image features.

METHODS

The Z-score standardization method was adopted to normalize the samples from 43 patients who underwent HFNC and six EIT features were selected as model input variables through the random forest feature selection method. Machine-learning methods including discriminant, ensembles, k-nearest neighbour (KNN), artificial neural network (ANN), support vector machine (SVM), AdaBoost, xgboost, logistic, random forest, bernoulli bayes, gaussian bayes and gradient-boosted decision trees (GBDT) were used to build prediction models with the original data and balanced data proceeded by the synthetic minority oversampling technique.

RESULTS

Prior to data balancing, an extremely low specificity (less than 33.33%) as well as a high accuracy in the validation data set were observed in all the methods. After data balancing, the specificity of KNN, xgboost, random forest, GBDT, bernoulli bayes and AdaBoost significantly reduced (p<0.05) while the area under curve did not improve considerably (p>0.05); and the accuracy and recall decreased significantly (p<0.05).

CONCLUSIONS

The xgboost method showed better overall performance for balanced EIT image features, which may be considered as the ideal machine learning method for early prediction of HFNC outcomes.

Cytokine Biomarker Phenotype for Early Prediction and Triage of Sepsis in Blunt Trauma Patients

BACKGROUND

Altered levels of inflammatory markers secondary to severe trauma present a major problem to physicians and are prone to interfering with the clinical identification of sepsis events. This study aimed to establish the profiles of cytokines in trauma patients to characterize the nature of immune responses to sepsis, which might enable early prediction and individualized treatments to be developed for targeted intervention.

METHODS

A 15-plex human cytokine magnetic bead assay system was used to measure analytes in citrated plasma samples. Analysis of the kinetics of these cytokines was performed in 40 patients with severe blunt trauma admitted to our trauma center between March 2016 and February 2017, with an Injury Severity Score (ISS) greater than 20 with regard to sepsis (Sepsis-3) over a 14-d time course.

RESULTS

In total, the levels of six cytokines were altered in trauma patients across the 1-, 3-, 5-, 7-, and 14-d time points. Additionally, IL-6, IL-10, IL-15, macrophage derived chemokine (MDC), GRO, sCD40 L, granulocyte colony-stimulating factor (G-CSF), and fibroblast growth factor (FGF)-2 levels could be used to provide a significant discrimination between sepsis and nonsepsis patients at day 3 and afterward, with an area under the curve (AUC) of up to 0.90 through a combined analysis of the eight biomarkers (P < 0.001). Event-related analysis demonstrated 1.5- to 4-fold serum level changes for these cytokines within 72 h before clinically apparent sepsis.

CONCLUSIONS

Cytokine profiles demonstrate a high discriminatory ability enabling the timely identification of evolving sepsis in trauma patients. These abrupt changes enable sepsis to be detected up to 72 h before clinically overt deterioration. Defining cytokine release patterns that distinguish sepsis risk from trauma patients might enable physicians to initiate timely treatment and reduce mortality. Large prospective studies are needed to validate and operationalize the findings.

TRIAL REGISTRATION

Clinicaltrials, NCT01713205. Registered October 22, 2012, https://register.

CLINICALTRIALS

gov/NCT01713205.

Prediction of Neurodevelopmental Disorders Based on De Novo Coding Variation

The early detection of neurodevelopmental disorders (NDDs) can significantly improve patient outcomes. The differential burden of non-synonymous de novo mutation among NDD cases and controls indicates that de novo coding variation can be used to identify a subset of samples that will likely display an NDD phenotype. Thus, we have developed an approach for the accurate prediction of NDDs with very low false positive rate (FPR) using de novo coding variation for a small subset of cases. We use a shallow neural network that integrates de novo likely gene-disruptive and missense variants, measures of gene constraint, and conservation information to predict a small subset of NDD cases at very low FPR and prioritizes NDD risk genes for future clinical study.

Predicting Persistent Acute Respiratory Failure in Acute Pancreatitis: The Accuracy of Two Lung Injury Indices

BACKGROUND/AIMS

Early and accurate identification of patients with acute pancreatitis (AP) at high risk of persistent acute respiratory failure (PARF) is crucial. We sought to determine the accuracy of simplified Lung Injury Prediction Score (sLIPS) and simplified Early Acute Lung Injury (sEALI) for predicting PARF in ward AP patients.

METHODS

Consecutive AP patients in a training cohort from West China Hospital of Sichuan University (n = 912) and a validation cohort from The First Affiliated Hospital of Nanchang University (n = 1033) were analyzed. PARF was defined as oxygen in arterial blood/fraction of inspired oxygen < 300 mmHg that lasts for > 48 h. The sLIPS was composed by shock (predisposing condition), alcohol abuse, obesity, high respiratory rate, low oxygen saturation, high oxygen requirement, hypoalbuminemia, and acidosis (risk modifiers). The sEALI was calculated from oxygen 2 to 6 L/min, oxygen > 6 L/min, and high respiratory rate. Both indices were calculated on admission.

RESULTS

PARF developed in 16% (145/912) and 22% (228/1033) (22%) of the training and validation cohorts, respectively. In these patients, sLIPS and sEALI were significantly increased. sLIPS ≥ 2 predicted PARF in the training (AUROC 0.87, 95% CI 0.84-0.89) and validation (AUROC 0.81, 95% CI 0.78-0.83) cohorts. sLIPS was significantly more accurate than sEALI and current clinical scoring systems in both cohorts (all P < 0.05).

CONCLUSIONS

Using routinely available clinical data, the sLIPS can accurately predict PARF in ward AP patients and outperforms the sEALI and current existing clinical scoring systems.

Predicting relapse from the time to remission during the acute treatment of depression: A re-analysis of the STAR*D data

BACKGROUND

Predicting relapse during maintenance treatment for depression is challenging. The objective of this analysis was to investigate the association between the time taken to achieve remission in the acute phase, and the subsequent relapse rate or time to relapse using the Sequenced Treatment Alternatives to Relieve Depression dataset.

METHOD

Data of 1296 outpatients with nonpsychotic depression who entered a 12-month naturalistic follow-up period after achieving remission with citalopram for up to 14 weeks were analyzed. One-way analysis of variance and the Jonckheere-Terpstra trend test were performed to compare the relapse rates and days to relapse during the follow-up period among those who achieved remission at weeks 2, 4, 6, 9, 12, and 14. Remission and relapse were defined as scores of ≤5 and ≥11, respectively, on the 16-Item Quick Inventory of Depressive Symptomatology and Self-Report.

RESULTS

The relapse rates were significantly different among those who achieved remission each week (F_{(5, 1087)} = 4.995, p < 0.001). The lowest and highest relapse rates were observed in those who achieved remission at weeks 4 (25.7 %) and 12 (42.4 %), respectively, with a significant difference (p = 0.006). There was also a significant negative trend between the weeks taken to achieve remission and the days to relapse (z = -6.13, p < 0.001).

CONCLUSIONS

Patients with depression who show a faster response to antidepressant treatment are more likely to maintain remission in the long term. This finding suggests that, to prevent relapse, close attention should be paid to patients who require a relatively long time to achieve remission.

Development of an acute kidney injury risk prediction model for patients undergoing extracorporeal membrane oxygenation

Background

Some studies have reported to use some predictors before extracorporeal membrane oxygenation (ECMO) initiation to predict the acute kidney injury (AKI) risk. However, injury during the ECMO operation and the response of patients to ECMO may significantly influence the prognosis, and they are unpredictable before ECMO initiation. This study aims to develop a potential model based clinical characteristics at the 2-hour time point during ECMO for the early prediction of AKI in patients receiving ECMO.

Methods

139 patients who underwent ECMO were enrolled in this study. The clinical characteristics and the laboratory examinations at 2-hour time point during ECMO were recorded. The least absolute shrinkage and selection operator (LASSO) regression method was performed to select predictors, and logistic regression and a nomogram were used to establish the prediction model. The area under curve (AUC) of the receiver operating characteristic and calibration curve were used to analyze the discrimination and calibration of the model. K-fold cross-validation method was performed to validate the accuracy of this model.

Results

Among the 139 patients receiving ECMO, 106 participants (76.26%) developed AKI. Four predictive variables including ECMO model, serum creatinine (Scr-2h), uric acid(UA-2h), and serum lactate (Lac-2h) at the 2-hour time point during ECMO were filtered from 39 clinical parameters by LASSO regression. These four predictors were incorporated to develop a model for predicting AKI risk using logistic regression. The AUC of the model was 0.905 (0.845-0.965), corresponding to 81.1% sensitivity, 90.9% specificity and 83.5% accuracy. Moreover, this model showed good consistency between observed and predicted probability based on the calibration curve (P > 0.05). The validation performed by K-fold cross-validation method showed that the accuracy was 0.874 ± 0.006 in training sets, 0.827 ± 0.053 in test sets, indicating a good capability for AKI risk prediction. Finally, a nomogram based on this model was constructed to facilitate its use in clinical practice.

Conclusion

The nomogram incorporating Scr-2h,Lac-2h, UA-2h, and ECMO model may facilitate the individualized prediction of the AKI risk among patients undergoing ECMO.

Matrix metalloproteinase -12: A marker of preeclampsia?

INTRODUCTION

Enzymes, including matrix metalloproteinases (MMPs), play a significant role in trophoblast invasion - the cornerstone of preeclampsia pathogenesis.

METHODS

This study aimed to explore the dynamics of the MMP-12 concentration in blood serum during the gestational period at determined weeks in preeclampsia and physiological pregnancy to compare the results with the expression of MMP-12 in placental tissue and reveal the MMP-12 predicting role in preeclampsia.

RESULTS

Circulating serum MMP-12 was significantly decreased. The level of 0.5 ng/ml had high sensitivity and low false positivity at 11-13 weeks of pregnancy in women destined to develop pre-eclampsia in the case-control study. The dynamics curve of serum MMP-12 varied between study groups: a sharp decrease in MMP-12 concentration was found from the first trimester to the second trimester, followed by a slight increase in the third trimester of pregnancy in controls compared to the increase in concentration from the first trimester to the second trimester in pre-eclampsia. The absence of a significant difference in the concentration of MMP-12 in the II and III trimesters as well as no difference in the expression of MMP-12 protein in placental tissue in the third trimester indicates a decrease in its role after the end of placentation.

DISCUSSION

To our knowledge, this is the first study to show the dynamics of serum MMP-12 concentration during the gestational period and indicates a significant role for MMP-12 in the initial stages of placentation. The data obtained may pave the way to new early prediction strategies for preeclampsia.

Prediction and evaluation of high-risk patients with primary biliary cholangitis receiving ursodeoxycholic acid therapy: an early criterion

BACKGROUND AND AIMS

Current treatment guidelines recommend ursodeoxycholic acid (UDCA) as the first-line treatment for new-diagnosed primary biliary cholangitis (PBC) patients. However, up to 40% patients are insensitive to UDCA monotherapy, and evaluation of UDCA response at 12 months may result in long period of ineffective treatment. We aimed to develop a new criterion to reliably identify non-response patients much earlier.

METHODS

Five hundred sixty-nine patients with an average of 59 months (Median: 53; IQR:32-79) follow-up periods were randomly divided into either the training (70%) or the validation cohort (30%). The efficiency of different combinations of total bilirubin (TBIL), alkaline phosphatase (ALP), and aspartate aminotransferase (AST) threshold values to predict outcomes was assessed at 1, 3 or 6 month after the initiation of UDCA therapy. The endpoints were defined as adverse outcomes, including liver-related death, liver transplantation and complications of cirrhosis. Adverse outcome-free survival was compared using various published criteria and a proposed new criterion.

RESULTS

A new criterion of evaluating UDCA responses at 1 month was established as: ALP ≤ 2.5 × upper limit of normal (ULN) and AST ≤ 2 × ULN, and TBIL ≤ 1 × ULN (Xi'an criterion). The 5 year adverse outcome-free survival rate of UDCA responders, defined by Xi'an criterion, was 97%, which was significantly higher than that of those non-responders (64%). An accurate distinguishing high-risk patients' capacity of Xi'an criterion was confirmed in both early and late-stage PBC.

CONCLUSIONS

Xi'an criterion has a similar or even higher ability to distinguish high-risk PBC patients than other published criteria. Xi'an criterion can facilitate early identification of patients requiring new therapeutic approaches.

Dynamic prediction of life-threatening events for patients in intensive care unit

Background

Early prediction of patients’ deterioration is helpful in early intervention for patients at greater risk of deterioration in Intensive Care Unit (ICU). This study aims to apply machine learning approaches to heterogeneous clinical data for predicting life-threatening events of patients in ICU.

Methods

We collected clinical data from a total of 3151 patients admitted to the Medical Intensive Care Unit of Peking Union Medical College Hospital in China from January 1st, 2014, to October 1st, 2019. After excluding the patients who were under 18 years old or stayed less than 24 h at the ICU, a total of 2170 patients were enrolled in this study. Multiple machine learning approaches were utilized to predict life-threatening events (i.e., death) in seven 24-h windows (day 1 to day 7) and their performance was compared.

Results

Light Gradient Boosting Machine showed the best performance. We found that life-threatening events during the short-term windows can be better predicted than those in the medium-term windows. For example, death in 24 h can be predicted with an Area Under Curve of 0.905. Features like infusion pump related fluid input were highly related to life-threatening events. Furthermore, the prediction power of static features such as age and cardio-pulmonary function increased with the extended prediction window.

Conclusion

This study demonstrates that the integration of machine learning approaches and large-scale high-quality clinical data in ICU could accurately predict life-threatening events for ICU patients for early intervention.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-022-02026-x.

Early prediction of treatment response to neoadjuvant chemotherapy based on longitudinal ultrasound images of HER2-positive breast cancer patients by Siamese multi-task network: A multicentre, retrospective cohort study

BACKGROUND

Early prediction of treatment response to neoadjuvant chemotherapy (NACT) in patients with human epidermal growth factor receptor 2 (HER2)-positive breast cancer can facilitate timely adjustment of treatment regimens. We aimed to develop and validate a Siamese multi-task network (SMTN) for predicting pathological complete response (pCR) based on longitudinal ultrasound images at the early stage of NACT.

METHODS

In this multicentre, retrospective cohort study, a total of 393 patients with biopsy-proven HER2-positive breast cancer were retrospectively enrolled from three hospitals in china between December 16, 2013 and March 05, 2021, and allocated into a training cohort and two external validation cohorts. Patients receiving full cycles of NACT and with surgical pathological results available were eligible for inclusion. The key exclusion criteria were missing ultrasound images and/or clinicopathological characteristics. The proposed SMTN consists of two subnetworks that could be joined at multiple layers, which allowed for the integration of multi-scale features and extraction of dynamic information from longitudinal ultrasound images before and after the first /second cycles of NACT. We constructed the clinical model as a baseline using multivariable logistic regression analysis. Then the performance of SMTN was evaluated and compared with the clinical model.

FINDINGS

The training cohort, comprising 215 patients, were selected from Yunnan Cancer Hospital. The two independent external validation cohorts, comprising 95 and 83 patients, were selected from Guangdong Provincial People's Hospital, and Shanxi Cancer Hospital, respectively. The SMTN yielded an area under the receiver operating characteristic curve (AUC) values of 0.986 (95% CI: 0.977-0.995), 0.902 (95%CI: 0.856-0.948), and 0.957 (95%CI: 0.924-0.990) in the training cohort and two external validation cohorts, respectively, which were significantly higher than that those of the clinical model (AUC: 0.524-0.588, P _all < 0.05). The AUCs values of the SMTN within the anti-HER2 therapy subgroups were 0.833-0.972 in the two external validation cohorts. Moreover, 272 of 279 (97.5%) non-pCR patients (159 of 160 (99.4%), 53 of 54 (98.1%), and 60 of 65 (92.3%) in the training and two external validation cohorts, respectively) were successfully identified by the SMTN, suggesting that they could benefit from regime adjustment at the early-stage of NACT.

INTERPRETATION

The SMTN was able to predict pCR in the early-stage of NACT for HER2-positive breast cancer patients, which could guide clinicians in adjusting treatment regimes.

FUNDING

Key-Area Research and Development Program of Guangdong Province (No.2021B0101420006); National Natural Science Foundation of China (No.82071892, 82171920); Guangdong Provincial Key Laboratory of Artificial Intelligence in Medical Image Analysis and Application (No.2022B1212010011); the National Science Foundation for Young Scientists of China (No.82102019, 82001986); Project Funded by China Postdoctoral Science Foundation (No.2020M682643); the Outstanding Youth Science Foundation of Yunnan Basic Research Project (202101AW070001); Scientific research fund project of Department of Education of Yunnan Province(2022J0249). Science and technology Projects in Guangzhou (202201020001;202201010513); High-level Hospital Construction Project (DFJH201805, DFJHBF202105).

A novel Ontology-guided Attribute Partitioning ensemble learning model for early prediction of cognitive deficits using quantitative Structural MRI in very preterm infants

Structural magnetic resonance imaging studies have shown that brain anatomical abnormalities are associated with cognitive deficits in preterm infants. Brain maturation and geometric features can be used with machine learning models for predicting later neurodevelopmental deficits. However, traditional machine learning models would suffer from a large feature-to-instance ratio (i.e., a large number of features but a small number of instances/samples). Ensemble learning is a paradigm that strategically generates and integrates a library of machine learning classifiers and has been successfully used on a wide variety of predictive modeling problems to boost model performance. Attribute (i.e., feature) bagging method is the most commonly used feature partitioning scheme, which randomly and repeatedly draws feature subsets from the entire feature set. Although attribute bagging method can effectively reduce feature dimensionality to handle the large feature-to-instance ratio, it lacks consideration of domain knowledge and latent relationship among features. In this study, we proposed a novel Ontology-guided Attribute Partitioning (OAP) method to better draw feature subsets by considering the domain-specific relationship among features. With the better-partitioned feature subsets, we developed an ensemble learning framework, which is referred to as OAP-Ensemble Learning (OAP-EL). We applied the OAP-EL to predict cognitive deficits at 2 years of age using quantitative brain maturation and geometric features obtained at term equivalent age in very preterm infants. We demonstrated that the proposed OAP-EL approach significantly outperformed the peer ensemble learning and traditional machine learning approaches.

Early severity prediction of BPD for premature infants from chest X-ray images using deep learning: A study at the 28th day of oxygen inhalation

BACKGROUND AND OBJECTIVE

Bronchopulmonary dysplasia is a common respiratory disease in premature infants. The severity is diagnosed at the 56th day after birth or discharge by analyzing the clinical indicators, which may cause the delay of the best treatment opportunity. Thus, we proposed a deep learning-based method using chest X-ray images of the 28th day of oxygen inhalation for the early severity prediction of bronchopulmonary dysplasia in clinic.

METHODS

We first adopted a two-step lung field extraction method by combining digital image processing and human-computer interaction to form the one-to-one corresponding image and label. The designed XSEG-Net model was then trained for segmenting the chest X-ray images, with the results being used for the analysis of heart development and clinical severity. Therein, Six-Point cardiothoracic ratio measurement algorithm based on corner detection was designed for the analysis of heart development; and the transfer learning of deep convolutional neural network models were used for the early prediction of clinical severities.

RESULTS

The dice and cross-entropy loss value of the training of XSEG-Net network reached 0.9794 and 0.0146. The dice, volumetric overlap error, relative volume difference, precision, and recall were used to evaluate the trained model in testing set with the result being 98.43 ± 0.39%, 0.49 ± 0.35%, 0.49 ± 0.35%, 98.67 ± 0.40%, and 98.20 ± 0.47%, respectively. The errors between the Six-Point cardiothoracic ratio measurement method and the gold standard were 0.0122 ± 0.0084. The deep convolutional neural network model based on VGGNet had the promising prediction performance, with the accuracy, precision, sensitivity, specificity, and F1 score reaching 95.58 ± 0.48%, 95.61 ± 0.55%, 95.67 ± 0.44%, 96.98 ± 0.42%, and 95.61±0.48%, respectively.

CONCLUSIONS

These experimental results of the proposed methods in lung field segmentation, cardiothoracic ratio measurement and clinic severity prediction were better than previous methods, which proved that this method had great potential for clinical application.

A machine learning-based procedure for leveraging clickstream data to investigate early predictability of failure on interactive tasks

Early detection of risk of failure on interactive tasks comes with great potential for better understanding how examinees differ in their initial behavior as well as for adaptively tailoring interactive tasks to examinees' competence levels. Drawing on procedures originating in shopper intent prediction on e-commerce platforms, we introduce and showcase a machine learning-based procedure that leverages early-window clickstream data for systematically investigating early predictability of behavioral outcomes on interactive tasks. We derive features related to the occurrence, frequency, sequentiality, and timing of performed actions from early-window clickstreams and use extreme gradient boosting for classification. Multiple measures are suggested to evaluate the quality and utility of early predictions. The procedure is outlined by investigating early predictability of failure on two PIAAC 2012 Problem Solving in Technology Rich Environments (PSTRE) tasks. We investigated early windows of varying size in terms of time and in terms of actions. We achieved good prediction performance at stages where examinees had, on average, at least two thirds of their solution process ahead of them, and the vast majority of examinees who failed could potentially be detected to be at risk before completing the task. In-depth analyses revealed different features to be indicative of success and failure at different stages of the solution process, thereby highlighting the potential of the applied procedure for gaining a finer-grained understanding of the trajectories of behavioral patterns on interactive tasks.

Early prediction of moderate-to-severe condition of inhalation-induced acute respiratory distress syndrome via interpretable machine learning

Background

Several studies have investigated the correlation between physiological parameters and the risk of acute respiratory distress syndrome (ARDS), in addition, etiology-associated heterogeneity in ARDS has become an emerging topic quite recently; however, the intersection between the two, which is early prediction of target conditions in etiology-specific ARDS, has not been well-studied. We aimed to develop and validate a machine-learning model for the early prediction of moderate-to-severe condition of inhalation-induced ARDS.

Methods

Clinical expertise was applied with data-driven analysis. Using data from electronic intensive care units (retrospective derivation cohort) and the three most accessible vital signs (i.e. heart rate, temperature, and respiratory rate) together with feature engineering, we applied a random forest approach during the time window of 90 h that ended 6 h prior to the onset of moderate-to-severe respiratory failure (the ratio of partial pressure of arterial oxygen to fraction of inspired oxygen ≤ 200 mmHg).

Results

The trained random forest classifier was validated using two independent validation cohorts, with an area under the curve of 0.9127 (95% confidence interval 0.8713–0.9542) and 0.9026 (95% confidence interval 0.8075–1), respectively. A Stable and Interpretable RUle Set (SIRUS) was used to extract rules from the RF to provide guidelines for clinicians. We identified several predictive factors, including resp_96h_6h_min < 9, resp_96h_6h_mean ≥ 16.1, HR_96h_6h_mean ≥ 102, and temp_96h_6h_max > 100, that could be used for predicting inhalation-induced ARDS (moderate-to-severe condition) 6 h prior to onset in critical care units. (‘xxx_96h_6h_min/mean/max’: the minimum/mean/maximum values of the xxx vital sign collected during a 90 h time window beginning 96 h prior to the onset of ARDS and ending 6 h prior to the onset from every recorded blood gas test).

Conclusions

This newly established random forest‑based interpretable model shows good predictive ability for moderate-to-severe inhalation-induced ARDS and may assist clinicians in decision-making, as well as facilitate the enrolment of patients in prevention programmes to improve their outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12890-022-01963-7.

Admission rate-pressure product as an early predictor for in-hospital mortality after aneurysmal subarachnoid hemorrhage

Early prediction of in-hospital mortality in aneurysmal subarachnoid hemorrhage (aSAH) is essential for the optimal management of these patients. Recently, a retrospective cohort observation has reported that the rate-pressure product (RPP, the product of systolic blood pressure and heart rate), an objective and easily calculated bedside index of cardiac hemodynamics, was predictively associated with in-hospital mortality following traumatic brain injury. We thus wondered whether this finding could also be generalized to aSAH patients. The current study aimed to examine the association of RPP at the time of emergency room (ER) admission with in-hospital mortality and its predictive performance among aSAH patients. We retrospectively included 515 aSAH patients who had been admitted to our ER between 2016 and 2020. Their baseline heart rate and systolic blood pressure at ER presentation were extracted for the calculation of the admission RPP. Meanwhile, we collected relevant clinical, laboratory, and neuroimaging data. Then, these data including the admission RPP were examined by univariate and multivariate analyses to identify independent predictors of hospital mortality. Eventually, continuous and ordinal variables were selected from those independent predictors, and the performance of these selected predictors was further evaluated and compared based on receiver operating characteristic (ROC) curve analyzes. We identified both low (< 10,000; adjusted odds ratio (OR) 3.49, 95% CI 1.93-6.29, p < 0.001) and high (> 15,000; adjusted OR 8.42, 95% CI 4.16-17.06, p < 0.001) RPP on ER admission to be independently associated with in-hospital mortality after aSAH. Furthermore, after centering the admission RPP by its median, the area under its ROC curve (0.761, 95% CI 0.722-0.798, p < 0.001) was found to be statistically superior to any of the other independent predictors included in the ROC analyzes (all p < 0.01). In light of the predictive superiority of the admission RPP, as well as its objectivity and easy accessibility, it is indeed a potentially more applicable predictor for in-hospital death in aSAH patients.

First but not second postoperative day growth hormone assessments as early predictive tests for long-term acromegaly persistence

PURPOSE

Postoperative assessment of acromegaly activity is typically performed at least 3 months after neurosurgery (NS). Few studies have evaluated the use of early postoperative growth hormone (GH) levels as a test to predict short- and long-term remission of acromegaly. Our objective was to evaluate the diagnostic performance of serum random GH on a postoperative day one (D1-rGH) and two (D2-rGH), particularly in predicting long-term disease persistence.

MATERIALS AND METHODS

Forty-one subjects with acromegaly who were undergoing NS were enrolled (mean age ± SD 47.4 ± 13.1 years at diagnosis; women 54%; macroadenomas 71%). The final assessment of disease activity was performed one year after NS. ROC curves were used to evaluate the diagnostic performance of D1-rGH and D2-rGH.

RESULTS

After a 1-year follow-up, the overall remission rate was 55%. ROC analysis identified an optimal D1-rGH cut-off value of 2.1 ng/mL for diagnosing long-term disease persistence (55.6% SE; 90.9% SP). The cut-off point became 2.5 ng/mL after maximizing specificity for disease persistence (yielding a 100% positive predictive value) and 0.3 ng/mL after maximizing sensitivity for disease remission. The optimal D2-rGH cut-off value was 0.6 ng/mL (81.8% SE; 50% SP); the cut-off point became 2.9 ng/mL after maximizing specificity and 0.1 ng/mL after maximizing sensitivity, with no clinical utility.

CONCLUSIONS

D1-rGH could be a highly specific test for the early diagnosis of long-term acromegaly persistence, which is predicted by a value > 2.5 ng/mL with a great degree of certainty. The diagnostic performance of D2-rGH was insufficient. Further research is required to validate these preliminary results prior to modifying the postoperative management of acromegaly.

Increased proportion of Th17/Treg cells at the new diagnosed stage of chronic immune thrombocytopenia in pediatrics: the pilot study from a multi-center

Chronic immune thrombocytopenia (CITP) is an autoimmune disease with many immune dysfunctions, including T helper type 17 cell (Th17)/regulatory T cells (Tregs) imbalance. Low quality of life and side effects of drugs are severe, especially in pediatrics. This study aimed to determine Th17/Treg polarization in pediatric CITP when first diagnosing ITP and evaluate its use as a predictive marker for pediatric CITP. This was a pilot study from a multi-center. Setting the effective data size to 100 patients, data entry ended in the 142nd patient who had completed a 1-year follow-up. The percentages of Treg cells and Th17 cells were quantified by flow cytometry when new diagnosed ITP patients first arrived. The association between the Th17/Treg ratio and CITP was analyzed statistically. The percentages of Treg cells and Th17 cells were lower (P = 0.0008) and higher (P = 0.0001), respectively, in the CITP-outcome group compared with the remission group. The receiver operating characteristic analysis showed that the area under the curve (AUC) of Treg and Th17 cells was 0.811 and 0.834, respectively. The ratio of Th17/Treg exhibited the largest AUC of 0.897 (cutoff value 0.076).Conclusions: Thus, the percentage of Th17 /Treg ratio of pediatric CITP is elevated at new diagnosed ITP stage. It is a promising predictive marker for the development of CITP to some extent.Trial registration: ChiCTR1900022419 (10^th April 2019) What is Known: • The percentage of Th17 /Treg ratio of pediatric CITP is elevated. What is New: • This study shows that the percentage of Th17 /Treg ratio of pediatric CITP is elevated at new diagnosed ITP stage. This work may provide a new point for pediatric CITP's prediction. The imbalanced ratio of Th17/Treg was obvious when first diagnose ITP in pediatric CITP patients, rendering them as potential predictive tools for discriminating CITP to facilitate with the management of pediatric patient care. In addition, the combination of them may serve as a predictive marker in pediatric CITP.

Sepsis in Trauma: A Deadly Complication

Sepsis is a major cause of death following a traumatic injury. As a life-threatening medical emergency, it is defined as the body's extreme response to an infection. Without timely treatment, sepsis can rapidly lead to tissue damage, and organ failure The capacity to limit tissue damage through metabolic adaptation and repair processes is associated with an excessive immune response of the host. It is important to make an early prediction of sepsis, based on the quick Sepsis associated Organ Failure Assessment Score (qSOFA), so an accurate treatment can be initiated reducing the morbidity and mortality at the emergency and UCI services. Many factors increase the rate of complications and the development of sepsis in a trauma patient, representing a challenge to orthopedic surgeons. Several early biomarkers that help to identify and predict the inflammatory and immune responses of the host going through polytrauma and sepsis have been studied; procalcitonin (PCT), C-reactive protein (CRP), glycosylated hemoglobin (HbA1c), the Neutrophil/lymphocyte ratio (NLR), Interleukin-17 (IL-17), Caspase-1, Vanin-1, High-density lipoproteins (HDL), and the Thrombin-activable fibrinolysis inhibitor (TAFI). Once sepsis is diagnosed, treatment must be immediately initiated with an appropriate empiric antimicrobial, an all-purpose supporting treatment, and metabolic control, followed by the specific antibiotic therapy based on blood culture. Since the participation of sepsis in polytrauma has been recognized as a key event in the outcome of patients at the ICU, the ability of the specialist to early recognize a septic process has become a key feature to reduce mortality and improve clinical prognosis.

A risk score for early predicting bloodstream infections in febrile obstetric patients: a pilot study

PURPOSE

Early prediction of bloodstream infections (BSI) among obstetric patients remains to be a challenge for clinicians. The objective of this study was to develop a risk score and assess its discriminative ability in febrile obstetric patients in a maternal intensive care unit (ICU).

METHODS

Between May 2015 and August 2020, a total of 497 febrile obstetric patients were categorized into BSI group (n = 276) and Non-BSI group (n = 221) based on the result of blood cultures. White blood cell count, C-reactive protein (CRP), procalcitonin (PCT), time of interval from amniorrhea to fever (IFAF) and maximum body temperature (T_max) were compared between the two groups. All patients were divided into training set (n = 298) and validation set (n = 199). The risk score was established using univariate and multivariate logistic regression from patients in the training set, and its discriminative ability was tested among patients in the validation set.

RESULTS

The levels of neutrophil, CRP, PCT, IFAF and T_max were significantly higher in BSI group than those in Non-BSI group. PROM, T_max, neutrophil and CRP acted as independent predictive factors for BSI in the training set. The area under the receiver operating characteristic curve of risk score for early prediction of BSI in the training, validation set and the whole population was 0.829 (95% CI 0.783-0.876), 0.848 (95% CI 0.792-0.903) and 0.838 (95% CI 0.803-0.873), respectively.

CONCLUSION

The risk score has a feasible discriminatory ability in early prediction of BSI in febrile obstetric patients.

A machine learning approach to predicting risk of myelodysplastic syndrome

BACKGROUND

Early myelodysplastic syndrome (MDS) diagnosis can allow physicians to provide early treatment, which may delay advancement of MDS and improve quality of life. However, MDS often goes unrecognized and is difficult to distinguish from other disorders. We developed a machine learning algorithm for the prediction of MDS one year prior to clinical diagnosis of the disease.

METHODS

Retrospective analysis was performed on 790,470 patients over the age of 45 seen in the United States between 2007 and 2020. A gradient boosted decision tree model (XGB) was built to predict MDS diagnosis using vital signs, lab results, and demographics from the prior two years of patient data. The XGB model was compared to logistic regression (LR) and artificial neural network (ANN) models. The models did not use blast percentage and cytogenetics information as inputs. Predictions were made one year prior to MDS diagnosis as determined by International Classification of Diseases (ICD) codes, 9th and 10th revisions. Performance was assessed with regard to area under the receiver operating characteristic curve (AUROC).

RESULTS

On a hold-out test set, the XGB model achieved an AUROC value of 0.87 for prediction of MDS one year prior to diagnosis, with a sensitivity of 0.79 and specificity of 0.80. The XGB model was compared against LR and ANN models, which achieved an AUROC of 0.838 and 0.832, respectively.

CONCLUSIONS

Machine learning may allow for early MDS diagnosis MDS and more appropriate treatment administration.

Predictive value of pentraxin-3 on disease severity and mortality risk in patients with hemorrhagic fever with renal syndrome

BACKGROUND

Hemorrhagic fever with renal syndrome (HFRS) caused by Hantaan virus is characterized by systemic immunopathological injury. Pentraxin-3 is an acute-phase reactant involved in the processes of inflammation and infection. This study aimed to investigate the levels of plasma pentraxin-3 and evaluate its predictive value on disease severity and mortality risk in patients with HFRS.

METHODS

This was a prospective real-world observational study. The concentrations of plasma pentraxin-3 were measured by enzyme linked immunosorbent assay (ELISA) in 105 HFRS patients and 27 healthy controls. We analyzed the clinical relevance between pentraxin-3 and clinical subtyping, hospital stay and conventional laboratory parameters of HFRS patients. Considering the prognosis (death) as the primary endpoint, the levels of pentraxin-3 between survivors and non-survivors were compared, and its association with mortality was assessed by Kaplan-Meier survival analysis. The predictive potency of pentraxin-3 for mortality risk in HFRS patients was evaluated by receiver operating characteristic (ROC) curve analysis.

RESULTS

The levels of pentraxin-3 during the acute phase were increased with the aggravation of the disease, and showed the highest expression in critical-type patients (P < 0.05). Pentraxin-3 demonstrated significant correlations with conventional laboratory parameters (WBC, PLT, AST, ALB, APTT, Fib) and the length of hospital stay. Compared with the survivors, non-survivors showed higher levels of pentraxin-3 and worse expressions of conventional laboratory parameters during the acute phase. The Kaplan-Meier survival curves showed that high levels of pentraxin-3 during the acute phase were significantly associated with the death in HFRS patients. Pentraxin-3 demonstrated significant predictive value for the mortality risk of HFRS patients, with the area under ROC curve (AUC) of 0.753 (95%CI: 0.593 ~ 0.914, P = 0.003).

CONCLUSIONS

The detection of plasma pentraxin-3 might be beneficial to the evaluation of disease severity and to the prediction of mortality risk in HFRS patients.

A predictive framework in healthcare: Case study on cardiac arrest prediction

Data-driven healthcare uses predictive analytics to enhance decision-making and personalized healthcare. Developing prognostic models is one of the applications of predictive analytics in medical environments. Various studies have used machine learning techniques for this purpose. However, there is no specific standard for choosing prediction models for different medical purposes. In this paper, the ISAF framework was proposed for choosing appropriate prediction models with regard to the properties of the classification methods. As one of the case study applications, a prognostic model for predicting cardiac arrests in sepsis patients was developed step by step through the ISAF framework. Finally, a new modified stacking model produced the best results. We predict 85 % of heart arrest cases one hour before the incidence (sensitivity> = 0.85) and 73 % of arrest cases 25 h before the occurrence (sensitivity> = 0.73). The results indicated that the proposed prognostic model has significantly improved the prediction results compared to the two standard systems of APACHE II and MEWS. Furthermore, compared to previous research, the proposed model has extended the prediction interval and improved the performance criteria.

Tensor learning of pointwise mutual information from EHR data for early prediction of sepsis

Early detection of sepsis can facilitate early clinical intervention with effective treatment and may reduce sepsis mortality rates. In view of this, machine learning-based automated diagnosis of sepsis using easily recordable physiological data can be more promising as compared to the gold standard rule-based clinical criteria in current practice. This study aims to develop such a machine learning framework that demonstrates the quantification of heterogeneity within the tabular electronic health records (EHR) data of clinical covariates to capture both linear relationships and nonlinear correlation for the early prediction of sepsis. Here, the statistics of pairwise association for each hour-covariate pair within the EHR data for every 6-hours window-duration with selected 24 covariates is described using pointwise mutual information (PMI) matrix. This matrix gives the heterogeneity of data as a two-dimensional map. Such matrices are fused horizontally along the z-axis as vertical slices in the xy plane to form a 3-way tensor for each record with the corresponding Length of Stay (L). Tensor factorization of such fused tensor for every record is performed using Tucker decomposition, and only the core tensors are retained later, excluding the 3 unitary matrices to provide the latent feature set for the prediction of sepsis onset. A five-fold cross-validation scheme is employed wherein the obtained 120 latent features from the reshaped core tensor, are fed to Light Gradient Boosting Machine Learning models (LightGBM) for binary classification, further alleviating the involved class imbalance. The machine-learning framework is designed via Bayesian optimization, yielding an average normalized utility score of 0.4519 as defined by challenge organizers and area under the receiver operating characteristic curve (AUROC) of 0.8621 on publicly available PhysioNet/Computing in Cardiology Challenge 2019 training data. The proposed tensor decomposition of 3-way fused tensor formulated using PMI matrices leverages higher-order temporal interactions between the pairwise associations among the clinical values for early prediction of sepsis. This is validated with improved risk prediction power for every hour of admission to the ICU in terms of utility score, AUROC, and F1 score. The results obtained show a significant improvement particularly in terms of utility score of ~1.5-2% under a 5-fold cross-validation scheme on entire training data as compared to a top entrant research study that participated in the challenge.

Predicting oil content in ripe Macaw fruits (Acrocomia aculeata) from unripe ones by near infrared spectroscopy and PLS regression

A method for early quantification of unripe macaw fruits oil content using near-infrared spectroscopy (NIR) and partial least squares (PLS) is presented. After harvest, the fruit takes about 30 days to reach its maximum oil accumulation. The oil content was quantified thirty days after harvest using Soxhlet extraction. PLS models were built using NIR spectra of shell obtained five days after harvest (Shell5). The Shell5 model was compared with models built using NIR spectra of the shell (Shell30) and mesocarp thirty days after harvest (Pulp30). Ordered predictors selection was used to select the most informative variables. The best models presented root mean square error of prediction and correlation coefficient of prediction of 4.87% and 0.89 for Shell5; 5.83% and 0.85 for Shell30; 4.76% and 0.92 for Pulp30. Thus, the anticipated prediction of oil content could reduce the time and costs of macaw palm quality control and storage.

Sleep improvement is associated with the antidepressant efficacy of repeated-dose ketamine and serum BDNF levels: a post-hoc analysis

RATIONALE

Recently, the effects of ketamine on the circadian rhythm have suggested that ketamine's rapid antidepressant effects are associated with and without sleep disturbance improvement.

OBJECTIVES

Here, we evaluated the antidepressant efficacy of repeated ketamine infusions in patients with sleep disturbances.

METHODS

This study included 127 patients with major depressive disorder or bipolar disorder who received ketamine treatments during a 12-day period. Sleep quality was assessed by the 17-item Hamilton Depression Rating Scale sleep disturbance factor (SDF) (items 4, 5 and 6). Serum brain-derived neurotrophic factor (BDNF) was measured at baseline, day 13 and day 26. This study was a post-hoc analysis.

RESULTS

Significant differences were found in the HAMD-17 score at 13 post-infusion time points compared to baseline, as well as the scores in SDF score at each of the 7 post-infusion (4 h after each infusion excluded) time points among all patients. Logistic regression and linear correlation analyses revealed that a greater reduction in the SDF after 24 h of the first ketamine infusion resulted in a better antidepressant effect in the last two follow-up visits. Moreover, BDNF levels were significantly higher in sleep responders than in non-responders.

CONCLUSIONS

In the 127 patients, six ketamine infusions induced better therapeutic effects in sleep responders than in sleep non-responders and patients without sleep disturbances. The sleep response after repeated ketamine infusions was positively associated with high serum BDNF levels. Early sleep disturbance improvement (as early as 24 h after the first ketamine injection) may predict the antidepressant effect of repeated-dose ketamine.

SSP: Early prediction of sepsis using fully connected LSTM-CNN model

BACKGROUND

Sepsis is a life-threatening condition that occurs due to the body's reaction to infections, and it is a leading cause of morbidity and mortality in hospitals. Early prediction of sepsis onset facilitates early interventions that promote the survival of suspected patients. However, reliable and intelligent systems for predicting sepsis are scarce.

METHODS

This paper presents a novel technique called Smart Sepsis Predictor (SSP) to predict sepsis onset in patients admitted to an intensive care unit (ICU). SSP is a deep neural network architecture that encompasses long short-term memory (LSTM), convolutional, and fully connected layers to achieve early prediction of sepsis. SSP can work in two modes; Mode 1 uses demographic data and vital signs, and Mode 2 uses laboratory test results in addition to demographic data and vital signs. To evaluate SSP, we have used the 2019 PhysioNet/CinC Challenge dataset, which includes the records of 40,366 patients admitted to the ICU.

RESULTS

To compare SSP with existing state-of-the-art methods, we have measured the accuracy of the SSP in 4-, 8-, and 12-h prediction windows using publicly available data. Our results show that the SSP performance in Mode 1 and Mode 2 is much higher than existing methods, achieving an area under the receiver operating characteristic curve (AUROC) of 0.89 and 0.92, 0.88 and 0.87, and 0.86 and 0.84 for 4 h, 8 h, and 12 h before sepsis onset, respectively.

CONCLUSIONS

Using ICU data, sepsis onset can be predicted up to 12 h in advance. Our findings offer an early solution for mitigating the risk of sepsis onset.

Motor outcome after perinatal stroke and early prediction of unilateral spastic cerebral palsy

BACKGROUND

Unilateral spastic cerebral palsy (USCP) occurs in 30%-68% of infants with perinatal stroke. Early detection of USCP is essential for referring infants to early intervention. The aims of this study were to report motor outcomes after perinatal stroke, and to determine the predictive value of the General Movements Assessment (GMA) and Hand Assessment for Infants (HAI) for detection of USCP.

MATERIALS AND METHODS

This was a prospective observational study involving infants with perinatal stroke. GMA was conducted between 10 and 15 weeks post term-age (PTA). The HAI was performed between 3 and 5 months PTA. Motor outcome was collected between 12 and 36 months PTA.

RESULTS

The sample consisted of 46 infants. Fifteen children (32.6%) were diagnosed with CP, two children with bilateral CP and 13 with USCP. Abnormal GMA had a sensitivity of 85% (95% confidence interval [CI] 55-98%) and a specificity of 52% (95% CI 33-71%) to predict USCP. When asymmetrically presented FMs were also considered as abnormal, sensitivity increased to 100%, hence the specificity declined to 43%. A HAI asymmetry index cut-off of 23, had both a sensitivity and a specificity of 100% to detect USCP.

CONCLUSION

Using GMA and HAI can enable prediction of USCP before the age of 5 months in infants with perinatal stroke. Nevertheless, GMA must be interpreted with caution in this particular population. The HAI was found to be a very accurate screening tool for early detection of asymmetry and prediction of USCP.

Dynamic expression of hepatic GP73 mRNA and protein and circulating GP73 during hepatocytes malignant transformation

BACKGROUND

Hepatic Golgi protein-73 (GP73) expression is related to hepatocellular carcinoma (HCC) progression. The aim of this study was to investigate the dynamic expression of GP73 mRNA and protein during hepatocytes malignant transformation.

METHODS

Human GP73 expressions in 88 HCC tissues and their self-control surrounding tissues were examined by immunohistochemistry, and survival time of HCC patients was evaluated by the Kaplan-Meier method. HCC model of Sprague-Dawley rats was made by diet containing 2-fluorenylacetamide. The rats were divided into the control, hepatocyte degeneration, precanceration, and HCC groups to observe GP73 protein and mRNA alterations during hepatocytes malignant transformation.

RESULTS

The GP73 expression was significantly higher in the cancerous tissues than that in the surrounding tissues, with shorter survival time, and the positive rates of GP73 protein in human HCC tissues were 53.3% at stage I, 84.0% at stage II, 84.6% at stage III, and 60.0% at stage IV, respectively. The positive rates of hepatic GP73 protein and mRNA in the rat models were none in the control group, 66.7% and 44.4% in the hepatocytes degeneration group, 88.9% and 77.8% in the hepatocytes precanceration group, and 100% in the HCC group, respectively. There was a positive correlation (r = 0.91, P<0.01) between hepatic GP73 and serum GP73 during rat hepatocytes malignant transformation.

CONCLUSIONS

Abnormal GP73 expression may be a sensitive and valuable biomarker in hepatocarcinogensis.

Is the measurement of drain amylase content useful for predicting pancreas-related complications after gastrectomy with systematic lymphadenectomy?

Many studies investigating postoperative pancreatic fistula (POPF) after gastrectomy, including studies measuring drain amylase content (D-AMY) as a predictive factor have been reported. This article reviews previous studies and looks to the future of measuring D-AMY in patients after gastrectomy. The causes of pancreatic fluid leakage are; the parenchymal and/or thermal injury to the pancreas, and blunt injury to the pancreas by compression and retraction. Measurement of D-AMY to predict POPF has become common in clinical practice after pancreatic surgery and was later extended to the gastric surgery. Several studies have reported associations between D-AMY and POPF after gastrectomy, and the high value of D-AMY on postoperative day (POD) 1 was an independent risk factor. To improve both sensitivity and specificity, attempts have been made to enhance the predictive accuracy of factors on POD 1 as well as on POD 3 as combined markers. Although several studies have shown a high predictive ability of POPF, it has not necessarily been exploited in clinical practice. Many problems remain unresolved; ideal timing for measurement, optimal cut-off value, and means of intervention after prediction. Prospective clinical trial could be imperative in order to develop D-AMY measurement in common clinical practice for gastric surgery.

Predicting motor outcome in preterm infants from very early brain diffusion MRI using a deep learning convolutional neural network (CNN) model

BACKGROUND AND AIMS

Preterm birth imposes a high risk for developing neuromotor delay. Earlier prediction of adverse outcome in preterm infants is crucial for referral to earlier intervention. This study aimed to predict abnormal motor outcome at 2 years from early brain diffusion magnetic resonance imaging (MRI) acquired between 29 and 35 weeks postmenstrual age (PMA) using a deep learning convolutional neural network (CNN) model.

METHODS

Seventy-seven very preterm infants (born <31 weeks gestational age (GA)) in a prospective longitudinal cohort underwent diffusion MR imaging (3T Siemens Trio; 64 directions, b ​= ​2000 ​s/mm²). Motor outcome at 2 years corrected age (CA) was measured by Neuro-Sensory Motor Developmental Assessment (NSMDA). Scores were dichotomised into normal (functional score: 0, normal; n ​= ​48) and abnormal scores (functional score: 1-5, mild-profound; n ​= ​29). MRIs were pre-processed to reduce artefacts, upsampled to 1.25 ​mm isotropic resolution and maps of fractional anisotropy (FA) were estimated. Patches extracted from each image were used as inputs to train a CNN, wherein each image patch predicted either normal or abnormal outcome. In a postprocessing step, an image was classified as predicting abnormal outcome if at least 27% (determined by a grid search to maximise the model performance) of its patches predicted abnormal outcome. Otherwise, it was considered as normal. Ten-fold cross-validation was used to estimate performance. Finally, heatmaps of model predictions for patches in abnormal scans were generated to explore the locations associated with abnormal outcome.

RESULTS

For the identification of infants with abnormal motor outcome based on the FA data from early MRI, we achieved mean sensitivity 70% (standard deviation SD 19%), mean specificity 74% (SD 39%), mean AUC (area under the receiver operating characteristic curve) 72% (SD 14%), mean F1 score of 68% (SD 13%) and mean accuracy 73% (SD 19%) on an unseen test data set. Patch-based prediction heatmaps showed that the patches around the motor cortex and somatosensory regions were most frequently identified by the model with high precision (74%) as a location associated with abnormal outcome. Part of the cerebellum, and occipital and frontal lobes were also highly associated with abnormal NSMDA/motor outcome.

DISCUSSION/CONCLUSION

This study established the potential of an early brain MRI-based deep learning CNN model to identify preterm infants at risk of a later motor impairment and to identify brain regions predictive of adverse outcome. Results suggest that predictions can be made from FA maps of diffusion MRIs well before term equivalent age (TEA) without any prior knowledge of which MRI features to extract and associated feature extraction steps. This method, therefore, is suitable for any case of brain condition/abnormality. Future studies should be conducted on a larger cohort to re-validate the robustness and effectiveness of these models.

Identifying predictors of probable posttraumatic stress disorder in children and adolescents with earthquake exposure: A longitudinal study using a machine learning approach

BACKGROUND

Evidence has identified risk factors associated with individuals with trauma exposure who develop posttraumatic stress disorder (PTSD). How to combine risk factors to predict probable PTSD in young survivors using machine learning is limited. The study aimed to integrated multiple measures at 2 weeks after the earthquake using machine learning for the prediction of probable PTSD at 3 months after earthquake.

METHODS

A total of 2099 young survivors with earthquake exposure were included. We integrated multiple domains of variables to 'train' a machine learning algorithm (XGBoost). Thirty-one combination types were implemented and evaluated. The resulting XGBoost was utilized in identifying individual participants as either probable PTSD or no PTSD.

RESULTS

Any combination type predicted young survivor probable PTSD, with prediction accuracies ranging between 66%-80% (p < 0.05). In particular, the combination of earthquake experience, everyday functioning, somatic symptoms and sleeping correctly predicted 683 out of 802 cases of probable PTSD, translating to a classical accuracy of 74.476% (85.156% sensitivity and 60.366% specificity) and an area under the curve of 0.80. The most relevant variables (e.g. age, sex, property loss and a sedentary lifestyle) revealed in the present study.

LIMITATIONS

Participants from a specific district might limit the generalizability of our results. Self-report questionnaires and non-standardized measures were used to assess symptoms.

CONCLUSION

Detection of probable PTSD according to self-reported measurement data is feasible, may improve operational efficiencies via enabling targeted intervention, before manifestation of symptoms.

C-reactive protein and drain amylase accurately predict clinically relevant pancreatic fistula after partial pancreaticoduodenectomy

BACKROUND

C-reactive protein (CRP) and procalcitonin (PCT) have shown to be reliable predictors of inflammatory complications and anastomotic leak after colorectal surgery. Their predictive value after partial pancreaticoduodenectomy (PD) remains unclear.

MATERIALS AND METHODS

All consecutive pancreaticoduodenectomies (2009-2018) at our hospital were included. Drain amylase was evaluated on postoperative day (POD) 1, serum CRP and PCT were evaluated on POD 1-3. Receiver-operating characteristics curves were performed and significant cut-off values were tested using logistic regression.

RESULTS

Among 188 patients who underwent partial PD, clinically relevant pancreatic fistulas (POPF) occurred in 30 (16%) patients, including 20 (10.6%) with Grade B and 10 (5.3%) patients with Grade C. Postoperative complications (Clavien-Dindo ≥ III) were reported in 46 (24.5%) patients, including Grade IIIa in 16 (8.5%), IIIb in 18 (9.6%), IVa in 3 (1.6%), IVb in 2 (1.1%) and V in 7 (3.7%) patients. Drain amylase on POD 1 showed the largest area under the curve (0.872, p < 0.001), followed by CRP (0.803, p < 0.001) and PCT on POD 3 (0.651, p < 0.011). Drain amylase on POD 1 > 303 U/l (OR 0.045, 95% CI 0.010-0.195, p < 0.001), CRP > 203 mg/l (OR 0.098, 95% CI 0.041-0.235, p < 0.001) and PCT > 0.85 μg/l (OR 0.393, 95%CI 0.178-0.869, p = 0.02) were significant predictors of relevant POPF in the univariate analysis. CRP > 203 mg/l (OR 0.098, 95% CI 0.024-0.403, p = 0.001) and drain amylase > 303 U/l (OR 0.064, 95% CI 0.007-0.554, p = 0.01) remained independent predictors in the multivariable analysis. The combination of drain amylase on POD 1 and CRP on POD 3 had a sensitivity and specificity of 87.4% and 90.9% to predict relevant POPF.

CONCLUSION

Drain amylase on POD 1 and CRP on POD 3 can accurately predict clinically relevant POPF after partial pancreaticoduodenectomy. The accuracy of PCT on POD 3 is limited.

Early prediction of acquiring acute kidney injury for older inpatients using most effective laboratory test results

BACKGROUND

Acute Kidney Injury (AKI) is common among inpatients. Severe AKI increases all-cause mortality especially in critically ill patients. Older patients are more at risk of AKI because of the declined renal function, increased comorbidities, aggressive medical treatments, and nephrotoxic drugs. Early prediction of AKI for older inpatients is therefore crucial.

METHODS

We use 80 different laboratory tests from the electronic health records and two types of representations for each laboratory test, that is, we consider 160 (laboratory test, type) pairs one by one to do the prediction. By proposing new similarity measures and employing the classification technique of the K nearest neighbors, we are able to identify the most effective (laboratory test, type) pairs for the prediction. Furthermore, in order to know how early and accurately can AKI be predicted to make our method clinically useful, we evaluate the prediction performance of up to 5 days prior to the AKI event.

RESULTS

We compare our method with two existing works and it shows our method outperforms the others. In addition, we implemented an existing method using our dataset, which also shows our method has a better performance. The most effective (laboratory test, type) pairs found for different prediction times are slightly different. However, Blood Urea Nitrogen (BUN) is found the most effective (laboratory test, type) pair for most prediction times.

CONCLUSION

Our study is first to consider the last value and the trend of the sequence for each laboratory test. In addition, we define the exclusion criteria to identify the inpatients who develop AKI during hospitalization and we set the length of the data collection window to ensure the laboratory data we collect is close to the AKI time. Furthermore, we individually select the most effective (laboratory test, type) pairs to do the prediction for different days of early prediction. In the future, we will extend this approach and develop a system for early prediction of major diseases to help better disease management for inpatients.

Inducing human induced pluripotent stem cell differentiation through embryoid bodies: A practical and stable approach

Human induced pluripotent stem cells (hiPSCs) are invaluable resources for producing high-quality differentiated cells in unlimited quantities for both basic research and clinical use. They are particularly useful for studying human disease mechanisms in vitro by making it possible to circumvent the ethical issues of human embryonic stem cell research. However, significant limitations exist when using conventional flat culturing methods especially concerning cell expansion, differentiation efficiency, stability maintenance and multicellular 3D structure establishment, differentiation prediction. Embryoid bodies (EBs), the multicellular aggregates spontaneously generated from iPSCs in the suspension system, might help to address these issues. Due to the unique microenvironment and cell communication in EB structure that a 2D culture system cannot achieve, EBs have been widely applied in hiPSC-derived differentiation and show significant advantages especially in scaling up culturing, differentiation efficiency enhancement, ex vivo simulation, and organoid establishment. EBs can potentially also be used in early prediction of iPSC differentiation capability. To improve the stability and feasibility of EB-mediated differentiation and generate high quality EBs, critical factors including iPSC pluripotency maintenance, generation of uniform morphology using micro-pattern 3D culture systems, proper cellular density inoculation, and EB size control are discussed on the basis of both published data and our own laboratory experiences. Collectively, the production of a large quantity of homogeneous EBs with high quality is important for the stability and feasibility of many PSCs related studies.

Prediction of Autism at 3 Years from Behavioural and Developmental Measures in High-Risk Infants: A Longitudinal Cross-Domain Classifier Analysis

We integrated multiple behavioural and developmental measures from multiple time-points using machine learning to improve early prediction of individual Autism Spectrum Disorder (ASD) outcome. We examined Mullen Scales of Early Learning, Vineland Adaptive Behavior Scales, and early ASD symptoms between 8 and 36 months in high-risk siblings (HR; n = 161) and low-risk controls (LR; n = 71). Longitudinally, LR and HR-Typical showed higher developmental level and functioning, and fewer ASD symptoms than HR-Atypical and HR-ASD. At 8 months, machine learning classified HR-ASD at chance level, and broader atypical development with 69.2% Area Under the Curve (AUC). At 14 months, ASD and broader atypical development were classified with approximately 71% AUC. Thus, prediction of ASD was only possible with moderate accuracy at 14 months.

Early prediction of olanzapine-induced weight gain for schizophrenia patients

The aim of this study was to determine whether weight changes at week 2 or other factors predicted weight gain at week 6 for schizophrenia patients receiving olanzapine. This study was the secondary analysis of a six-week trial for 94 patients receiving olanzapine (5 mg/d) plus trifluoperazine (5 mg/d), or olanzapine (10 mg/d) alone. Patients were included in analysis only if they had completed the 6-week trial (per protocol analysis). Weight gain was defined as a 7% or greater increase of the patient's baseline weight. The receiver operating characteristic curve was employed to determine the optimal cutoff points of statistically significant predictors. Eleven of the 67 patients completing the 6-week trial were classified as weight gainers. Weight change at week 2 was the statistically significant predictor for ultimate weight gain at week 6. A weight change of 1.0 kg at week 2 appeared to be the optimal cutoff point, with a sensitivity of 0.92, a specificity of 0.75, and an AUC of 0.85. Using weight change at week 2 to predict weight gain at week 6 is favorable in terms of both specificity and sensitivity. Weight change of 1.0 kg or more at 2 weeks is a reliable predictor.

Dynamic contrast-enhanced ultrasonography (D-CEUS) for the early prediction of bevacizumab efficacy in patients with metastatic colorectal cancer

OBJECTIVES

To investigate early changes in tumour perfusion parameters by dynamic contrast-enhanced ultrasonography (D-CEUS) and to identify any correlation with survival and tumour response in patients with metastatic colorectal cancer (CRC) treated with bevacizumab (B).

METHODS

Thirty-seven patients randomized to either chemotherapy (C) plus B or C alone were considered for this study. D-CEUS was performed at baseline and after the first treatment cycle (day 15). Four D-CEUS perfusion parameters were considered: derived peak intensity (DPI), area under the curve (AUC), slope of wash-in (A) and time to peak intensity (TPI).

RESULTS

In patients treated with C plus B, a ≥22.5 % reduction in DPI, ≥20 % increase in TPI and ≥10 % reduction in AUC were correlated with higher progression-free survival in the C+B arm (p = 0.048, 0.024 and 0.010, respectively) but not in the C arm. None of the evaluated parameter modifications had a correlation with tumour response or overall survival.

CONCLUSIONS

D-CEUS could be useful for detecting and quantifying dynamic changes in tumour vascularity as early as 15 days after the start of B-based therapy. Although these changes may be predictive of progression-free survival, no correlation with response or overall survival was found.

KEY POINTS

• D-CEUS showed early changes in liver metastasis perfusion in colorectal cancer. • A decrease in tumour perfusion was associated with longer progression-free survival. • The decrease in perfusion was not correlated with higher overall survival.

Early metabolic markers identify potential targets for the prevention of type 2 diabetes

AIMS/HYPOTHESIS

The aims of this study were to evaluate systematically the predictive power of comprehensive metabolomics profiles in predicting the future risk of type 2 diabetes, and to identify a panel of the most predictive metabolic markers.

METHODS

We applied an unbiased systems medicine approach to mine metabolite combinations that provide added value in predicting the future incidence of type 2 diabetes beyond known risk factors. We performed mass spectrometry-based targeted, as well as global untargeted, metabolomics, measuring a total of 568 metabolites, in a Finnish cohort of 543 non-diabetic individuals from the Botnia Prospective Study, which included 146 individuals who progressed to type 2 diabetes by the end of a 10 year follow-up period. Multivariate logistic regression was used to assess statistical associations, and regularised least-squares modelling was used to perform machine learning-based risk classification and marker selection. The predictive performance of the machine learning models and marker panels was evaluated using repeated nested cross-validation, and replicated in an independent French cohort of 1044 individuals including 231 participants who progressed to type 2 diabetes during a 9 year follow-up period in the DESIR (Data from an Epidemiological Study on the Insulin Resistance Syndrome) study.

RESULTS

Nine metabolites were negatively associated (potentially protective) and 25 were positively associated with progression to type 2 diabetes. Machine learning models based on the entire metabolome predicted progression to type 2 diabetes (area under the receiver operating characteristic curve, AUC = 0.77) significantly better than the reference model based on clinical risk factors alone (AUC = 0.68; DeLong's p = 0.0009). The panel of metabolic markers selected by the machine learning-based feature selection also significantly improved the predictive performance over the reference model (AUC = 0.78; p = 0.00019; integrated discrimination improvement, IDI = 66.7%). This approach identified novel predictive biomarkers, such as α-tocopherol, bradykinin hydroxyproline, X-12063 and X-13435, which showed added value in predicting progression to type 2 diabetes when combined with known biomarkers such as glucose, mannose and α-hydroxybutyrate and routinely used clinical risk factors.

CONCLUSIONS/INTERPRETATION

This study provides a panel of novel metabolic markers for future efforts aimed at the prevention of type 2 diabetes.

Early Prediction of Persistent Organ Failure by Serum Angiopoietin-2 in Patients with Acute Pancreatitis

BACKGROUND

Biomarkers for the early prediction of the severity of acute pancreatitis (AP) are urgently needed for clinical management of the disease. Angiopoietin-2 (Ang-2), one of the autocrine peptides that reduce endothelial permeability, has been found to be associated with various diseases, including inflammatory disorders.

AIMS

This study aimed to determine whether serum Ang-2 could serve as a noninvasive biomarker for the early prediction of persistent organ failure (POF) in acute pancreatitis.

METHODS

A total of 120 AP patients were prospectively enrolled at Jinling Hospital. Serum samples were collected on admission. Clinical and laboratory data were recorded. Ang-2 levels were measured by enzyme-linked immunosorbent assay.

RESULTS

A total of 37 patients developed POF and were classified as having severe AP (SAP). Ang-2 was significantly higher on admission in patients who developed POF than in those who did not (p < 0.001 for all). Furthermore, receiver operating characteristic (ROC) curve analysis revealed that Ang-2 could distinguish patients who developed POF from mild AP (MAP, area under ROC curve [AUC] = 0.88, 95 % CI 0.78-0.94) and moderately severe AP patients (MSAP, AUC = 0.74, 95 % CI 0.63-0.83). In addition, multivariate logistic regression showed that increased Ang-2 was an independent predictor of developing POF between subgroups with MSAP and SAP (OR 7.2, 95 % CI 2.7-19.4) and among all AP patients (OR 12.1, 95 % CI 4.8-30.3).

CONCLUSIONS

Elevated serum Ang-2 levels on admission may be a promising biomarker for the prediction of POF in AP.

Management of CML-blast crisis

Tyrosine kinase inhibitors (TKI) have moderately improved survival in BC, but a median survival of less than 1 year is still unsatisfactory. This article reviews the various tests required for diagnosis of BC, features at diagnosis, treatment modalities (intensive chemotherapy, TKI, allo-SCT and a selection of investigational agents), options of prevention and predictors of progression. The best prognosis is observed in patients that achieve a 2nd CP. Allo-SCT probably further improves prognosis of patients in 2nd CP. The choice of TKI should be directed by the mutation profile of the patient. BC can be prevented. A careful analysis of risk factors for progression may help. Current treatment options are combined in a concluding strategy for the management of BC.

First-trimester multimarker prediction of gestational diabetes mellitus using targeted mass spectrometry

AIMS/HYPOTHESIS

Gestational diabetes mellitus (GDM) is associated with an increased risk of pre-eclampsia, macrosomia and the future development of type 2 diabetes mellitus in both mother and child. Although an early and accurate prediction of GDM is needed to allow intervention and improve perinatal outcome, no single protein biomarker has yet proven useful for this purpose. In the present study, we hypothesised that multimarker panels of serum proteins can improve first-trimester prediction of GDM among obese and non-obese women compared with single markers.

METHODS

A nested case-control study was performed on first-trimester serum samples from 199 GDM cases and 208 controls, each divided into an obese group (BMI ≥27 kg/m(2)) and a non-obese group (BMI <27 kg/m(2)). Based on their biological relevance to GDM or type 2 diabetes mellitus or on their previously reported potential as biomarkers for these diseases, a number of proteins were selected for targeted nano-flow liquid chromatography (LC) MS analysis. This resulted in the development and validation of a 25-plex multiple reaction monitoring (MRM) MS assay.

RESULTS

After false discovery rate correction, six proteins remained significantly different (p<0.05) between obese GDM patients (n=135) and BMI-matched controls (n=139). These included adiponectin, apolipoprotein M and apolipoprotein D. Multimarker models combining protein levels and clinical data were then constructed and evaluated by receiver operating characteristic (ROC) analysis. For the obese, non-obese and all GDM groups, these models achieved marginally higher AUCs compared with adiponectin alone.

CONCLUSIONS/INTERPRETATION

Multimarker models combining protein markers and clinical data have the potential to predict women at a high risk of developing GDM.