Development and validation of a machine learning ASA-score to identify candidates for comprehensive preoperative screening and risk stratification

OBJECTIVE

The ASA physical status (ASA-PS) is determined by an anesthesia provider or surgeon to communicate co-morbidities relevant to perioperative risk. Assigning an ASA-PS is a clinical decision and there is substantial provider-dependent variability. We developed and externally validated a machine learning-derived algorithm to determine ASA-PS (ML-PS) based on data available in the medical record.

DESIGN

Retrospective multicenter hospital registry study.

SETTING

University-affiliated hospital networks.

PATIENTS

Patients who received anesthesia at Beth Israel Deaconess Medical Center (Boston, MA, training [n = 361,602] and internal validation cohorts [n = 90,400]) and Montefiore Medical Center (Bronx, NY, external validation cohort [n = 254,412]).

MEASUREMENTS

The ML-PS was created using a supervised random forest model with 35 preoperatively available variables. Its predictive ability for 30-day mortality, postoperative ICU admission, and adverse discharge were determined by logistic regression.

MAIN RESULTS

The anesthesiologist ASA-PS and ML-PS were in agreement in 57.2% of the cases (moderate inter-rater agreement). Compared with anesthesiologist rating, ML-PS assigned more patients into extreme ASA-PS (I and IV), (p < 0.01), and less patients in ASA II and III (p < 0.01). ML-PS and anesthesiologist ASA-PS had excellent predictive values for 30-day mortality, and good predictive values for postoperative ICU admission and adverse discharge. Among the 3594 patients who died within 30 days after surgery, net reclassification improvement analysis revealed that using the ML-PS, 1281 (35.6%) patients were reclassified into the higher clinical risk category compared with anesthesiologist rating. However, in a subgroup of multiple co-morbidity patients, anesthesiologist ASA-PS had a better predictive accuracy than ML-PS.

CONCLUSIONS

We created and validated a machine learning physical status based on preoperatively available data. The ability to identify patients at high risk early in the preoperative process independent of the provider's decision is a part of the process we use to standardize the stratified preoperative evaluation of patients scheduled for ambulatory surgery.

Age-related differences in response to plasma exchange in male rat liver tissues: insights from histopathological and machine-learning assisted spectrochemical analyses

This study aimed to examine the biological effects of blood plasma exchange in liver tissues of aged and young rats using machine learning methods and spectrochemical and histopathological approaches. Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) were the machine learning algorithms employed. Young plasma was given to old male rats (24 months), while old plasma was given to young male rats (5 weeks) for thirty days. LDA (95.83-100%) and SVM (87.5-91.67%) detected significant qualitative changes in liver biomolecules. In old rats, young plasma infusion increased the length of fatty acids, triglyceride, lipid carbonyl, and glycogen levels. Nucleic acid concentration, phosphorylation, and carbonylation rates of proteins were also increased, whereas a decrease in protein concentration was measured. Aged plasma decreased protein carbonylation, triglyceride, and lipid carbonyl levels. Young plasma infusion improved hepatic fibrosis and cellular degeneration and reduced hepatic microvesicular steatosis in aged rats. Otherwise, old plasma infusion in young rats caused disrupted cellular organization, steatosis, and increased fibrosis. Young plasma administration increased liver glycogen accumulation and serum albumin levels. Aged plasma infusion raised serum ALT levels while diminished ALP concentrations in young rats, suggesting possible liver dysfunction. Young plasma increased serum albumin levels in old rats. The study concluded that young plasma infusion might be associated with declined liver damage and fibrosis in aged rats, while aged plasma infusion negatively impacted liver health in young rats. These results imply that young blood plasma holds potential as a rejuvenation therapy for liver health and function.

PTM-X: Prediction of Post-Translational Modification Crosstalk Within and Across Proteins

Posttranslational modifications (PTMs), which are processes of adding covalent groups in protein amino acids after the translation, play an important role in regulating proteins' localization, degradation, and functions. Different PTMs both within a single protein and across multiple proteins can work together or regulate reciprocally, known as PTM cross talk. However, high-throughput experimental identifications of PTM cross talk are lack due to technical limitations. In this chapter, we review in silico prediction approaches and illustrate the usage of PTM-X, a suite of recently proposed machine learning methods to predict both intra- and interprotein PTM cross talk.

Examining the predictability and prognostication of multimorbidity among older Delayed-Discharge Patients: A Machine learning analytics

BACKGROUND

Patient complexity among older delayed-discharge patients complicates discharge planning, resulting in a higher rate of adverse outcomes, such as readmission and mortality. Early prediction of multimorbidity, as a common indicator of patient complexity, can support proactive discharge planning by prioritizing complex patients and reducing healthcare inefficiencies.

OBJECTIVE

We set out to accomplish the following two objectives: 1) to examine the predictability of three common multimorbidity indices, including Charlson-Deyo Comorbidity Index (CDCI), the Elixhauser Comorbidity Index (ECI), and the Functional Comorbidity Index (FCI) using machine learning (ML), and 2) to assess the prognostic power of these indices in predicting 30-day readmission and mortality.

MATERIALS AND METHODS

We used data including 163,983 observations of patients aged 65 and older who experienced discharge delay in Ontario, Canada, during 2004 - 2017. First, we utilized various classification ML algorithms, including classification and regression trees, random forests, bagging trees, extreme gradient boosting, and logistic regression, to predict the multimorbidity status based on CDCI, ECI, and FCI. Second, we used adjusted multinomial logistic regression to assess the association between multimorbidity indices and the patient-important outcomes, including 30-day mortality and readmission.

RESULTS

For all ML algorithms and regardless of the predictive performance criteria, better predictions were established for the CDCI compared with the ECI and FCI. Remarkably, the most predictable multimorbidity index (i.e., CDCI with Area Under the Receiver Operating Characteristic Curve = 0.80, 95% CI = 0.79 - 0.81) also offered the highest prognostications regarding adverse events (RRR_mortality = 3.44, 95% CI = 3.21 - 3.68 and RRR_readmission = 1.36, 95% CI = 1.31 - 1.40).

CONCLUSIONS

Our findings highlight the feasibility and utility of predicting multimorbidity status using ML algorithms, resulting in the early detection of patients at risk of mortality and readmission. This can support proactive triage and decision-making about staffing and resource allocation, with the goal of optimizing patient outcomes and facilitating an upstream and informed discharge process through prioritizing complex patients for discharge and providing patient-centered care.

Is Cr(III) re-oxidation occurring in Cr-contaminated soils after remediation: Meta-analysis and machine learning prediction

Whether Cr(III) in Cr(III)-containing sites formed after Cr(VI) reduction and stabilization remediation are re-oxidized and pose toxicity risks again has been a growing concern. In this study, 1030 data were collected to perform a meta-analysis to clarify the effects of various factors (oxidant type, soil and Cr(III) solid compound properties, aging conditions, and testing methods) on Cr(III) oxidation. We observed that the soil properties of clay, pH ≥ 8, the lower CEC capacity, easily reducible Mn content, and Cr(III) content, and the higher Eh value and Fe content can promote the re-oxidation of Cr(III). Publication bias and sensitivity analyses confirmed the stability and reliability of the meta-analysis. Subsequently, we used five machine learning algorithms to construct and optimize the models. The prediction results of the RF model (RMSE <1.36, R2 >0.71) with good algorithm performance showed that after ten years of remediation, the extractable Cr(VI) concentration in the soil was 0.0087 mg/L, indicating a negligible secondary pollution risk of Cr(III) re-oxidation. This study provides theoretical support for subsequent risk management and control after Cr(VI) soil remediation and provides a solution for the quantitative prediction of Cr(III) re-oxidation.

Hematoma expansion prediction based on SMOTE and XGBoost algorithm

Hematoma expansion (HE) is a high risky symptom with high rate of occurrence for patients who have undergone spontaneous intracerebral hemorrhage (ICH) after a major accident or illness. Correct prediction of the occurrence of HE in advance is critical to help the doctors to determine the next step medical treatment. Most existing studies focus only on the occurrence of HE within 6 h after the occurrence of ICH, while in reality a considerable number of patients have HE after the first 6 h but within 24 h. In this study, based on the medical doctors recommendation, we focus on prediction of the occurrence of HE within 24 h, as well as the occurrence of HE every 6 h within 24 h. Based on the demographics and computer tomography (CT) image extraction information, we used the XGBoost method to predict the occurrence of HE within 24 h. In this study, to solve the issue of highly imbalanced data set, which is a frequent case in medical data analysis, we used the SMOTE algorithm for data augmentation. To evaluate our method, we used a data set consisting of 582 patients records, and compared the results of proposed method as well as few machine learning methods. Our experiments show that XGBoost achieved the best prediction performance on the balanced dataset processed by the SMOTE algorithm with an accuracy of 0.82 and F1-score of 0.82. Moreover, our proposed method predicts the occurrence of HE within 6, 12, 18 and 24 h at the accuracy of 0.89, 0.82, 0.87 and 0.94, indicating that the HE occurrence within 24 h can be predicted accurately by the proposed method.

A method to describe attenuation of river contamination under peak flows: Can the public water supply from Paraopeba River finally return after the Brumadinho dam disaster?

Tailings dams' disasters begin a stage of river water contamination with no endpoint at first sight. But when the river was formerly used for public water supply and the use was suspended as consequence of a dam break, a time window for safe suspension lift must be anticipated to help water managers. The purpose of this study was to seek for that moment in the case of Brumadinho dam disaster which occurred in 2019 and injected millions of cubic meters of iron- and manganese-rich tailings into the Paraopeba River, leading to the suspension of public water supply to Belo Horizonte metropolitan region with this resource, until now. To accomplish the proposed goal, an assemblage of artificial intelligence and socio-economic development models were used to anticipate precipitation, river discharge and metal concentrations (iron, manganese) until 2033. Then, the ratios of metal concentrations between impacted and non-impacted sites were determined and values representing extreme events of river discharge were selected for further assessment. A ratio ≈1 generally indicates a similarity between impacted and non-impacted areas or, put another way, a return of impacted areas to a pre-rupture condition. Moreover, when the ratio is estimated under the influence of peak flows, then a value of ≈1 indicates a return to pre-rupture conditions under the most unfavorable hydrologic regimes, thus a safe return. So, the extreme ratios were plotted against time and fitted to a straight line with intercept-x representing the requested safe time. The results pointed to 6.57 years after the accident, while using iron as contaminant indicator, or 8.71 years when manganese was considered. Despite of being a relatively low-risk timeframe, the suspension lift should be implemented in phases and monitored for precaution of potential sporadic contamination events, while dredging of the tailings from impacted areas should continue and be accelerated.

Machine Learning Models to Predict Bone Metastasis Risk in Patients With Lung Cancer

INTRODUCTION

The aim of this study was to find the most appropriate variables to input into machine learning algorithms to identify those patients with primary lung malignancy with high risk for metastasis to the bone.

PATIENT INCLUSION

Patients with either histological or radiological diagnoses of lung cancer were included in this study.

RESULTS

The patient cohort comprised 1864 patients diagnosed from 2016 to 2021. A total of 25 variables were considered as potential risk factors. These variables have been identified in previous studies as independent risk factors for bone metastasis. Treatment methods for lung cancer were taken into account during model development. The outcome variable was binary, (presence or absence of bone metastasis) with follow-up until death or 12-month survival, whichever is the sooner. Results showed that American Joint Committee on Cancer staging, the use of EGFR inhibitor, age, T-staging, and lymphovascular invasion were the five input features contributing the most to the model algorithm. High AJCC staging (OR 1.98; p < 0.05), the use of EGFR inhibitor (OR 6.14; p < 0.05), high T-staging (OR 1.47; p < 0.05), and the presence of lymphovascular invasion (OR 4.92; p < 0.05) increase predicted risk of bone metastasis. Conversely, older age reduces predicted bone metastasis risk (OR 0.98; p < 0.05).

CONCLUSION

The machine learning model developed in this study can be easily incorporated into the hospital's Clinical Management System so that input variables can be immediately utilized to give an accurate prediction of bone metastatic risk, therefore informing clinicians on the best treatment strategy for that individual patient.

Spatiotemporal trends in hernia disease burden and health workforce correlations in aging populations: a global analysis with projections to 2050

BACKGROUND

Inguinal, femoral, and abdominal wall hernias represent significant health and economic burdens globally, particularly among adults aged 45 and older. In 2021, the Global Burden of Disease Study reported 1.72 million new cases, 6.75 million prevalent cases, and over 41,000 deaths in this population. While age-standardized rates have declined with improved healthcare, absolute burden continues to rise due to population growth and aging. Gender disparities remain pronounced, with men experiencing sevenfold higher incidence than women. This study analyzes global hernia trends, determinants, future projections, and the association between health workforce distribution and hernia burden to inform targeted interventions.

METHODS

Using data from the Global Burden of Disease Study 2021, we analyzed incidence, prevalence, mortality, and disability-adjusted life years (DALYs) for inguinal, femoral, and abdominal wall hernias. Long-term trends were assessed using average annual percentage change (EAPC), with decomposition analyses exploring factors influencing disease burden changes. Spatial and temporal patterns were examined using age-period-cohort and frontier analyses. We conducted health inequality analyses and utilized eight time-series machine learning models to project disease burden from 2022 to 2050. Additionally, we analyzed correlations between health workforce distribution and hernia burden across 204 countries and territories for 1990 and 2019.

RESULTS

In 2021, global incidence of hernias was 1,720,177, with 6,748,203 prevalent cases and 41,834 deaths among individuals aged 45 years and older. Although age-standardized incidence rate (ASIR) decreased from 153.98/100,000 in 1990 to 112.29/100,000 in 2021 (EAPC = -0.83%, 95% CI: -0.95% to -0.70%), and age-standardized mortality rate (ASMR) decreased from 3.19/100,000 to 1.86/100,000 (EAPC = -1.77%, 95% CI: -1.94% to -1.59%), absolute burden continued increasing. Socioeconomic differences were significant, with higher ASIR in high SDI areas (141.94/100,000) than low SDI areas (104.60/100,000) in 2021, but much higher ASMR in low SDI areas (4.14/100,000) than high SDI areas (1.23/100,000). Decomposition analysis revealed population growth as the main driver of increased disease burden, contributing 173.80% to incidence increases. Age-period-cohort analysis showed incidence peaked in the 65-69 age group (RR = 1.43, 95% CI: 1.42-1.43). Male ASIR in 2021 (203.41/100,000) was approximately 7.3 times higher than female ASIR (27.94/100,000). Correlation analyses revealed significant negative associations between pharmaceutical personnel density and hernia disease burden, with correlation coefficients strengthening from 1990 (DALYs: r = -0.39, p < 0.001) to 2019 (DALYs: r = -0.57, p < 0.001). Similar trends were observed for dentistry personnel (DALYs: r = -0.26 in 1990 to r = -0.47 in 2019, p < 0.001). Countries with high hernia burden (Guatemala, Paraguay, Indonesia) consistently demonstrated lower health workforce density compared to low-burden countries. ARIMA model projections showed that by 2050, ASIR would increase slightly from 112.32/100,000 in 2022 to 112.64/100,000, with absolute new cases increasing by 19.70%. ASMR is projected to increase from 1.84/100,000 to 2.11/100,000, with deaths increasing by 8.50%.

CONCLUSIONS

Despite declining age-standardized rates for inguinal, femoral, and abdominal wall hernias, absolute disease burden continues increasing due to demographic factors. Socioeconomic development significantly impacts disease patterns, with higher morbidity but lower mortality in high SDI areas. The strong negative correlation between pharmaceutical and dentistry personnel density and hernia burden suggests potential protective effects of healthcare workforce investment, particularly in resource-constrained settings. Future projections indicate growing absolute burden despite relatively stable age-standardized rates, highlighting the urgent need to strengthen preventive measures, improve treatments, and strategically allocate health workforce resources to address this growing public health challenge.

Leveraging machine learning for sustainable cultivation of Zn-enriched crops in Cd-contaminated karst regions

Karst soils often exhibit elevated zinc (Zn) levels, providing an opportunity to cultivate Zn-enriched crops. (meanwhile) However, these soils also frequently contain high background levels of toxic metals, particularly cadmium (Cd), posing potential health risks. Understanding the bioaccumulation of Cd and Zn and the related drivers in a high geochemical background area can provide important insights for the safe development of Zn-enriched crops. Traditional models often struggle to accurately predict metal levels in crop systems grown on soils with high geochemical background. This study employed machine learning models, including Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGBoost), to explore effective strategies for sustainable cultivation of Zn-enriched crops in karst regions, focusing on bioaccumulation factors (BAF). A total of 10,986 topsoil samples and 181 paired rhizosphere soil-crop samples, including early rice, late rice, and maize, were collected from a karst region in Guangxi. The SVM and XGBoost models demonstrated superior performance, achieving R2 values of 0.84 and 0.60 for estimating the BAFs of Zn and Cd, respectively. Key determinants of the BAFs were identified, including soil iron and manganese contents, pH level, and the interaction between Zn and Cd. By integrating these soil properties with machine learning, a framework for the safe cultivation of Zn-enriched crops was developed. This research contributes to the development of strategies for mitigating Zn deficiency in crops grown on Cd-contaminated soils.

Impacts of micro/nano plastics on the ecotoxicological effects of antibiotics in agricultural soil: A comprehensive study based on meta-analysis and machine learning prediction

Micro/nano plastics (M/NPs) and antibiotics, as widely coexisting pollutants in environment, pose serious threats to soil ecosystem. The purpose of this study was to systematically evaluate the ecological effects of the co-exposure of M/NPs and antibiotics on soil organisms through the meta-analysis and machine learning prediction. Totally, 1002 data set from 38 articles were studied. The co-exposure of M/NPs significantly promoted the abundance (62.68 %) and migration level (55.22 %) of antibiotic contamination in soil, and caused serious biotoxicity to plants (-12.31 %), animals (-12.03 %), and microorganisms (35.07 %). Using 10 variables, such as risk response categories, basic physicochemical properties, exposure objects, and exposure time of M/NPs, as data sources, Random Forests (RF) and eXtreme Gradient Boosting (XGBoost) models were developed to predict the impacts of M/NPs on the ecotoxicological effects of antibiotics in agricultural soil. The effective R2 values (0.58 and 0.60, respectively) indicated that both models can be used to predict the future ecological risk of M/NPs and antibiotics coexistence in soil. Particle size (13.54 %), concentration (5.02 %), and type (11.18 %) of M/NPs were the key characteristic parameters that affected the prediction results. The findings of this study indicate that the co-exposure of M/NPs and antibiotics in soil not only exacerbates antibiotic contamination levels but also causes severe toxic effects to soil organism. Furthermore, this study provides an effective approach for ecological risk assessment of the coexistence of M/NPs and antibiotics in environment.

Intestinal Microbiome Modulation of Therapeutic Efficacy of Cancer Immunotherapy

Bacteria are associated with certain cancers and may induce genetic instability and cancer progression. The gut microbiome modulates the response to cancer therapy. Training machine learning models with response associated taxa or bacterial genes predict patients' response to immunotherapies with moderate accuracy. Clinical trials targeting the gut microbiome to improve immunotherapy efficacy have been conducted. While single bacterial strains or small consortia have not be reported yet to be successful, encouraging results have been reported in small single arm and randomized studies using transplant of fecal microbiome from cancer patients who successfully responded to therapy or from healthy volunteers.

Spatial variation in stability of wheat (Triticum aestivum L.) straw phytolith-occluded carbon in China

Global climate warming, driven by human activities emitting greenhouse gases like CO2, results in adverse effects, posing significant challenges to human health and food security. In response to this challenge, it is imperative to enhance long-term carbon sequestration, including phytolith-occluded carbon (PhytOC). Currently, there is a dearth of research on the assessment and distribution of the stability of PhytOC. Additionally, the intricate relationships and effects between the stability and environmental factors such as climate and soil remain insufficiently elucidated. Our study provided a composite assessment index for PhytOC stability based on a rapid solubility assay and principal component analysis. The machine learning models that we developed in this study, utilize experimentally and publicly accessible environmental data on large spatial scales, facilitating the prediction and spatial distribution mapping of the PhytOC stability using simple kriging interpolation in wheat ecosystems across China. We compared and evaluated 10 common classification machine learning models at 10-fold cross-validation. Based on the overall performance, the Stochastic Gradient Boosting model (GBM) was selected as predictive model. The stability is influenced by dynamic and complex environments with climate having a more significant impact. It was evident that light and temperature had a significant positive direct relationship with the stability, while the other factors showed indirect effects on the stability. PhytOC stability exhibited obvious zonal difference and spatial heterogeneity, with the distribution trend gradually decreasing from the southeast to the northwest in China. Overall, our research contributed to reducing greenhouse gas emissions and achieving global climate targets, working towards a more sustainable and climate-resilient future.

Identification of Recurrence-associated Gene Signatures and Machine Learning-based Prediction in IDH-Wildtype Histological Glioblastoma

Glioblastoma (GBM) is a highly aggressive brain tumor with frequent recurrence, yet the molecular mechanisms driving recurrence remain poorly understood. Identifying recurrence-associated genes may improve prognosis and treatment strategies. We applied weighted gene co-expression network analysis (WGCNA) to transcriptomic data from IDH-wildtype histological GBM in the CGGA-693 (n = 190) and CGGA-325 (n = 111) cohorts to identify recurrence-associated genes. These genes were validated using RT-qPCR and single-cell RNA sequencing (scRNA-seq) datasets (GSE174554, GSE131928). Their associations with immune cell composition were analyzed. Finally, we evaluated 113 machine learning algorithms to develop a multi-gene predictive model for GBM recurrence, with model performance assessed using receiver operating characteristic (ROC) curves and confusion matrix analysis. We identified eight recurrence-associated genes (CERS2, EML2, FNBP1, ICOSLG, MFAP3L, NPC1, ROGDI, SLAIN1) that were significantly differentially expressed between primary and recurrent GBM. The scRNA-seq analysis revealed cell-type-specific expression patterns, with eight genes predominantly enriched in oligodendrocytes, malignant GBM subtypes, and immune cells. Immune cell deconvolution showed significant alterations in macrophage polarization and NK cell activation in recurrent GBM. Machine learning analysis demonstrated that random forest (RF) was the most effective model, achieving AUC values of 0.998, 0.968, and 0.998 in the training, CGGA-693 validation, and CGGA-325 validation cohorts, respectively, suggesting high predictive accuracy. This study identifies novel recurrence-associated molecular signatures and establishes a machine learning-based predictive model in IDH-wildtype histological GBM.