External validation of the prediction model of intradialytic hypotension: a multicenter prospective cohort study

OBJECTIVE

Intradialytic hypotension (IDH) is a common and serious complication in patients with Maintenance Hemodialysis (MHD). The purpose of this study is to externally verify three IDH risk prediction models recently developed by Ma et al. and recalibrate, update and present the optimal model to improve the accuracy and applicability of the model in clinical environment.

METHODS

A multicenter prospective cohort study of patients from 11 hemodialysis centers in Sichuan Province, China, was conducted using convenience sampling from March 2022 to July 2022, with a follow-up period of 1 month. Model performance was assessed by: (1) Discrimination: Evaluated through the computation of the Area Under Curve (AUC) and its corresponding 95% confidence intervals. (2) Calibration: scrutinized through visual inspection of the calibration plot and utilization of the Brier score. (3) The incremental value of risk prediction and the utility of updating the model were gauged using NRI (Net Reclassification Improvement) and IDI (Integrated Discrimination Improvement). Decision Curve Analysis (DCA) was employed to evaluate the clinical benefit of updating the model.

RESULTS

The final cohort comprised 2235 individuals undergoing maintenance hemodialysis, exhibiting a 14.6% occurrence rate of IDH. The externally validated Area Under the Curve (AUC) values for the three original prediction models were 0.746 (95% CI: 0.718 to 0.775), 0.709 (95% CI: 0.679 to 0.739), and 0.735 (95% CI: 0.706 to 0.764) respectively. Conversely, the AUC value for the recalibrated and updated columnar plot model reached 0.817 (95% CI: 0.791 to 0.842), accompanied by a Brier score of 0.081. Furthermore, Decision Curve Analysis (DCA) exhibited a net benefit within the threshold probability range of 15.2% to 87.1%.

CONCLUSION

Externally validated, recalibrated, updated, and presented IDH prediction models may serve as a valuable instrument for evaluating IDH risk in clinical practice. Furthermore, they hold the potential to guide clinical providers in discerning individuals at risk and facilitating judicious clinical intervention decisions.

Prognostic Models for Chronic Low Back Pain Outcomes in Primary Care Are at High Risk of Bias and Lack Validation-High-Quality Studies Are Needed: A Systematic Review

OBJECTIVE: To provide an updated overview of available prognostic models for people with chronic low back pain (LBP) in primary care. DESIGN: Prognosis systematic review LITERATURE SEARCH: We searched for relevant studies on MEDLINE, Embase, Web of Science, and CINAHL databases (up to July 13, 2022), and performed citation tracking in Web of Science. STUDY SELECTION CRITERIA: We included observational (cohort or nested case-control) studies and randomized controlled trials that developed or validated prognostic models for adults with chronic LBP in primary care. The outcomes of interest were physical functioning, pain intensity, and health-related quality of life at any follow-up time-point. DATA SYNTHESIS: Data were extracted using the CHecklist for critical Appraisal and data extraction for systematic Reviews of prediction Modelling Studies (CHARMS), and the Prediction model Risk of Bias Assessment Tool (PROBAST) tool was used to evaluate the risk of bias of the models. Due to the number of studies retrieved and the heterogeneity, we reported the results descriptively. RESULTS: Ten studies (out of 5593 hits screened) with 34 models met our inclusion criteria, of which six are development studies and four are external validation studies. Five studies reported the area under the curve of the models (ranging from 0.48 to 0.84), whereas no study reported calibration indices. The most promising model is the Örebro Musculoskeletal Pain Screening Questionnaire Short-Form. CONCLUSIONS: Given the high risk of bias and lack of external validation, we cannot recommend that clinicians use prognostic models for patients with chronic LBP in primary care settings. J Orthop Sports Phys Ther 2024;54(5):1-13. Epub 15 February 2024. doi:10.2519/jospt.2024.12081.

Ventilatory efficiency as a prognostic factor for postoperative complications in patients undergoing elective major surgery: a systematic review

BACKGROUND

Major surgery is associated with high complication rates. Several risk scores exist to assess individual patient risk before surgery but have limited precision. Novel prognostic factors can be included as additional building blocks in existing prediction models. A candidate prognostic factor, measured by cardiopulmonary exercise testing, is ventilatory efficiency (VE/VCO2). The aim of this systematic review was to summarise evidence regarding VE/VCO2 as a prognostic factor for postoperative complications in patients undergoing major surgery.

METHODS

A medical library specialist developed the search strategy. No database-provided limits, considering study types, languages, publication years, or any other formal criteria were applied to any of the sources. Two reviewers assessed eligibility of each record and rated risk of bias in included studies.

RESULTS

From 10,082 screened records, 65 studies were identified as eligible. We extracted adjusted associations from 32 studies and unadjusted from 33 studies. Risk of bias was a concern in the domains 'study confounding' and 'statistical analysis'. VE/VCO2 was reported as a prognostic factor for short-term complications after thoracic and abdominal surgery. VE/VCO2 was also reported as a prognostic factor for mid- to long-term mortality. Data-driven covariable selection was applied in 31 studies. Eighteen studies excluded VE/VCO2 from the final multivariable regression owing to data-driven model-building approaches.

CONCLUSIONS

This systematic review identifies VE/VCO2 as a predictor for short-term complications after thoracic and abdominal surgery. However, the available data do not allow conclusions about clinical decision-making. Future studies should select covariables for adjustment a priori based on external knowledge.

SYSTEMATIC REVIEW PROTOCOL

PROSPERO (CRD42022369944).

Ensemble Machine Learning for Predicting 90-Day Outcomes and Analyzing Risk Factors in Acute Kidney Injury Requiring Dialysis

Purpose

Our objectives were to (1) employ ensemble machine learning algorithms utilizing real-world clinical data to predict 90-day prognosis, including dialysis dependence and mortality, following the first hospitalized dialysis and (2) identify the significant factors associated with overall outcomes.

Patients and Methods

We identified hospitalized patients with Acute kidney injury requiring dialysis (AKI-D) from a dataset of the Taipei Medical University Clinical Research Database (TMUCRD) from January 2008 to December 2020. The extracted data comprise demographics, comorbidities, medications, and laboratory parameters. Ensemble machine learning models were developed utilizing real-world clinical data through the Google Cloud Platform.

Results

The Study Analyzed 1080 Patients in the Dialysis-Dependent Module, Out of Which 616 Received Regular Dialysis After 90 Days. Our Ensemble Model, Consisting of 25 Feedforward Neural Network Models, Demonstrated the Best Performance with an Auroc of 0.846. We Identified the Baseline Creatinine Value, Assessed at Least 90 Days Before the Initial Dialysis, as the Most Crucial Factor. We selected 2358 patients, 984 of whom were deceased after 90 days, for the survival module. The ensemble model, comprising 15 feedforward neural network models and 10 gradient-boosted decision tree models, achieved superior performance with an AUROC of 0.865. The pre-dialysis creatinine value, tested within 90 days prior to the initial dialysis, was identified as the most significant factor.

Conclusion

Ensemble machine learning models outperform logistic regression models in predicting outcomes of AKI-D, compared to existing literature. Our study, which includes a large sample size from three different hospitals, supports the significance of the creatinine value tested before the first hospitalized dialysis in determining overall prognosis. Healthcare providers could benefit from utilizing our validated prediction model to improve clinical decision-making and enhance patient care for the high-risk population.

A Systematic Review of Predictor Composition, Outcomes, Risk of Bias, and Validation of COVID-19 Prognostic Scores

BACKGROUND

Numerous prognostic scores have been published to support risk stratification for patients with coronavirus disease 2019 (COVID-19).

METHODS

We performed a systematic review to identify the scores for confirmed or clinically assumed COVID-19 cases. An in-depth assessment and risk of bias (ROB) analysis (Prediction model Risk Of Bias ASsessment Tool [PROBAST]) was conducted for scores fulfilling predefined criteria ([I] area under the curve [AUC)] ≥ 0.75; [II] a separate validation cohort present; [III] training data from a multicenter setting [≥2 centers]; [IV] point-scale scoring system).

RESULTS

Out of 1522 studies extracted from MEDLINE/Web of Science (20/02/2023), we identified 242 scores for COVID-19 outcome prognosis (mortality 109, severity 116, hospitalization 14, long-term sequelae 3). Most scores were developed using retrospective (75.2%) or single-center (57.1%) cohorts. Predictor analysis revealed the primary use of laboratory data and sociodemographic information in mortality and severity scores. Forty-nine scores were included in the in-depth analysis. The results indicated heterogeneous quality and predictor selection, with only five scores featuring low ROB. Among those, based on the number and heterogeneity of validation studies, only the 4C Mortality Score can be recommended for clinical application so far.

CONCLUSIONS

The application and translation of most existing COVID scores appear unreliable. Guided development and predictor selection would have improved the generalizability of the scores and may enhance pandemic preparedness in the future.

Effects of IV fluid restriction according to site-specific intensity of standard fluid treatment-protocol

BACKGROUND

Variation in usual practice in fluid trials assessing lower versus higher volumes may affect overall comparisons. To address this, we will evaluate the effects of heterogeneity in treatment intensity in the Conservative versus Liberal Approach to Fluid Therapy of Septic Shock in Intensive Care trial. This will reflect the effects of differences in site-specific intensities of standard fluid treatment due to local practice preferences while considering participant characteristics.

METHODS

We will assess the effects of heterogeneity in treatment intensity across one primary (all-cause mortality) and three secondary outcomes (serious adverse events or reactions, days alive without life support and days alive out of hospital) after 90 days. We will classify sites based on the site-specific intensity of standard fluid treatment, defined as the mean differences in observed versus predicted intravenous fluid volumes in the first 24 h in the standard-fluid group while accounting for differences in participant characteristics. Predictions will be made using a machine learning model including 22 baseline predictors using the extreme gradient boosting algorithm. Subsequently, sites will be grouped into fluid treatment intensity subgroups containing at least 100 participants each. Subgroups differences will be assessed using hierarchical Bayesian regression models with weakly informative priors. We will present the full posterior distributions of relative (risk ratios and ratios of means) and absolute differences (risk differences and mean differences) in each subgroup.

DISCUSSION

This study will provide data on the effects of heterogeneity in treatment intensity while accounting for patient characteristics in critically ill adult patients with septic shock.

REGISTRATIONS

The European Clinical Trials Database (EudraCT): 2018-000404-42, ClinicalTrials. gov: NCT03668236.

A review of disease risk prediction methods and applications in the omics era

Risk prediction and disease prevention are the innovative care challenges of the 21st century. Apart from freeing the individual from the pain of disease, it will lead to low medical costs for society. Until very recently, risk assessments have ushered in a new era with the emergence of omics technologies, including genomics, transcriptomics, epigenomics, proteomics, and so on, which potentially advance the ability of biomarkers to aid prediction models. While risk prediction has achieved great success, there are still some challenges and limitations. We reviewed the general process of omics-based disease risk model construction and the applications in four typical diseases. Meanwhile, we highlighted the problems in current studies and explored the potential opportunities and challenges for future clinical practice.

Prognosis Impact and Prediction of Trans-Radial Access Failure in Patients With STEMI, A Nationwide Observational Study

Trans-radial access (TRA) is the primary arterial approach for percutaneous coronary intervention (PCI) in ST-elevation myocardial infarction (STEMI). However, occasionally, a crossover to trans-femoral access is necessary because of unsuccessful TRA. The impact of failed TRA on the prognosis in STEMI patients and the utility of predictive models for TRA failure remains uncertain. Data from the Hungarian Myocardial Infarction Registry (January 2014 to December 2020) were analyzed. Primary endpoints were 1-year mortality and major adverse cardiovascular events. Propensity score matching was employed to create a balanced cohort for comparing successful and failed TRA. The impact of unsuccessful TRA on prognosis was evaluated using Cox regression analysis. Machine learning techniques were applied to predict TRA failure. The performance and the clinical applicability of the novel and previous prediction models were comprehensively evaluated. Of 76,625 registered patients, 34,293 (69.8 ± 13.4 years, male/female: 21,893/12,400) underwent TRA (33,573) or failed TRA (720) PCI for STEMI. After propensity score matching, in the unsuccessful TRA group, the risk of mortality (34.3% vs 22.5%, hazard ratio 1.6, 95% confidence interval 1.3 to 2.0, p <0.001) and major adverse cardiovascular events (37.4% vs 26.8%, hazard ratio 1.5, 95% confidence interval 1.3 to 1.8, p <0.001) were significantly higher. Door-to-balloon time did not differ significantly (p = 0.835). In predictive analysis, Regularized Discriminant Analysis emerged as the most promising model, surpassing previous prediction models (area under the curve: 0.66, sensitivity: 0.32, specificity: 0.86). Nevertheless, Global Registry of Acute Coronary Events (GRACE) 2.0 score demonstrated a remarkable performance (area under the curve: 0.65, sensitivity: 0.51, specificity: 0.73). This study underscores the pivotal role of successful TRA in enhancing outcomes in STEMI cases, advocating for its prioritization. The inability to conclude interventions through this approach is linked to a poorer prognosis, even in risk-adjusted analyses. Our findings indicate that prediction models utilizing clinical parameters do not outperform the established GRACE 2.0 algorithm, questioning their utility. In conclusion, the results emphasize the significance of TRA success and the continued relevance of the GRACE score in clinical decision-making to optimize patient outcomes.

Climate change impacts on temperate fruit and nut production: a systematic review

Temperate fruit and nut crops require distinctive cold and warm seasons to meet their physiological requirements and progress through their phenological stages. Consequently, they have been traditionally cultivated in warm temperate climate regions characterized by dry-summer and wet-winter seasons. However, fruit and nut production in these areas faces new challenging conditions due to increasingly severe and erratic weather patterns caused by climate change. This review represents an effort towards identifying the current state of knowledge, key challenges, and gaps that emerge from studies of climate change effects on fruit and nut crops produced in warm temperate climates. Following the PRISMA methodology for systematic reviews, we analyzed 403 articles published between 2000 and 2023 that met the defined eligibility criteria. A 44-fold increase in the number of publications during the last two decades reflects a growing interest in research related to both a better understanding of the effects of climate anomalies on temperate fruit and nut production and the need to find strategies that allow this industry to adapt to current and future weather conditions while reducing its environmental impacts. In an extended analysis beyond the scope of the systematic review methodology, we classified the literature into six main areas of research, including responses to environmental conditions, water management, sustainable agriculture, breeding and genetics, prediction models, and production systems. Given the rapid expansion of climate change-related literature, our analysis provides valuable information for researchers, as it can help them identify aspects that are well understood, topics that remain unexplored, and urgent questions that need to be addressed in the future.

A comparison of hyperparameter tuning procedures for clinical prediction models: A simulation study

Tuning hyperparameters, such as the regularization parameter in Ridge or Lasso regression, is often aimed at improving the predictive performance of risk prediction models. In this study, various hyperparameter tuning procedures for clinical prediction models were systematically compared and evaluated in low-dimensional data. The focus was on out-of-sample predictive performance (discrimination, calibration, and overall prediction error) of risk prediction models developed using Ridge, Lasso, Elastic Net, or Random Forest. The influence of sample size, number of predictors and events fraction on performance of the hyperparameter tuning procedures was studied using extensive simulations. The results indicate important differences between tuning procedures in calibration performance, while generally showing similar discriminative performance. The one-standard-error rule for tuning applied to cross-validation (1SE CV) often resulted in severe miscalibration. Standard non-repeated and repeated cross-validation (both 5-fold and 10-fold) performed similarly well and outperformed the other tuning procedures. Bootstrap showed a slight tendency to more severe miscalibration than standard cross-validation-based tuning procedures. Differences between tuning procedures were larger for smaller sample sizes, lower events fractions and fewer predictors. These results imply that the choice of tuning procedure can have a profound influence on the predictive performance of prediction models. The results support the application of standard 5-fold or 10-fold cross-validation that minimizes out-of-sample prediction error. Despite an increased computational burden, we found no clear benefit of repeated over non-repeated cross-validation for hyperparameter tuning. We warn against the potentially detrimental effects on model calibration of the popular 1SE CV rule for tuning prediction models in low-dimensional settings.

Estimation of aortic stiffness by finger photoplethysmography using enhanced pulse wave analysis and machine learning

Introduction

Aortic stiffness plays a critical role in the evolution of cardiovascular diseases, but the assessment requires specialized equipment. Photoplethysmography (PPG) and single-lead electrocardiogram (ECG) are readily available in healthcare and wearable devices. We studied whether a brief PPG registration, alone or in combination with single-lead ECG, could be used to reliably estimate aortic stiffness.

Methods

A proof-of-concept study with simultaneous high-resolution index finger recordings of infrared PPG, single-lead ECG, and finger blood pressure (Finapres) was performed in 33 participants [median age 44 (range 21-66) years, 19 men] and repeated within 2 weeks. Carotid-femoral pulse wave velocity (cfPWV; two-site tonometry with SphygmoCor) was used as a reference. A brachial single-cuff oscillometric device assessed aortic pulse wave velocity (aoPWV; Arteriograph) for further comparisons. We extracted 136 established PPG waveform features and engineered 13 new with improved coupling to the finger blood pressure curve. Height-normalized pulse arrival time (NPAT) was derived using ECG. Machine learning methods were used to develop prediction models.

Results

The best PPG-based models predicted cfPWV and aoPWV well (root-mean-square errors of 0.70 and 0.52 m/s, respectively), with minor improvements by adding NPAT. Repeatability and agreement were on par with the reference equipment. A new PPG feature, an amplitude ratio from the early phase of the waveform, was most important in modelling, showing strong correlations with cfPWV and aoPWV (r = -0.81 and -0.75, respectively, both P < 0.001).

Conclusion

Using new features and machine learning methods, a brief finger PPG registration can estimate aortic stiffness without requiring additional information on age, anthropometry, or blood pressure. Repeatability and agreement were comparable to those obtained using non-invasive reference equipment. Provided further validation, this readily available simple method could improve cardiovascular risk evaluation, treatment, and prognosis.

Using Artificial Intelligence to Predict Mechanical Ventilation Weaning Success in Patients with Respiratory Failure, Including Those with Acute Respiratory Distress Syndrome

The management of mechanical ventilation (MV) remains a challenge in intensive care units (ICUs). The digitalization of healthcare and the implementation of artificial intelligence (AI) and machine learning (ML) has significantly influenced medical decision-making capabilities, potentially enhancing patient outcomes. Acute respiratory distress syndrome, an overwhelming inflammatory lung disease, is common in ICUs. Most patients require MV. Prolonged MV is associated with an increased length of stay, morbidity, and mortality. Shortening the MV duration has both clinical and economic benefits and emphasizes the need for better MV weaning management. AI and ML models can assist the physician in weaning patients from MV by providing predictive tools based on big data. Many ML models have been developed in recent years, dealing with this unmet need. Such models provide an important prediction regarding the success of the individual patient's MV weaning. Some AI models have shown a notable impact on clinical outcomes. However, there are challenges in integrating AI models into clinical practice due to the unfamiliar nature of AI for many physicians and the complexity of some AI models. Our review explores the evolution of weaning methods up to and including AI and ML as weaning aids.

Validation of PRE-DELIRIC and E-PRE-DELIRIC in a Danish population of intensive care unit patients-A prospective observational multicenter study

BACKGROUND

Delirium is a clinical condition characterized by an acute change in brain function and is frequently observed in critically ill patients. The condition has been associated with negative outcomes, making it crucial to identify patients who are at risk. Two recent prediction models have been developed to estimate the risk of delirium in intensive care unit (ICU) patients; the prediction model for delirium (PRE-DELIRIC) and the early prediction model for delirium (E-PRE-DELIRIC). We aimed to perform an external validation of these models in a Danish cohort of critically ill patients.

METHODS

We conducted a prospective, observational multicenter study to validate the PRE-DELIRIC and E-PRE-DELIRIC models in a population of patients admitted to four general ICUs in the Zealand Region of Denmark. From January 2022 to January 2023 all adult patients acutely admitted to the participating ICUs were assessed for eligibility. Patients had to be admitted to the ICU for >24 h to be included in the study. Included patients were screened with E-PRE-DELIRIC upon ICU admission and PRE-DELIRIC after 24 h of admission and followed throughout their ICU stay with CAM-ICU delirium assessments. Our primary outcomes were the prognostic accuracy measured by Area Under the Receiver Operating Characteristics (AUROC) and the calibration plot for the E-PRE-DELIRIC and PRE-DELIRIC prediction models.

RESULTS

We included 660 patients, of whom 660 were assessed with E-PRE-DELIRIC, and 622 were assessed with PRE-DELIRIC. PRE-DELIRIC showed acceptable discrimination with AUROC of 0.70 (95% CI 0.66 to 0.74) and good calibration. E-PRE-DELIRIC had inadequate discrimination AUROC of 0.63 (95% CI 0.58 to 0.67) and poor calibration.

CONCLUSION

In a Danish cohort, we found that the PRE-DELIRIC model demonstrated acceptable performance and E-PRE-DELIRIC demonstrated poor performance. In critically ill adult patients PRE-DELIRIC may be useful in identifying patients at high risk of delirium.

Predicting risk of preterm birth in singleton pregnancies using machine learning algorithms

We aimed to develop, train, and validate machine learning models for predicting preterm birth (<37 weeks' gestation) in singleton pregnancies at different gestational intervals. Models were developed based on complete data from 22,603 singleton pregnancies from a prospective population-based cohort study that was conducted in 51 midwifery clinics and hospitals in Wenzhou City of China between 2014 and 2016. We applied Catboost, Random Forest, Stacked Model, Deep Neural Networks (DNN), and Support Vector Machine (SVM) algorithms, as well as logistic regression, to conduct feature selection and predictive modeling. Feature selection was implemented based on permutation-based feature importance lists derived from the machine learning models including all features, using a balanced training data set. To develop prediction models, the top 10%, 25%, and 50% most important predictive features were selected. Prediction models were developed with the training data set with 5-fold cross-validation for internal validation. Model performance was assessed using area under the receiver operating curve (AUC) values. The CatBoost-based prediction model after 26 weeks' gestation performed best with an AUC value of 0.70 (0.67, 0.73), accuracy of 0.81, sensitivity of 0.47, and specificity of 0.83. Number of antenatal care visits before 24 weeks' gestation, aspartate aminotransferase level at registration, symphysis fundal height, maternal weight, abdominal circumference, and blood pressure emerged as strong predictors after 26 completed weeks. The application of machine learning on pregnancy surveillance data is a promising approach to predict preterm birth and we identified several modifiable antenatal predictors.

Prediction Models for Dysphagia in Intensive Care Unit after Mechanical Ventilation: A Systematic Review and Meta-analysis

OBJECTIVE

Dysphagia is a common condition that can independently lead to death in patients in the intensive care unit (ICU), particularly those who require mechanical ventilation. Despite extensive research on the predictors of dysphagia development, consistency across these studies is lacking. Therefore, this study aimed to identify predictors and summarize existing prediction models for dysphagia in ICU patients undergoing invasive mechanical ventilation.

METHODS

We searched five databases: PubMed, EMBASE, Web of Science, Cochrane Library, and the China National Knowledge Infrastructure. Studies that developed a post-extubation dysphagia risk prediction model in ICU were included. A meta-analysis of individual predictor variables was performed with mixed-effects models. The risk of bias was assessed using the prediction model risk of bias assessment tool (PROBAST).

RESULTS

After screening 1,923 references, we ultimately included nine studies in our analysis. The most commonly identified risk predictors included in the final risk prediction model were the length of indwelling endotracheal tube ≥72 h, Acute Physiology and Chronic Health Evaluation (APACHE) II score ≥15, age ≥65 years, and duration of gastric tube ≥72 h. However, PROBAST analysis revealed a high risk of bias in the performance of these prediction models, mainly because of the lack of external validation, inadequate pre-screening of variables, and improper treatment of continuous and categorical predictors.

CONCLUSIONS

These models are particularly susceptible to bias because of numerous limitations in their development and inadequate external validation. Future research should focus on externally validating the existing model in ICU patients with varying characteristics. Moreover, assessing the acceptance and effectiveness of the model in clinical practice is needed.

LEVEL OF EVIDENCE

NA Laryngoscope, 134:517-525, 2024.

A machine learning model to predict the risk of depression in US adults with obstructive sleep apnea hypopnea syndrome: a cross-sectional study

Objective

Depression is very common and harmful in patients with obstructive sleep apnea hypopnea syndrome (OSAHS). It is necessary to screen OSAHS patients for depression early. However, there are no validated tools to assess the likelihood of depression in patients with OSAHS. This study used data from the National Health and Nutrition Examination Survey (NHANES) database and machine learning (ML) methods to construct a risk prediction model for depression, aiming to predict the probability of depression in the OSAHS population. Relevant features were analyzed and a nomogram was drawn to visually predict and easily estimate the risk of depression according to the best performing model.

Study design

This is a cross-sectional study.

Methods

Data from three cycles (2005-2006, 2007-2008, and 2015-2016) were selected from the NHANES database, and 16 influencing factors were screened and included. Three prediction models were established by the logistic regression algorithm, least absolute shrinkage and selection operator (LASSO) algorithm, and random forest algorithm, respectively. The receiver operating characteristic (ROC) area under the curve (AUC), specificity, sensitivity, and decision curve analysis (DCA) were used to assess evaluate and compare the different ML models.

Results

The logistic regression model had lower sensitivity than the lasso model, while the specificity and AUC area were higher than the random forest and lasso models. Moreover, when the threshold probability range was 0.19-0.25 and 0.45-0.82, the net benefit of the logistic regression model was the largest. The logistic regression model clarified the factors contributing to depression, including gender, general health condition, body mass index (BMI), smoking, OSAHS severity, age, education level, ratio of family income to poverty (PIR), and asthma.

Conclusion

This study developed three machine learning (ML) models (logistic regression model, lasso model, and random forest model) using the NHANES database to predict depression and identify influencing factors among OSAHS patients. Among them, the logistic regression model was superior to the lasso and random forest models in overall prediction performance. By drawing the nomogram and applying it to the sleep testing center or sleep clinic, sleep technicians and medical staff can quickly and easily identify whether OSAHS patients have depression to carry out the necessary referral and psychological treatment.

Evaluation of gut microbiota predictive potential associated with phenotypic characteristics to identify multifactorial diseases

Gut microbiota has been implicated in various clinical conditions, yet the substantial heterogeneity in gut microbiota research results necessitates a more sophisticated approach than merely identifying statistically different microbial taxa between healthy and unhealthy individuals. Our study seeks to not only select microbial taxa but also explore their synergy with phenotypic host variables to develop novel predictive models for specific clinical conditions.

DESIGN

We assessed 50 healthy and 152 unhealthy individuals for phenotypic variables (PV) and gut microbiota (GM) composition by 16S rRNA gene sequencing. The entire modeling process was conducted in the R environment using the Random Forest algorithm. Model performance was assessed through ROC curve construction.

RESULTS

We evaluated 52 bacterial taxa and pre-selected PV (p < 0.05) for their contribution to the final models. Across all diseases, the models achieved their best performance when GM and PV data were integrated. Notably, the integrated predictive models demonstrated exceptional performance for rheumatoid arthritis (AUC = 88.03%), type 2 diabetes (AUC = 96.96%), systemic lupus erythematosus (AUC = 98.4%), and type 1 diabetes (AUC = 86.19%).

CONCLUSION

Our findings underscore that the selection of bacterial taxa based solely on differences in relative abundance between groups is insufficient to serve as clinical markers. Machine learning techniques are essential for mitigating the considerable variability observed within gut microbiota. In our study, the use of microbial taxa alone exhibited limited predictive power for health outcomes, while the integration of phenotypic variables into predictive models substantially enhanced their predictive capabilities.

Using Discrete Form Data in the Electronic Medical Record to Predict the Likelihood of Psychiatric Consultation

BACKGROUND

Manually screening for mental health needs in acute medical-surgical settings is thorough but time-intensive. Automated approaches to screening can enhance efficiency and reliability, but the predictive accuracy of automated screening remains largely unknown.

OBJECTIVE

The aims of this project are to develop an automated screening list using discrete form data in the electronic medical record that identify medical inpatients with psychiatric needs and to evaluate its ability to predict the likelihood of psychiatric consultation.

METHODS

An automated screening list was incorporated into an existing manual screening process for 1 year. Screening items were applied to the year's implementation data to determine whether they predicted consultation likelihood. Consultation likelihood was designated high, medium, or low. This prediction model was applied hospital-wide to characterize mental health needs.

RESULTS

The screening items were derived from nursing screens, orders, and medication and diagnosis groupers. We excluded safety or suicide sitters from the model because all patients with sitters received psychiatric consultation. Area under the receiver operating characteristic curve for the regression model was 84%. The two most predictive items in the model were "3 or more psychiatric diagnoses" (odds ratio 15.7) and "prior suicide attempt" (odds ratio 4.7). The low likelihood category had a negative predictive value of 97.2%; the high likelihood category had a positive predictive value of 46.7%.

CONCLUSIONS

Electronic medical record discrete data elements predict the likelihood of psychiatric consultation. Automated approaches to screening deserve further investigation.

Machine Learning Models for Predicting Stroke Mortality in Malaysia: An Application and Comparative Analysis

Background Stroke is a significant public health concern characterized by increasing mortality and morbidity. Accurate long-term outcome prediction for acute stroke patients, particularly stroke mortality, is vital for clinical decision-making and prognostic management. This study aimed to develop and compare various prognostic models for stroke mortality prediction. Methods In a retrospective cohort study from January 2016 to December 2021, we collected data from patients diagnosed with acute stroke from five selected hospitals. Data contained variables on demographics, comorbidities, and interventions retrieved from medical records. The cohort comprised 950 patients with 20 features. Outcomes (censored vs. death) were determined by linking data with the Malaysian National Mortality Registry. We employed three common survival modeling approaches, the Cox proportional hazard regression (Cox), support vector machine (SVM), and random survival forest (RSF), while enhancing the Cox model with Elastic Net (Cox-EN) for feature selection. Models were compared using the concordance index (C-index), time-dependent area under the curve (AUC), and discrimination index (D-index), with calibration assessed by the Brier score. Results The support vector machine (SVM) model excelled among the four, with three-month, one-year, and three-year time-dependent AUC values of 0.842, 0.846, and 0.791; a D-index of 5.31 (95% CI: 3.86, 7.30); and a C-index of 0.803 (95% CI: 0.758, 0.847). All models exhibited robust calibration, with three-month, one-year, and three-year Brier scores ranging from 0.103 to 0.220, all below 0.25. Conclusion The support vector machine (SVM) model demonstrated superior discriminative performance, suggesting its efficacy in developing prognostic models for stroke mortality. This study enhances stroke mortality prediction and supports clinical decision-making, emphasizing the utility of the support vector machine method.

Predicting moderate drinking behaviors in National Health and Nutrition Examination Survey participants using biochemical and demographical factors with machine learning

Recent studies revealed that any amount of alcohol consumption is an overall health detriment to multiple populations, contrary to popular beliefs. In addition, very few alcohol use studies utilized machine learning methods to compare the biological health of moderate drinkers compared to those that abstain from alcohol consumption, opting instead to focus on binge drinking and heavy drinking. Using participant data of multiple factor types from the National Health and Nutrition Examination Survey, we created prediction models with stacked ensembles and gradient boosting models. Machine learning models were used to identify which factors most enabled the prediction of moderate drinking behaviors. Our combined factor runs produced a cross-validation area under the curve (AUC) of 0.929 and a validation area under the curve of 0.806. Runs that only included biochemical or demographical factors received cross-validation AUC values of 0.825 and 0.925, and validation AUC values of 0.757 and 0.783, respectively. The top predictive factors for our machine learning runs, including gamma glutamyl transferase, gender, iron levels, and cigarette and marijuana usage, corroborate past studies that link those factors to alcohol consumption. Our findings identified key differences in the biological health of moderate drinkers compared to those that abstain from drinking. These results reveal a need to further explore the health effects of moderate drinking, especially for vulnerable populations.

The Not-So-Distant Future or Just Hype? Utilizing Machine Learning to Predict 30-Day Post-Operative Complications in Laparoscopic Colectomy Patients

BACKGROUND

Complex machine learning (ML) models have revolutionized predictions in clinical care. However, for laparoscopic colectomy (LC), prediction of morbidity by ML has not been adequately analyzed nor compared against traditional logistic regression (LR) models.

METHODS

All LC patients, between 2017 and 2019, in the National Surgical Quality Improvement Program (NSQIP) were identified. A composite outcome of 17 variables defined any post-operative morbidity. Seven of the most common complications were additionally analyzed. Three ML models (Random Forests, XGBoost, and L1-L2-RFE) were compared with LR.

RESULTS

Random Forests, XGBoost, and L1-L2-RFE predicted 30-day post-operative morbidity with average area under the curve (AUC): .709, .712, and .712, respectively. LR predicted morbidity with AUC = .712. Septic shock was predicted with AUC ≤ .9, by ML and LR.

CONCLUSION

There was negligible difference in the predictive ability of ML and LR in post-LC morbidity prediction. Possibly, the computational power of ML cannot be realized in limited datasets.

A synthetic data integration framework to leverage external summary-level information from heterogeneous populations

There is a growing need for flexible general frameworks that integrate individual-level data with external summary information for improved statistical inference. External information relevant for a risk prediction model may come in multiple forms, through regression coefficient estimates or predicted values of the outcome variable. Different external models may use different sets of predictors and the algorithm they used to predict the outcome Y given these predictors may or may not be known. The underlying populations corresponding to each external model may be different from each other and from the internal study population. Motivated by a prostate cancer risk prediction problem where novel biomarkers are measured only in the internal study, this paper proposes an imputation-based methodology, where the goal is to fit a target regression model with all available predictors in the internal study while utilizing summary information from external models that may have used only a subset of the predictors. The method allows for heterogeneity of covariate effects across the external populations. The proposed approach generates synthetic outcome data in each external population, uses stacked multiple imputation to create a long dataset with complete covariate information. The final analysis of the stacked imputed data is conducted by weighted regression. This flexible and unified approach can improve statistical efficiency of the estimated coefficients in the internal study, improve predictions by utilizing even partial information available from models that use a subset of the full set of covariates used in the internal study, and provide statistical inference for the external population with potentially different covariate effects from the internal population.

Machine Learning-Based Operational State Recognition and Compressive Property Prediction in Fused Filament Fabrication

3D printing has exhibited significant potential in outer space and medical implants. To use this technology in the specific high-value scenarios, 3D-printed parts need to satisfy quality-related requirements. In this article, the influence of the filament feeder operating states of 3D printer on the compressive properties of 3D-printed parts is studied in the fused filament fabrication process. A machine learning approach, back-propagation neural network with a genetic algorithm (GA-BPNN) optimized by k-fold cross-validation, is proposed to monitor the operating states and predict the compressive properties. Vibration and current sensors are used in situ to monitor the operating states of the filament feeder, and a set of features are extracted and selected from raw sensor data in time and frequency domains. Results show that the operating states of the filament feeder significantly affected the compressive properties of the fabricated samples, the operating states were accurately recognized with 96.3% rate, and compressive properties were successfully predicted by the GA-BPNN. This proposed method has the potential for use in industrial applications after 3D printing without requiring any further quality control.

Predicting amyloid-beta pathology in the general population

INTRODUCTION

Reliable models to predict amyloid beta (Aβ) positivity in the general aging population are lacking but could become cost-efficient tools to identify individuals at risk of developing Alzheimer's disease.

METHODS

We developed Aβ prediction models in the clinical Anti-Amyloid Treatment in Asymptomatic Alzheimer's (A4) Study (n = 4,119) including a broad range of easily ascertainable predictors (demographics, cognition and daily functioning, health and lifestyle factors). Importantly, we determined the generalizability of our models in the population-based Rotterdam Study (n = 500).

RESULTS

The best performing model in the A4 Study (area under the curve [AUC] = 0.73 [0.69-0.76]), including age, apolipoprotein E (APOE) ε4 genotype, family history of dementia, and subjective and objective measures of cognition, walking duration and sleep behavior, was validated in the independent Rotterdam Study with higher accuracy (AUC = 0.85 [0.81-0.89]). Yet, the improvement relative to a model including only age and APOE ε4 was marginal.

DISCUSSION

Aβ prediction models including inexpensive and non-invasive measures were successfully applied to a general population-derived sample more representative of typical older non-demented adults.

Validation and Comparison of Four Mortality Prediction Models in a Geriatric Ward in China

Purpose

The efficacy of mortality risk prediction models among older patients in China remains uncertain. We aimed to validate and compare the performances of the Walter Index, Geriatric Prognostic Index (GPI), Charlson Comorbidity Index (CCI), and FRAIL Scale in predicting 1-year all-cause mortality post-discharge in geriatric inpatients in China.

Patients and Methods

This study was conducted at a geriatric ward of a tertiary Hospital in Beijing, including patients aged 70 years or older with a documented comprehensive geriatric assessment, discharged between January 1, 2016, and December 31, 2021. Patients with a hospital stay ≤24 h or >60 days were excluded. All-cause mortality data within one year of discharge were collected from medical files and telephone interviews between August 2022 and February 2023. Multiple imputation, Logistic regression analysis, Brier scores, C-statistics, Hosmer-Lemeshow goodness-of-fit-test, and calibration plots were employed for statistical analysis.

Results

We included 832 patients with a median (interquartile range) age of 77 (74-82) years. One-hundred patients (12.0%) died within one year. After adjusting for covariates-marital status, social support, cigarette use, length of stay, number of medications, hemoglobin levels, handgrip strength, and Short Physical Performance Battery-CCI scores of 3-4 and >4, and increased Walter Index, GPI, and FRAIL Scale scores were significantly associated with 1-year mortality risk. The Brier scores varied from 0.07 (Walter Index) to 0.10 (FRAIL Scale). The C-statistic ranged from 0.74 (95% confidence interval, 0.69-0.78) for FRAIL Scale to 0.88 (95% confidence interval, 0.84-0.91) for the Walter Index. Calibration curves showed that the Walter Index, GPI, and FRAIL Scale were well calibrated, while the CCI was poor.

Conclusion

Combining the Brier score, discrimination and calibration, the Walter Index was confirmed for the first time to be the best model to predict the 1-year mortality risk of geriatric inpatients in China among the four models.

Using feature optimization and LightGBM algorithm to predict the clinical pregnancy outcomes after in vitro fertilization

Background

According to a recent report by the WHO, approximately 17.5\% (about one-sixth) of the global adult population is affected by infertility. Consequently, researchers worldwide have proposed various machine learning models to improve the prediction of clinical pregnancy outcomes during IVF cycles. The objective of this study is to develop a machine learning(ML) model that predicts the outcomes of pregnancies following in vitro fertilization (IVF) and assists in clinical treatment.

Methods

This study conducted a retrospective analysis on provincial reproductive centers in China from March 2020 to March 2021, utilizing 13 selected features. The algorithms used included XGBoost, LightGBM, KNN, Naïve Bayes, Random Forest, and Decision Tree. The results were evaluated using performance metrics such as precision, recall, F1-score, accuracy and AUC, employing five-fold cross-validation repeated five times.

Results

Among the models, LightGBM achieved the best performance, with an accuracy of 92.31%, recall of 87.80%, F1-score of 90.00\%, and an AUC of 90.41%. The model identified the estrogen concentration at the HCG injection(etwo), endometrium thickness (mm) on HCG day(EM TNK), years of infertility(Years), and body mass index(BMI) as the most important features.

Conclusion

This study successfully demonstrates the LightGBM model has the best predictive effect on pregnancy outcomes during IVF cycles. Additionally, etwo was found to be the most significant predictor for successful IVF compared to other variables. This machine learning approach has the potential to assist fertility specialists in providing counseling and adjusting treatment strategies for patients.

Characterizing the pre-clinical phase of inflammatory bowel disease

Understanding the biological changes that precede a diagnosis of inflammatory bowel disease (IBD) could facilitate pre-emptive interventions, including risk factor modification, but this pre-clinical phase of disease remains poorly characterized. Using measurements from 17 hematological and biochemical parameters taken up to 10 years before diagnosis in over 20,000 IBD patients and population controls, we address this at massive scale. We observe widespread significant changes in multiple biochemical and hematological parameters that occur up to 8 years before diagnosis of Crohn's disease (CD) and up to 3 years before diagnosis of ulcerative colitis. These changes far exceed previous expectations regarding the length of this pre-diagnostic phase, revealing an opportunity for earlier intervention, especially in CD. In summary, using a nationwide, case-control dataset-obtained from the Danish registers-we provide a comprehensive characterization of the hematological and biochemical changes that occur in the pre-clinical phase of IBD.

External Validation of Models for Predicting Disability in Community-Dwelling Older People in the Netherlands: A Comparative Study

Background

Advanced statistical modeling techniques may help predict health outcomes. However, it is not the case that these modeling techniques always outperform traditional techniques such as regression techniques. In this study, external validation was carried out for five modeling strategies for the prediction of the disability of community-dwelling older people in the Netherlands.

Methods

We analyzed data from five studies consisting of community-dwelling older people in the Netherlands. For the prediction of the total disability score as measured with the Groningen Activity Restriction Scale (GARS), we used fourteen predictors as measured with the Tilburg Frailty Indicator (TFI). Both the TFI and the GARS are self-report questionnaires. For the modeling, five statistical modeling techniques were evaluated: general linear model (GLM), support vector machine (SVM), neural net (NN), recursive partitioning (RP), and random forest (RF). Each model was developed on one of the five data sets and then applied to each of the four remaining data sets. We assessed the performance of the models with calibration characteristics, the correlation coefficient, and the root of the mean squared error.

Results

The models GLM, SVM, RP, and RF showed satisfactory performance characteristics when validated on the validation data sets. All models showed poor performance characteristics for the deviating data set both for development and validation due to the deviating baseline characteristics compared to those of the other data sets.

Conclusion

The performance of four models (GLM, SVM, RP, RF) on the development data sets was satisfactory. This was also the case for the validation data sets, except when these models were developed on the deviating data set. The NN models showed a much worse performance on the validation data sets than on the development data sets.

Impact of Modified Atmosphere Packaging Conditions on Quality of Dates: Experimental Study and Predictive Analysis Using Artificial Neural Networks

Dates are highly perishable fruits, and maintaining their quality during storage is crucial. The current study aims to investigate the impact of storage conditions on the quality of dates (Khalas and Sukary cultivars) at the Tamer stage and predict their quality attributes during storage using artificial neural networks (ANN). The studied storage conditions were the modified atmosphere packing (MAP) gases (CO2, O2, and N), packaging materials, storage temperature, and storage time, and the evaluated quality attributes were moisture content, firmness, color parameters (L*, a*, b*, and ∆E), pH, water activity, total soluble solids, and microbial contamination. The findings demonstrated that the storage conditions significantly impacted (p < 0.05) the quality of the two stored date cultivars. The use of MAP with 20% CO2 + 80% N had a high potential to decrease the rate of color transformation and microbial growth of dates stored at 4 °C for both stored date cultivars. The developed ANN models efficiently predicted the quality changes of stored dates closely aligned with observed values under the different storage conditions, as evidenced by low Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) values. In addition, the reliability of the developed ANN models was further affirmed by the linear regression between predicted and measured values, which closely follow the 1:1 line, with R2 values ranging from 0.766 to 0.980, the ANN models demonstrate accurate estimating of fruit quality attributes. The study's findings contribute to food quality and supply chain management through the identification of optimal storage conditions and predicting the fruit quality during storage under different atmosphere conditions, thereby minimizing food waste and enhancing food safety.

Enhanced phenotypes for identifying opioid overdose in emergency department visit electronic health record data

Background

Accurate identification of opioid overdose (OOD) cases in electronic healthcare record (EHR) data is an important element in surveillance, empirical research, and clinical intervention. We sought to improve existing OOD electronic phenotypes by incorporating new data types beyond diagnostic codes and by applying several statistical and machine learning methods.

Materials and Methods

We developed an EHR dataset of emergency department visits involving OOD cases or patients considered at risk for an OOD and ascertained true OOD status through manual chart reviews. We developed and validated prediction models using Random Forest, Extreme Gradient Boost, and Elastic Net models that incorporated 717 features involving primary and second diagnoses, chief complaints, medications prescribed, vital signs, laboratory results, and procedural codes. We also developed models limited to single data types.

Results

A total of 1718 records involving 1485 patients were manually reviewed; 541 (36.4%) patients had one or more OOD. Prediction performance was similar for all models; sensitivity varied from 94% to 97%; and area under the receiver operating characteristic curve (AUC) was 98% for all methods. The primary diagnosis and chief complaint were the most important contributors to AUC performance; primary diagnoses and medication class contributed most to sensitivity; chief complaint, primary diagnosis, and vital signs were most important for specificity. Models limited to decision support data types available in real time demonstrated robust prediction performance.

Conclusions

Substantial prediction performance improvements were demonstrated for identifying OODs in EHR data. Our e-phenotypes could be applied in surveillance, retrospective empirical applications, or clinical decision support systems.

Machine learning approaches for predicting 5-year breast cancer survival: A multicenter study

The study used clinical data to develop a prediction model for breast cancer survival. Breast cancer prognostic factors were explored using machine learning techniques. We conducted a retrospective study using data from the Taipei Medical University Clinical Research Database, which contains electronic medical records from three affiliated hospitals in Taiwan. The study included female patients aged over 20 years who were diagnosed with primary breast cancer and had medical records in hospitals between January 1, 2009 and December 31, 2020. The data were divided into training and external testing datasets. Nine different machine learning algorithms were applied to develop the models. The performances of the algorithms were measured using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and F1-score. A total of 3914 patients were included in the study. The highest AUC of 0.95 was observed with the artificial neural network model (accuracy, 0.90; sensitivity, 0.71; specificity, 0.73; PPV, 0.28; NPV, 0.94; and F1-score, 0.37). Other models showed relatively high AUC, ranging from 0.75 to 0.83. According to the optimal model results, cancer stage, tumor size, diagnosis age, surgery, and body mass index were the most critical factors for predicting breast cancer survival. The study successfully established accurate 5-year survival predictive models for breast cancer. Furthermore, the study found key factors that could affect breast cancer survival in Taiwanese women. Its results might be used as a reference for the clinical practice of breast cancer treatment.

Modeling Gastrointestinal Tract Wet Pool Size in Small Ruminants

The gastrointestinal tract (GIT) wet pool size (GITwps) refers to the total amount of wet contents in GIT, which in small ruminants can reach up to 19% of their body weight (BW). This study aimed to develop models to comprehensively predict GITwps in small ruminants using a meta-regression approach. A dataset was created based on 21 studies, comprising 750 individual records of sheep and goats. Various predictor variables, including BW, sex, breed, species, intake level, physiological states, stages and types of pregnancy, dry matter intake, and neutral detergent fiber intake (NDFI), were initially analyzed through simple linear regression. Subsequently, the variables were fitted using natural logarithm transformations, considering the random effect of the study and residual error, employing a supervised forward selection procedure. Overall, no significant relationship between GITwps and BW (p = 0.326) was observed for animals fed a milk-based diet. However, a strong negative linear relationship (p < 0.001) was found for animals on a solid diet, with the level of restriction influencing GITwps only at the intercept. Furthermore, the prediction of GITwps was independent of sex and influenced by species in cases where individuals were fed ad libitum. Pregnant females showed a noticeable reduction in GITwps, which was more pronounced in cases of multiple pregnancies, regardless of species (p < 0.01). The composition of the diet was found to be the primary factor affecting the modulation of GITwps, with NDFI able to override the species effect (p < 0.0001). Overall, this study sheds light on the factors influencing GITwps in small ruminants, providing valuable insights into their digestive processes and nutritional requirements.

Predicting Patterns of Distant Metastasis in Breast Cancer Patients following Local Regional Therapy Using Machine Learning

Up to 30% of breast cancer (BC) patients will develop distant metastases (DM), for which there is no cure. Here, statistical and machine learning (ML) models were developed to estimate the risk of site-specific DM following local-regional therapy. This retrospective study cohort included 175 patients diagnosed with invasive BC who later developed DM. Clinicopathological information was collected for analysis. Outcome variables were the first site of metastasis (brain, bone or visceral) and the time interval (months) to developing DM. Multivariate statistical analysis and ML-based multivariable gradient boosting machines identified factors associated with these outcomes. Machine learning models predicted the site of DM, demonstrating an area under the curve of 0.74, 0.75, and 0.73 for brain, bone and visceral sites, respectively. Overall, most patients (57%) developed bone metastases, with increased odds associated with estrogen receptor (ER) positivity. Human epidermal growth factor receptor-2 (HER2) positivity and non-anthracycline chemotherapy regimens were associated with a decreased risk of bone DM, while brain metastasis was associated with ER-negativity. Furthermore, non-anthracycline chemotherapy alone was a significant predictor of visceral metastasis. Here, clinicopathologic and treatment variables used in ML prediction models predict the first site of metastasis in BC. Further validation may guide focused patient-specific surveillance practices.

Estimating the area under the ROC curve when transporting a prediction model to a target population

We propose methods for estimating the area under the receiver operating characteristic (ROC) curve (AUC) of a prediction model in a target population that differs from the source population that provided the data used for original model development. If covariates that are associated with model performance, as measured by the AUC, have a different distribution in the source and target populations, then AUC estimators that only use data from the source population will not reflect model performance in the target population. Here, we provide identification results for the AUC in the target population when outcome and covariate data are available from the sample of the source population, but only covariate data are available from the sample of the target population. In this setting, we propose three estimators for the AUC in the target population and show that they are consistent and asymptotically normal. We evaluate the finite-sample performance of the estimators using simulations and use them to estimate the AUC in a nationally representative target population from the National Health and Nutrition Examination Survey for a lung cancer risk prediction model developed using source population data from the National Lung Screening Trial.

Consolidated Reporting Guidelines for Prognostic and Diagnostic Machine Learning Modeling Studies: Development and Validation

BACKGROUND

The reporting of machine learning (ML) prognostic and diagnostic modeling studies is often inadequate, making it difficult to understand and replicate such studies. To address this issue, multiple consensus and expert reporting guidelines for ML studies have been published. However, these guidelines cover different parts of the analytics lifecycle, and individually, none of them provide a complete set of reporting requirements.

OBJECTIVE

We aimed to consolidate the ML reporting guidelines and checklists in the literature to provide reporting items for prognostic and diagnostic ML in in-silico and shadow mode studies.

METHODS

We conducted a literature search that identified 192 unique peer-reviewed English articles that provide guidance and checklists for reporting ML studies. The articles were screened by their title and abstract against a set of 9 inclusion and exclusion criteria. Articles that were filtered through had their quality evaluated by 2 raters using a 9-point checklist constructed from guideline development good practices. The average κ was 0.71 across all quality criteria. The resulting 17 high-quality source papers were defined as having a quality score equal to or higher than the median. The reporting items in these 17 articles were consolidated and screened against a set of 6 inclusion and exclusion criteria. The resulting reporting items were sent to an external group of 11 ML experts for review and updated accordingly. The updated checklist was used to assess the reporting in 6 recent modeling papers in JMIR AI. Feedback from the external review and initial validation efforts was used to improve the reporting items.

RESULTS

In total, 37 reporting items were identified and grouped into 5 categories based on the stage of the ML project: defining the study details, defining and collecting the data, modeling methodology, model evaluation, and explainability. None of the 17 source articles covered all the reporting items. The study details and data description reporting items were the most common in the source literature, with explainability and methodology guidance (ie, data preparation and model training) having the least coverage. For instance, a median of 75% of the data description reporting items appeared in each of the 17 high-quality source guidelines, but only a median of 33% of the data explainability reporting items appeared. The highest-quality source articles tended to have more items on reporting study details. Other categories of reporting items were not related to the source article quality. We converted the reporting items into a checklist to support more complete reporting.

CONCLUSIONS

Our findings supported the need for a set of consolidated reporting items, given that existing high-quality guidelines and checklists do not individually provide complete coverage. The consolidated set of reporting items is expected to improve the quality and reproducibility of ML modeling studies.

Construction and evaluation of hourly average indoor PM2.5 concentration prediction models based on multiple types of places

Background

People usually spend most of their time indoors, so indoor fine particulate matter (PM2.5) concentrations are crucial for refining individual PM2.5 exposure evaluation. The development of indoor PM2.5 concentration prediction models is essential for the health risk assessment of PM2.5 in epidemiological studies involving large populations.

Methods

In this study, based on the monitoring data of multiple types of places, the classical multiple linear regression (MLR) method and random forest regression (RFR) algorithm of machine learning were used to develop hourly average indoor PM2.5 concentration prediction models. Indoor PM2.5 concentration data, which included 11,712 records from five types of places, were obtained by on-site monitoring. Moreover, the potential predictor variable data were derived from outdoor monitoring stations and meteorological databases. A ten-fold cross-validation was conducted to examine the performance of all proposed models.

Results

The final predictor variables incorporated in the MLR model were outdoor PM2.5 concentration, type of place, season, wind direction, surface wind speed, hour, precipitation, air pressure, and relative humidity. The ten-fold cross-validation results indicated that both models constructed had good predictive performance, with the determination coefficients (R2) of RFR and MLR were 72.20 and 60.35%, respectively. Generally, the RFR model had better predictive performance than the MLR model (RFR model developed using the same predictor variables as the MLR model, R2 = 71.86%). In terms of predictors, the importance results of predictor variables for both types of models suggested that outdoor PM2.5 concentration, type of place, season, hour, wind direction, and surface wind speed were the most important predictor variables.

Conclusion

In this research, hourly average indoor PM2.5 concentration prediction models based on multiple types of places were developed for the first time. Both the MLR and RFR models based on easily accessible indicators displayed promising predictive performance, in which the machine learning domain RFR model outperformed the classical MLR model, and this result suggests the potential application of RFR algorithms for indoor air pollutant concentration prediction.

TinyML-Sensor for Shelf Life Estimation of Fresh Date Fruits

Fresh dates have a limited shelf life and are susceptible to spoilage, which can lead to economic losses for producers and suppliers. The problem of accurate shelf life estimation for fresh dates is essential for various stakeholders involved in the production, supply, and consumption of dates. Modified atmosphere packaging (MAP) is one of the essential methods that improves the quality and increases the shelf life of fresh dates by reducing the rate of ripening. Therefore, this study aims to apply fast and cost-effective non-destructive techniques based on machine learning (ML) to predict and estimate the shelf life of stored fresh date fruits under different conditions. Predicting and estimating the shelf life of stored date fruits is essential for scheduling them for consumption at the right time in the supply chain to benefit from the nutritional advantages of fresh dates. The study observed the physicochemical attributes of fresh date fruits, including moisture content, total soluble solids, sugar content, tannin content, pH, and firmness, during storage in a vacuum and MAP at 5 and 24 ∘C every 7 days to determine the shelf life using a non-destructive approach. TinyML-compatible regression models were employed to predict the stages of fruit development during the storage period. The decrease in the shelf life of the fruits begins when they transition from the Khalal stage to the Rutab stage, and the shelf life ends when they start to spoil or ripen to the Tamr stage. Low-cost Visible-Near-Infrared (VisNIR) spectral sensors (AS7265x-multi-spectral) were used to capture the internal physicochemical attributes of the fresh fruit. Regression models were employed for shelf life estimation. The findings indicated that vacuum and modified atmosphere packaging with 20% CO2 and N balance efficiently increased the shelf life of the stored fresh fruit to 53 days and 44 days, respectively, when maintained at 5 ∘C. However, the shelf life decreased to 44 and 23 days when the vacuum and modified atmosphere packaging with 20% CO2 and N balance were maintained at room temperature (24 ∘C). Edge Impulse supports the training and deployment of models on low-cost microcontrollers, which can be used to predict real-time estimations of the shelf life of fresh dates using TinyML sensors.

Validated Pretreatment Prediction Models for Response to Neoadjuvant Therapy in Patients with Rectal Cancer: A Systematic Review and Critical Appraisal

Pretreatment response prediction is crucial to select those patients with rectal cancer who will benefit from organ preservation strategies following (intensified) neoadjuvant therapy and to avoid unnecessary toxicity in those who will not. The combination of individual predictors in multivariable prediction models might improve predictive accuracy. The aim of this systematic review was to summarize and critically appraise validated pretreatment prediction models (other than radiomics-based models or image-based deep learning models) for response to neoadjuvant therapy in patients with rectal cancer and provide evidence-based recommendations for future research. MEDLINE via Ovid, Embase.com, and Scopus were searched for eligible studies published up to November 2022. A total of 5006 studies were screened and 16 were included for data extraction and risk of bias assessment using Prediction model Risk Of Bias Assessment Tool (PROBAST). All selected models were unique and grouped into five predictor categories: clinical, combined, genetics, metabolites, and pathology. Studies generally included patients with intermediate or advanced tumor stages who were treated with neoadjuvant chemoradiotherapy. Evaluated outcomes were pathological complete response and pathological tumor response. All studies were considered to have a high risk of bias and none of the models were externally validated in an independent study. Discriminative performances, estimated with the area under the curve (AUC), ranged per predictor category from 0.60 to 0.70 (clinical), 0.78 to 0.81 (combined), 0.66 to 0.91 (genetics), 0.54 to 0.80 (metabolites), and 0.71 to 0.91 (pathology). Model calibration outcomes were reported in five studies. Two collagen feature-based models showed the best predictive performance (AUCs 0.83-0.91 and good calibration). In conclusion, some pretreatment models for response prediction in rectal cancer show encouraging predictive potential but, given the high risk of bias in these studies, their value should be evaluated in future, well-designed studies.

A pooled analysis of the risk prediction models for mortality in acute exacerbation of chronic obstructive pulmonary disease

OBJECTIVE

The prognosis for acute exacerbation of chronic obstructive pulmonary disease (AECOPD) is not optimistic, and severe AECOPD leads to an increased risk of mortality. Prediction models help distinguish between high- and low-risk groups. At present, many prediction models have been established and validated, which need to be systematically reviewed to screen out more suitable models that can be used in the clinic and provide evidence for future research.

METHODS

We searched PubMed, EMBASE, Cochrane Library and Web of Science databases for studies on risk models for AECOPD mortality from their inception to 10 April 2022. The risk of bias was assessed using the prediction model risk of bias assessment tool (PROBAST). Stata software (version 16) was used to synthesize the C-statistics for each model.

RESULTS

A total of 37 studies were included. The development of risk prediction models for mortality in patients with AECOPD was described in 26 articles, in which the most common predictors were age (n = 17), dyspnea grade (n = 11), altered mental status (n = 8), pneumonia (n = 6) and blood urea nitrogen (BUN, n = 6). The remaining 11 articles only externally validated existing models. All 37 studies were evaluated at a high risk of bias using PROBAST. We performed a meta-analysis of five models included in 15 studies. DECAF (dyspnoea, eosinopenia, consolidation, acidemia and atrial fibrillation) performed well in predicting in-hospital death [C-statistic = 0.91, 95% confidence interval (CI): 0.83, 0.98] and 90-day death [C-statistic = 0.76, 95% CI: 0.69, 0.82] and CURB-65 (confusion, urea, respiratory rate, blood pressure and age) performed well in predicting 30-day death [C-statistic = 0.74, 95% CI: 0.70, 0.77].

CONCLUSIONS

This study provides information on the characteristics, performance and risk of bias of a risk model for AECOPD mortality. This pooled analysis of the present study suggests that the DECAF performs well in predicting in-hospital and 90-day deaths. Yet, external validation in different populations is still needed to prove this performance.

Prediction of In-hospital Mortality Among Intensive Care Unit Patients Using Modified Daily Laboratory-based Acute Physiology Score, Version 2

BACKGROUND

Mortality prediction for intensive care unit (ICU) patients frequently relies on single ICU admission acuity measures without accounting for subsequent clinical changes.

OBJECTIVE

Evaluate novel models incorporating modified admission and daily, time-updating Laboratory-based Acute Physiology Score, version 2 (LAPS2) to predict in-hospital mortality among ICU patients.

RESEARCH DESIGN

Retrospective cohort study.

PATIENTS

ICU patients in 5 hospitals from October 2017 through September 2019.

MEASURES

We used logistic regression, penalized logistic regression, and random forest models to predict in-hospital mortality within 30 days of ICU admission using admission LAPS2 alone in patient-level and patient-day-level models, or admission and daily LAPS2 at the patient-day level. Multivariable models included patient and admission characteristics. We performed internal-external validation using 4 hospitals for training and the fifth for validation, repeating analyses for each hospital as the validation set. We assessed performance using scaled Brier scores (SBS), c -statistics, and calibration plots.

RESULTS

The cohort included 13,993 patients and 107,699 ICU days. Across validation hospitals, patient-day-level models including daily LAPS2 (SBS: 0.119-0.235; c -statistic: 0.772-0.878) consistently outperformed models with admission LAPS2 alone in patient-level (SBS: 0.109-0.175; c -statistic: 0.768-0.867) and patient-day-level (SBS: 0.064-0.153; c -statistic: 0.714-0.861) models. Across all predicted mortalities, daily models were better calibrated than models with admission LAPS2 alone.

CONCLUSIONS

Patient-day-level models incorporating daily, time-updating LAPS2 to predict mortality among an ICU population performs as well or better than models incorporating modified admission LAPS2 alone. The use of daily LAPS2 may offer an improved tool for clinical prognostication and risk adjustment in research in this population.

Health Equity in Clinical Research Informatics

OBJECTIVES

Through a scoping review, we examine in this survey what ways health equity has been promoted in clinical research informatics with patient implications and especially published in the year of 2021 (and some in 2022).

METHOD

A scoping review was conducted guided by using methods described in the Joanna Briggs Institute Manual. The review process consisted of five stages: 1) development of aim and research question, 2) literature search, 3) literature screening and selection, 4) data extraction, and 5) accumulate and report results.

RESULTS

From the 478 identified papers in 2021 on the topic of clinical research informatics with focus on health equity as a patient implication, 8 papers met our inclusion criteria. All included papers focused on artificial intelligence (AI) technology. The papers addressed health equity in clinical research informatics either through the exposure of inequity in AI-based solutions or using AI as a tool for promoting health equity in the delivery of healthcare services. While algorithmic bias poses a risk to health equity within AI-based solutions, AI has also uncovered inequity in traditional treatment and demonstrated effective complements and alternatives that promotes health equity.

CONCLUSIONS

Clinical research informatics with implications for patients still face challenges of ethical nature and clinical value. However, used prudently-for the right purpose in the right context-clinical research informatics could bring powerful tools in advancing health equity in patient care.

Predicting Survival in Patients Presenting With Spinal Epidural Metastases: The Limburg Spinal Metastasis Score

BACKGROUND

Patients with spinal epidural metastases (SEM) often experience a reduction in ambulatory status and, thus, the quality of life. Predicting which patients will benefit from a surgical intervention remains a challenge. Life expectancy is an essential factor to be considered in surgical decision-making, although not the only one. Prediction models can add value in surgical decision-making. The goal of this study was to develop and internally validate a novel model (Limburg spinal metastases score [LSMS]) and compare the predictive value with 2 commonly used models: modified Bauer score and Oswestry Spinal Risk Index (OSRI).

METHODS

We retrospectively analyzed 144 consecutive patients who underwent surgical decompression for SEM in our centers between November 2006 and December 2020. Clinical and surgical parameters were evaluated. The novel prediction model was based on multivariate analysis and was internally validated. External validation of the 2 most commonly used prediction models was performed.

RESULTS

The median survival was 17 months, 55.7% of the immobile patients regained ambulation postoperatively. In 50 patients (34.7%), at least 1 complication occurred within 30 days after surgery. The LSMS consists of 4 parameters: primary tumor type, Karnofsky performance score, presence of visceral metastases, and presence of multiple spinal metastases. Bootstrap internal validation of the model developed on this cohort yielded an optimism-corrected c-statistic of 0.75 (95% CI: 0.71-0.80). The c-statistic of the OSRI score and the Bauer score was 0.69 (95% CI: 0.64-0.74) and 0.67 (95% CI: 0.62-0.72), respectively.

CONCLUSION

The LSMS consists of 4 parameters to assist surgical decision-making for patients with SEM. The score is easy to use and appears more accurate in our population in comparison with previous existing models.

CLINICAL RELEVANCE

A novel prediction model was developed to aid in surgical decision-making for patients with spinal epidural metastases.

LEVEL OF EVIDENCE: 3

Characterizing Ultrafine Particle Mobile Monitoring Data for Epidemiology

Mobile monitoring is increasingly used to assess exposure to traffic-related air pollutants (TRAPs), including ultrafine particles (UFPs). Due to the rapid spatial decrease in the concentration of UFPs and other TRAPs with distance from roadways, mobile measurements may be non-representative of residential exposures, which are commonly used for epidemiologic studies. Our goal was to develop, apply, and test one possible approach for using mobile measurements in exposure assessment for epidemiology. We used an absolute principal component score model to adjust the contribution of on-road sources in mobile measurements to provide exposure predictions representative of cohort locations. We then compared UFP predictions at residential locations from mobile on-road plume-adjusted versus stationary measurements to understand the contribution of mobile measurements and characterize their differences. We found that predictions from mobile measurements are more representative of cohort locations after down-weighting the contribution of localized on-road plumes. Further, predictions at cohort locations derived from mobile measurements incorporate more spatial variation compared to those from short-term stationary data. Sensitivity analyses suggest that this additional spatial information captures features in the exposure surface not identified from the stationary data alone. We recommend the correction of mobile measurements to create exposure predictions representative of residential exposure for epidemiology.

Development and Internal Validation of a Prediction Model for Falls Using Electronic Health Records in a Hospital Setting

OBJECTIVE

Fall prevention is important in many hospitals. Current fall-risk-screening tools have limited predictive accuracy specifically for older inpatients. Their administration can be time-consuming. A reliable and easy-to-administer tool is desirable to identify older inpatients at higher fall risk. We aimed to develop and internally validate a prognostic prediction model for inpatient falls for older patients.

DESIGN

Retrospective analysis of a large cohort drawn from hospital electronic health record data.

SETTING AND PARTICIPANTS

Older patients (≥70 years) admitted to a university medical center (2016 until 2021).

METHODS

The outcome was an inpatient fall (≥24 hours of admission). Two prediction models were developed using regularized logistic regression in 5 imputed data sets: one model without predictors indicating missing values (Model-without) and one model with these additional predictors indicating missing values (Model-with). We internally validated our whole model development strategy using 10-fold stratified cross-validation. The models were evaluated using discrimination (area under the receiver operating characteristic curve) and calibration (plot assessment). We determined whether the areas under the receiver operating characteristic curves (AUCs) of the models were significantly different using DeLong test.

RESULTS

Our data set included 21,286 admissions. In total, 470 (2.2%) had a fall after 24 hours of admission. The Model-without had 12 predictors and Model-with 13, of which 4 were indicators of missing values. The AUCs of the Model-without and Model-with were 0.676 (95% CI 0.646-0.707) and 0.695 (95% CI 0.667-0.724). The AUCs between both models were significantly different (P = .013). Calibration was good for both models.

CONCLUSIONS AND IMPLICATIONS

Both the Model-with and Model-without indicators of missing values showed good calibration and fair discrimination, where the Model-with performed better. Our models showed competitive performance to well-established fall-risk-screening tools, and they have the advantage of being based on routinely collected data. This may substantially reduce the burden on nurses, compared with nonautomatic fall-risk-screening tools.

An Apparent Diffusion Coefficient-Based Machine Learning Model Can Improve Prostate Cancer Detection in the Grey Area of the Prostate Imaging Reporting and Data System Category 3: A Single-Centre Experience

The Prostate Imaging and Reporting Data System (PI-RADS) has a key role in the management of prostate cancer (PCa). However, the clinical interpretation of PI-RADS 3 score lesions may be challenging and misleading, thus postponing PCa diagnosis to biopsy outcome. Multiparametric magnetic resonance imaging (mpMRI) radiomic analysis may represent a stand-alone noninvasive tool for PCa diagnosis. Hence, this study aims at developing a mpMRI-based radiomic PCa diagnostic model in a cohort of PI-RADS 3 lesions. We enrolled 133 patients with 155 PI-RADS 3 lesions, 84 of which had PCa confirmation by fusion biopsy. Local radiomic features were generated from apparent diffusion coefficient maps, and the four most informative were selected using LASSO, the Wilcoxon rank-sum test (p < 0.001), and support vector machines (SVMs). The selected features where augmented and used to train an SVM classifier, externally validated on a holdout subset. Linear and second-order polynomial kernels were exploited, and their predictive performance compared through receiver operating characteristics (ROC)-related metrics. On the test set, the highest performance, equally for both kernels, was specificity = 76%, sensitivity = 78%, positive predictive value = 80%, and negative predictive value = 74%. Our findings substantially improve radiologist interpretation of PI-RADS 3 lesions and let us advance towards an image-driven PCa diagnosis.

Development and economic assessment of machine learning models to predict glycosylated hemoglobin in type 2 diabetes

Background: Glycosylated hemoglobin (HbA1c) is recommended for diagnosing and monitoring type 2 diabetes. However, the monitoring frequency in real-world applications has not yet reached the recommended frequency in the guidelines. Developing machine learning models to screen patients with poor glycemic control in patients with T2D could optimize management and decrease medical service costs. Methods: This study was carried out on patients with T2D who were examined for HbA1c at the Sichuan Provincial People's Hospital from April 2018 to December 2019. Characteristics were extracted from interviews and electronic medical records. The data (excluded FBG or included FBG) were randomly divided into a training dataset and a test dataset with a radio of 8:2 after data pre-processing. Four imputing methods, four screening methods, and six machine learning algorithms were used to optimize data and develop models. Models were compared on the basis of predictive performance metrics, especially on the model benefit (MB, a confusion matrix combined with economic burden associated with therapeutic inertia). The contributions of features were interpreted using SHapley Additive exPlanation (SHAP). Finally, we validated the sample size on the best model. Results: The study included 980 patients with T2D, of whom 513 (52.3%) were defined as positive (need to perform the HbA1c test). The results indicated that the model trained in the data (included FBG) presented better forecast performance than the models that excluded the FBG value. The best model used modified random forest as the imputation method, ElasticNet as the feature screening method, and the LightGBM algorithms and had the best performance. The MB, AUC, and AUPRC of the best model, among a total of 192 trained models, were 43475.750 (¥), 0.972, 0.944, and 0.974, respectively. The FBG values, previous HbA1c values, having a rational and reasonable diet, health status scores, type of manufacturers of metformin, interval of measurement, EQ-5D scores, occupational status, and age were the most significant contributors to the prediction model. Conclusion: We found that MB could be an indicator to evaluate the model prediction performance. The proposed model performed well in identifying patients with T2D who need to undergo the HbA1c test and could help improve individualized T2D management.

Individualized Treatment Effects of Bougie versus Stylet for Tracheal Intubation in Critical Illness

Rationale: A recent randomized trial found that using a bougie did not increase the incidence of successful intubation on first attempt in critically ill adults. The average effect of treatment in a trial population, however, may differ from effects for individuals. Objective: We hypothesized that application of a machine learning model to data from a clinical trial could estimate the effect of treatment (bougie vs. stylet) for individual patients based on their baseline characteristics ("individualized treatment effects"). Methods: This was a secondary analysis of the BOUGIE (Bougie or Stylet in Patients Undergoing Intubation Emergently) trial. A causal forest algorithm was used to model differences in outcome probabilities by randomized group assignment (bougie vs. stylet) for each patient in the first half of the trial (training cohort). This model was used to predict individualized treatment effects for each patient in the second half (validation cohort). Measurements and Main Results: Of 1,102 patients in the BOUGIE trial, 558 (50.6%) were the training cohort, and 544 (49.4%) were the validation cohort. In the validation cohort, individualized treatment effects predicted by the model significantly modified the effect of trial group assignment on the primary outcome (P value for interaction = 0.02; adjusted qini coefficient, 2.46). The most important model variables were difficult airway characteristics, body mass index, and Acute Physiology and Chronic Health Evaluation II score. Conclusions: In this hypothesis-generating secondary analysis of a randomized trial with no average treatment effect and no treatment effect in any prespecified subgroups, a causal forest machine learning algorithm identified patients who appeared to benefit from the use of a bougie over a stylet and from the use of a stylet over a bougie using complex interactions between baseline patient and operator characteristics.

Using Machine-Learning to Assess the Prognostic Value of Early Enteral Feeding Intolerance in Critically Ill Patients: A Retrospective Study

BACKGROUND

The association between gastrointestinal intolerance during early enteral nutrition (EN) and adverse clinical outcomes in critically ill patients is controversial. We aimed to assess the prognostic value of enteral feeding intolerance (EFI) markers during early ICU stays and to predict early EN failure using a machine learning (ML) approach.

METHODS

We performed a retrospective analysis of data from adult patients admitted to Beilinson Hospital ICU between January 2011 and December 2018 for more than 48 h and received EN. Clinical data, including demographics, severity scores, EFI markers, and medications, along with 72 h after admission, were analyzed by ML algorithms. Prediction performance was assessed by the area under the receiver operating characteristics (AUCROC) of a ten-fold cross-validation set.

RESULTS

The datasets comprised 1584 patients. The means of the cross-validation AUCROCs for 90-day mortality and early EN failure were 0.73 (95% CI 0.71-0.75) and 0.71 (95% CI 0.67-0.74), respectively. Gastric residual volume above 250 mL on the second day was an important component of both prediction models.

CONCLUSIONS

ML underlined the EFI markers that predict poor 90-day outcomes and early EN failure and supports early recognition of at-risk patients. Results have to be confirmed in further prospective and external validation studies.

Prediction Models for Intrauterine Growth Restriction Using Artificial Intelligence and Machine Learning: A Systematic Review and Meta-Analysis

BACKGROUND

IntraUterine Growth Restriction (IUGR) is a global public health concern and has major implications for neonatal health. The early diagnosis of this condition is crucial for obtaining positive outcomes for the newborn. In recent years Artificial intelligence (AI) and machine learning (ML) techniques are being used to identify risk factors and provide early prediction of IUGR. We performed a systematic review (SR) and meta-analysis (MA) aimed to evaluate the use and performance of AI/ML models in detecting fetuses at risk of IUGR.

METHODS

We conducted a systematic review according to the PRISMA checklist. We searched for studies in all the principal medical databases (MEDLINE, EMBASE, CINAHL, Scopus, Web of Science, and Cochrane). To assess the quality of the studies we used the JBI and CASP tools. We performed a meta-analysis of the diagnostic test accuracy, along with the calculation of the pooled principal measures.

RESULTS

We included 20 studies reporting the use of AI/ML models for the prediction of IUGR. Out of these, 10 studies were used for the quantitative meta-analysis. The most common input variable to predict IUGR was the fetal heart rate variability (n = 8, 40%), followed by the biochemical or biological markers (n = 5, 25%), DNA profiling data (n = 2, 10%), Doppler indices (n = 3, 15%), MRI data (n = 1, 5%), and physiological, clinical, or socioeconomic data (n = 1, 5%). Overall, we found that AI/ML techniques could be effective in predicting and identifying fetuses at risk for IUGR during pregnancy with the following pooled overall diagnostic performance: sensitivity = 0.84 (95% CI 0.80-0.88), specificity = 0.87 (95% CI 0.83-0.90), positive predictive value = 0.78 (95% CI 0.68-0.86), negative predictive value = 0.91 (95% CI 0.86-0.94) and diagnostic odds ratio = 30.97 (95% CI 19.34-49.59). In detail, the RF-SVM (Random Forest-Support Vector Machine) model (with 97% accuracy) showed the best results in predicting IUGR from FHR parameters derived from CTG.

CONCLUSIONS

our findings showed that AI/ML could be part of a more accurate and cost-effective screening method for IUGR and be of help in optimizing pregnancy outcomes. However, before the introduction into clinical daily practice, an appropriate algorithmic improvement and refinement is needed, and the importance of quality assessment and uniform diagnostic criteria should be further emphasized.