Comparing machine learning algorithms for predicting ICU admission and mortality in COVID-19

The role of AI in emergency department triage: An integrative systematic review

BACKGROUND

Overcrowding in emergency departments (EDs) leads to delayed treatments, poor patient outcomes, and increased staff workloads. Artificial intelligence (AI) and machine learning (ML) have emerged as promising tools to optimize triage.

OBJECTIVE

This systematic review evaluates AI/ML-driven triage and risk stratification models in EDs, focusing on predictive performance, key predictors, clinical and operational outcomes, and implementation challenges.

METHODS

Following PRISMA 2020 guidelines, we systematically searched PubMed, CINAHL, Scopus, Web of Science, and IEEE Xplore for studies on AI/ML-driven ED triage published through January 2025. Two independent reviewers screened studies, extracted data, and assessed quality using PROBAST, with findings synthesized thematically.

RESULTS

Twenty-six studies met inclusion criteria. ML-based triage models consistently outperformed traditional tools, often achieving AUCs > 0.80 for high acuity outcomes (e.g., hospital admission, ICU transfer). Key predictors included vital signs, age, arrival mode, and disease-specific markers. Incorporating free-text data via natural language processing enhances accuracy and sensitivity. Advanced ML techniques, such as gradient boosting and random forests, generally surpassed simpler models across diverse populations. Reported benefits included reduced ED overcrowding, improved resource allocation, fewer mis-triaged patients, and potential patient outcome improvements.

CONCLUSION

AI/ML-based triage models hold substantial promise in improving ED efficiency and patient outcomes. Prospective, multi-center trials with transparent reporting and seamless electronic health record integration are essential to confirm these benefits.

IMPLICATIONS FOR CLINICAL PRACTICE

Integrating AI and ML into ED triage can enhance assessment accuracy and resource allocation. Early identification of high-risk patients supports better clinical decision-making, including critical care and ICU nurses, by streamlining patient transitions and reducing overcrowding. Explainable AI models foster trust and enable informed decisions under pressure. To realize these benefits, healthcare organizations must invest in robust infrastructure, provide comprehensive training for all clinical staff, and implement ethical, standardized practices that support interdisciplinary collaboration between ED and ICU teams.

Classification of the ICU Admission for COVID-19 Patients with Transfer Learning Models Using Chest X-Ray Images

Objectives: Predicting intensive care unit (ICU) admissions during pandemic outbreaks such as COVID-19 can assist clinicians in early intervention and the better allocation of medical resources. Artificial intelligence (AI) tools are promising for this task, but their development can be hindered by the limited availability of training data. This study aims to explore model development strategies in data-limited scenarios, specifically in detecting the need for ICU admission using chest X-rays of COVID-19 patients by leveraging transfer learning and data extension to improve model performance. Methods: We explored convolutional neural networks (CNNs) pre-trained on either natural images or chest X-rays, fine-tuning them on a relatively limited dataset (COVID-19-NY-SBU, n = 899) of lung-segmented X-ray images for ICU admission classification. To further address data scarcity, we introduced a dataset extension strategy that integrates an additional dataset (MIDRC-RICORD-1c, n = 417) with different but clinically relevant labels. Results: The TorchX-SBU-RSNA and ELIXR-SBU-RSNA models, leveraging X-ray-pre-trained models with our training data extension approach, enhanced ICU admission classification performance from a baseline AUC of 0.66 (56% sensitivity and 68% specificity) to AUCs of 0.77-0.78 (58-62% sensitivity and 78-80% specificity). The gradient-weighted class activation mapping (Grad-CAM) analysis demonstrated that the TorchX-SBU-RSNA model focused more precisely on the relevant lung regions and reduced the distractions from non-relevant areas compared to the natural image-pre-trained model without data expansion. Conclusions: This study demonstrates the benefits of medical image-specific pre-training and strategic dataset expansion in enhancing the model performance of imaging AI models. Moreover, this approach demonstrates the potential of using diverse but limited data sources to alleviate the limitations of model development for medical imaging AI. The developed AI models and training strategies may facilitate more effective and efficient patient management and resource allocation in future outbreaks of infectious respiratory diseases.

Utilizing machine learning to predict hospital admissions for pediatric COVID-19 patients (PrepCOVID-Machine)

The COVID-19 pandemic has burdened healthcare systems globally. To curb high hospital admission rates, only patients with genuine medical needs are admitted. However, machine learning (ML) models to predict COVID-19 hospitalization in Asian children are lacking. This study aimed to develop and validate ML models to predict pediatric COVID-19 hospitalization. We collected secondary data with 2200 patients and 65 variables from Malaysian aged 0 to 12 with COVID-19 between 1st February 2020 and 31st March 2022. The sample was partitioned into training, internal, and external validation groups. Recursive Feature Elimination (RFE) was employed for feature selection, and we trained seven supervised classifiers. Grid Search was used to optimize the hyperparameters of each algorithm. The study analyzed 1988 children and 30 study variables after data were processed. The RFE algorithm selected 12 highly predicted variables for COVID-19 hospitalization, including age, male sex, fever, cough, rhinorrhea, shortness of breath, vomiting, diarrhea, seizures, body temperature, chest indrawing, and abnormal breath sounds. With external validation, Adaptive Boosting was the highest-performing classifier (AUROC = 0.95) to predict COVID-19 hospital admission in children. We validated AdaBoost as the best to predict COVID-19 hospitalization among children. This model may assist front-line clinicians in making medical disposition decisions.

Predictive modelling and identification of critical variables of mortality risk in COVID-19 patients

South Africa was the most affected country in Africa by the coronavirus disease 2019 (COVID-19) pandemic, where over 4 million confirmed cases of COVID-19 and over 102,000 deaths have been recorded since 2019. Aside from clinical methods, artificial intelligence (AI)-based solutions such as machine learning (ML) models have been employed in treating COVID-19 cases. However, limited application of AI for COVID-19 in Africa has been reported in the literature. This study aimed to investigate the performance and interpretability of several ML algorithms, including deep multilayer perceptron (Deep MLP), support vector machine (SVM) and Extreme gradient boosting trees (XGBoost) for predicting COVID-19 mortality risk with an emphasis on the effect of cross-validation (CV) and principal component analysis (PCA) on the results. For this purpose, a dataset with 154 features from 490 COVID-19 patients admitted into the intensive care unit (ICU) of Tygerberg Hospital in Cape Town, South Africa, during the first wave of COVID-19 in 2020 was retrospectively analysed. Our results show that Deep MLP had the best overall performance (F1 = 0.92; area under the curve (AUC) = 0.94) when CV and the synthetic minority oversampling technique (SMOTE) were applied without PCA. By using the Shapley Additive exPlanations (SHAP) model to interpret the mortality risk predictions, we identified the Length of stay (LOS) in the hospital, LOS in the ICU, Time to ICU from admission, days discharged alive or death, D-dimer (blood clotting factor), and blood pH as the six most critical variables for mortality risk prediction. Also, Age at admission, Pf ratio (PaO2/FiO2 ratio), troponin T (TropT), ferritin, ventilation, C-reactive protein (CRP), and symptoms of acute respiratory distress syndrome (ARDS) were associated with the severity and fatality of COVID-19 cases. The study reveals how ML could assist medical practitioners in making informed decisions on handling critically ill COVID-19 patients with comorbidities. It also offers insight into the combined effect of CV, PCA, and SMOTE on the performance of ML models for COVID-19 mortality risk prediction, which has been little explored.

Prediction of prolonged mechanical ventilation in the intensive care unit via machine learning: a COVID-19 perspective

Early recognition of risk factors for prolonged mechanical ventilation (PMV) could allow for early clinical interventions, prevention of secondary complications such as nosocomial infections, and effective triage of hospital resources. This study tested the hypothesis that an ensemble machine learning (ML) analysis of clinical data at time of intubation could identify patients at risk of PMV, using a COVID-19 dataset to classify patients into PMV (> 14 days) and non-PMV (≤ 14 days) groups. While several factors are known to cause PMV, including acid-base, weakness, and delirium, lesser-utilized but routinely measured parameters such as platelet count, glucose levels and fevers may also be relevant. Patient data from a single University Hospital were analyzed via the ML workflow to predict patients at risk of PMV and identify key clinical markers. Model performance was evaluated on a chronologically distinct cohort. The ML workflow identified patients at risk of PMV with AUROCTRAIN=0.960 (F1TRAIN = 0.935) and AUROCTEST=0.804 (F1TEST = 0.800). Top key features for classification included glucose, platelet count, temperature, LVEF, bicarbonate (arterial blood gas), and BMI. Data analysis at intubation time via the proposed workflow offers the potential to accurately predict patients at risk of PMV, with the goal to improve patient management and triage of hospital resources.

Establishing a differential diagnosis model between primary membranous nephropathy and non-primary membranous nephropathy by machine learning algorithms

CONTEXT

Four algorithms with relatively balanced complexity and accuracy in deep learning classification algorithm were selected for differential diagnosis of primary membranous nephropathy (PMN).

OBJECTIVE

This study explored the most suitable classification algorithm for PMN identification, and to provide data reference for PMN diagnosis research.

METHODS

A total of 500 patients were referred to Luo-he Central Hospital from 2019 to 2021. All patients were diagnosed with primary glomerular disease confirmed by renal biopsy, contained 322 cases of PMN, the 178 cases of non-PMN. Using the decision tree, random forest, support vector machine, and extreme gradient boosting (Xgboost) to establish a differential diagnosis model for PMN and non-PMN. Based on the true positive rate, true negative rate, false-positive rate, false-negative rate, accuracy, feature work area under the curve (AUC) of subjects, the best performance of the model was chosen.

RESULTS

The efficiency of the Xgboost model based on the above evaluation indicators was the highest, which the diagnosis of PMN of the sensitivity and specificity, respectively 92% and 96%.

CONCLUSIONS

The differential diagnosis model for PMN was established successfully and the efficiency performance of the Xgboost model was the best. It could be used for the clinical diagnosis of PMN.

Prediction models for COVID-19 disease outcomes

SARS-CoV-2 has caused over 6.9 million deaths and continues to produce lasting health consequences. COVID-19 manifests broadly from no symptoms to death. In a retrospective cross-sectional study, we developed personalized risk assessment models that predict clinical outcomes for individuals with COVID-19 and inform targeted interventions. We sequenced viruses from SARS-CoV-2-positive nasopharyngeal swab samples between July 2020 and July 2022 from 4450 individuals in Missouri and retrieved associated disease courses, clinical history, and urban-rural classification. We integrated this data to develop machine learning-based predictive models to predict hospitalization, ICU admission, and long COVID.The mean age was 38.3 years (standard deviation = 21.4) with 55.2% (N = 2453) females and 44.8% (N = 1994) males (not reported, N = 4). Our analyses revealed a comprehensive set of predictors for each outcome, encompassing human, environment, and virus genome-wide genetic markers. Immunosuppression, cardiovascular disease, older age, cardiac, gastrointestinal, and constitutional symptoms, rural residence, and specific amino acid substitutions were associated with hospitalization. ICU admission was associated with acute respiratory distress syndrome, ventilation, bacterial co-infection, rural residence, and non-wild type SARS-CoV-2 variants. Finally, long COVID was associated with hospital admission, ventilation, and female sex.Overall, we developed risk assessment models that offer the capability to identify patients with COVID-19 necessitating enhanced monitoring or early interventions. Of importance, we demonstrate the value of including key elements of virus, host, and environmental factors to predict patient outcomes, serving as a valuable platform in the field of personalized medicine with the potential for adaptation to other infectious diseases.

Advantages of Metabolomics-Based Multivariate Machine Learning to Predict Disease Severity: Example of COVID

The COVID-19 outbreak caused saturations of hospitals, highlighting the importance of early patient triage to optimize resource prioritization. Herein, our objective was to test if high definition metabolomics, combined with ML, can improve prognostication and triage performance over standard clinical parameters using COVID infection as an example. Using high resolution mass spectrometry, we obtained metabolomics profiles of patients and combined them with clinical parameters to design machine learning (ML) algorithms predicting severity (herein determined as the need for mechanical ventilation during patient care). A total of 64 PCR-positive COVID patients at the Poitiers CHU were recruited. Clinical and metabolomics investigations were conducted 8 days after the onset of symptoms. We show that standard clinical parameters could predict severity with good performance (AUC of the ROC curve: 0.85), using SpO2, first respiratory rate, Horowitz quotient and age as the most important variables. However, the performance of the prediction was substantially improved by the use of metabolomics (AUC = 0.92). Our small-scale study demonstrates that metabolomics can improve the performance of diagnosis and prognosis algorithms, and thus be a key player in the future discovery of new biological signals. This technique is easily deployable in the clinic, and combined with machine learning, it can help design the mathematical models needed to advance towards personalized medicine.

A machine learning model to predict spontaneous vaginal delivery failure for term nulliparous women: An observational study

OBJECTIVE

This study aims to construct and evaluate a model to predict spontaneous vaginal delivery (SVD) failure in term nulliparous women based on machine learning algorithms.

METHODS

In this retrospective observational study, data on nulliparous women without contraindications for vaginal delivery with a singleton pregnancy ≥37 weeks and before the onset of labor from September 2020 to September 2021 were divided into a training set and a temporal validation set. Transperineal ultrasound was performed to collect angle of progression, head-perineum distance, subpubic arch angle, and their levator hiatal dimensions. The cervical length was measured via transvaginal ultrasound. The delivery methods were later recorded. Through LASSO regression analysis, indicators that can affect SVD failure were selected. Seven common machine learning algorithms were selected for model training, and the optimal algorithm was selected based on the area under the curve (AUC) to evaluate the effectiveness of the validation model.

RESULTS

Four indicators related to SVD failure were identified through LASSO regression screening: angle of progression, cervical length, subpubic arch angle, and estimated fetal weight. The Gaussian NB algorithm was found to yield the highest AUC (0.82, 95% confidence interval [CI] 0.65-0.98) during model training, and hence it was chosen for verification with the temporal validation set, in which an AUC of 0.79 (95% CI 0.64-0.95) was obtained with accuracy, sensitivity, and specificity rates of 80.9%, 72.7%, and 75.0%, respectively.

CONCLUSION

The Gaussian NB model showed good predictive effect, proving its potential as a clinical reference for predicting SVD failure of term nulliparous women before actual delivery.

Enhanced SARS-CoV-2 case prediction using public health data and machine learning models

Objectives

The goal of this study is to propose and test a scalable framework for machine learning (ML) algorithms to predict near-term severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) cases by incorporating and evaluating the impact of real-time dynamic public health data.

Materials and Methods

Data used in this study include patient-level results, procurement, and location information of all SARS-CoV-2 tests reported in West Virginia as part of their mandatory reporting system from January 2021 to March 2022. We propose a method for incorporating and comparing widely available public health metrics inside of a ML framework, specifically a long-short-term memory network, to forecast SARS-CoV-2 cases across various feature sets.

Results

Our approach provides better prediction of localized case counts and indicates the impact of the dynamic elements of the pandemic on predictions, such as the influence of the mixture of viral variants in the population and variable testing and vaccination rates during various eras of the pandemic.

Discussion

Utilizing real-time public health metrics, including estimated Rt from multiple SARS-CoV-2 variants, vaccination rates, and testing information, provided a significant increase in the accuracy of the model during the Omicron and Delta period, thus providing more precise forecasting of daily case counts at the county level. This work provides insights on the influence of various features on predictive performance in rural and non-rural areas.

Conclusion

Our proposed framework incorporates available public health metrics with operational data on the impact of testing, vaccination, and current viral variant mixtures in the population to provide a foundation for combining dynamic public health metrics and ML models to deliver forecasting and insights in healthcare domains. It also shows the importance of developing and deploying ML frameworks in rural settings.

Explainable artificial intelligence model for mortality risk prediction in the intensive care unit: a derivation and validation study

BACKGROUND

The lack of transparency is a prevalent issue among the current machine-learning (ML) algorithms utilized for predicting mortality risk. Herein, we aimed to improve transparency by utilizing the latest ML explicable technology, SHapley Additive exPlanation (SHAP), to develop a predictive model for critically ill patients.

METHODS

We extracted data from the Medical Information Mart for Intensive Care IV database, encompassing all intensive care unit admissions. We employed nine different methods to develop the models. The most accurate model, with the highest area under the receiver operating characteristic curve, was selected as the optimal model. Additionally, we used SHAP to explain the workings of the ML model.

RESULTS

The study included 21 395 critically ill patients, with a median age of 68 years (interquartile range, 56-79 years), and most patients were male (56.9%). The cohort was randomly split into a training set (N = 16 046) and a validation set (N = 5349). Among the nine models developed, the Random Forest model had the highest accuracy (87.62%) and the best area under the receiver operating characteristic curve value (0.89). The SHAP summary analysis showed that Glasgow Coma Scale, urine output, and blood urea nitrogen were the top three risk factors for outcome prediction. Furthermore, SHAP dependency analysis and SHAP force analysis were used to interpret the Random Forest model at the factor level and individual level, respectively.

CONCLUSION

A transparent ML model for predicting outcomes in critically ill patients using SHAP methodology is feasible and effective. SHAP values significantly improve the explainability of ML models.

Development and evaluation of regression tree models for predicting in-hospital mortality of a national registry of COVID-19 patients over six pandemic surges

BACKGROUND

Objective prognostic information is essential for good clinical decision making. In case of unknown diseases, scarcity of evidence and limited tacit knowledge prevent obtaining this information. Prediction models can be useful, but need to be not only evaluated on how well they predict, but also how stable these models are under fast changing circumstances with respect to development of the disease and the corresponding clinical response. This study aims to provide interpretable and actionable insights, particularly for clinicians. We developed and evaluated two regression tree predictive models for in-hospital mortality of COVID-19 patient at admission and 24 hours (24 h) after admission, using a national registry. We performed a retrospective analysis of observational routinely collected data.

METHODS

Two regression tree models were developed for admission and 24 h after admission. The complexity of the trees was managed via cross validation to prevent overfitting. The predictive ability of the model was assessed via bootstrapping using the Area under the Receiver-Operating-Characteristic curve, Brier score and calibration curves. The tree models were assessed on the stability of their probabilities and predictive ability, on the selected variables, and compared to a full-fledged logistic regression model that uses variable selection and variable transformations using splines. Participants included COVID-19 patients from all ICUs participating in the Dutch National Intensive Care Evaluation (NICE) registry, who were admitted at the ICU between February 27, 2020, and November 23, 2021. From the NICE registry, we included concerned demographic data, minimum and maximum values of physiological data in the first 24 h of ICU admission and diagnoses (reason for admission as well as comorbidities) for model development. The main outcome measure was in-hospital mortality. We additionally analysed the Length-of-Stay (LoS) per patient subgroup per survival status.

RESULTS

A total of 13,369 confirmed COVID-19 patients from 70 ICUs were included (with mortality rate of 28%). The optimism-corrected AUROC of the admission tree (with seven paths) was 0.72 (95% CI: 0.71-0.74) and of the 24 h tree (with 11 paths) was 0.74 (0.74-0.77). Both regression trees yielded good calibration and variable selection for both trees was stable. Patient subgroups comprising the tree paths had comparable survival probabilities as the full-fledged logistic regression model, survival probabilities were stable over six COVID-19 surges, and subgroups were shown to have added predictive value over the individual patient variables.

CONCLUSIONS

We developed and evaluated regression trees, which operate at par with a carefully crafted logistic regression model. The trees consist of homogenous subgroups of patients that are described by simple interpretable constraints on patient characteristics thereby facilitating shared decision-making.

Utilizing predictive machine-learning modelling unveils feature-based risk assessment system for hyperinflammatory patterns and infectious outcomes in polytrauma

Purpose

Earlier research has identified several potentially predictive features including biomarkers associated with trauma, which can be used to assess the risk for harmful outcomes of polytraumatized patients. These features encompass various aspects such as the nature and severity of the injury, accompanying health conditions, immune and inflammatory markers, and blood parameters linked to organ functioning, however their applicability is limited. Numerous indicators relevant to the patients` outcome are routinely gathered in the intensive care unit (ICU) and recorded in electronic medical records, rendering them suitable predictors for risk assessment of polytraumatized patients.

Methods

317 polytraumatized patients were included, and the influence of 29 clinical and biological features on the complication patterns for systemic inflammatory response syndrome (SIRS), pneumonia and sepsis were analyzed with a machine learning workflow including clustering, classification and explainability using SHapley Additive exPlanations (SHAP) values. The predictive ability of the analyzed features within three days after admission to the hospital were compared based on patient-specific outcomes using receiver-operating characteristics.

Results

A correlation and clustering analysis revealed that distinct patterns of injury and biomarker patterns were observed for the major complication classes. A k-means clustering suggested four different clusters based on the major complications SIRS, pneumonia and sepsis as well as a patient subgroup that developed no complications. For classification of the outcome groups with no complications, pneumonia and sepsis based on boosting ensemble classification, 90% were correctly classified as low-risk group (no complications). For the high-risk groups associated with development of pneumonia and sepsis, 80% of the patients were correctly identified. The explainability analysis with SHAP values identified the top-ranking features that had the largest impact on the development of adverse outcome patterns. For both investigated risk scenarios (infectious complications and long ICU stay) the most important features are SOFA score, Glasgow Coma Scale, lactate, GGT and hemoglobin blood concentration.

Conclusion

The machine learning-based identification of prognostic feature patterns in patients with traumatic injuries may improve tailoring personalized treatment modalities to mitigate the adverse outcomes in high-risk patient clusters.

Machine learning analyses identify multi-modal frailty factors that selectively discriminate four cohorts in the Alzheimer's disease spectrum: a COMPASS-ND study

BACKGROUND

Frailty indicators can operate in dynamic amalgamations of disease conditions, clinical symptoms, biomarkers, medical signals, cognitive characteristics, and even health beliefs and practices. This study is the first to evaluate which, among these multiple frailty-related indicators, are important and differential predictors of clinical cohorts that represent progression along an Alzheimer's disease (AD) spectrum. We applied machine-learning technology to such indicators in order to identify the leading predictors of three AD spectrum cohorts; viz., subjective cognitive impairment (SCI), mild cognitive impairment (MCI), and AD. The common benchmark was a cohort of cognitively unimpaired (CU) older adults.

METHODS

The four cohorts were from the cross-sectional Comprehensive Assessment of Neurodegeneration and Dementia dataset. We used random forest analysis (Python 3.7) to simultaneously test the relative importance of 83 multi-modal frailty indicators in discriminating the cohorts. We performed an explainable artificial intelligence method (Tree Shapley Additive exPlanation values) for deep interpretation of prediction effects.

RESULTS

We observed strong concurrent prediction results, with clusters varying across cohorts. The SCI model demonstrated excellent prediction accuracy (AUC = 0.89). Three leading predictors were poorer quality of life ([QoL]; memory), abnormal lymphocyte count, and abnormal neutrophil count. The MCI model demonstrated a similarly high AUC (0.88). Five leading predictors were poorer QoL (memory, leisure), male sex, abnormal lymphocyte count, and poorer self-rated eyesight. The AD model demonstrated outstanding prediction accuracy (AUC = 0.98). Ten leading predictors were poorer QoL (memory), reduced olfaction, male sex, increased dependence in activities of daily living (n = 6), and poorer visual contrast.

CONCLUSIONS

Both convergent and cohort-specific frailty factors discriminated the AD spectrum cohorts. Convergence was observed as all cohorts were marked by lower quality of life (memory), supporting recent research and clinical attention to subjective experiences of memory aging and their potentially broad ramifications. Diversity was displayed in that, of the 14 leading predictors extracted across models, 11 were selectively sensitive to one cohort. A morbidity intensity trend was indicated by an increasing number and diversity of predictors corresponding to clinical severity, especially in AD. Knowledge of differential deficit predictors across AD clinical cohorts may promote precision interventions.

A time-course prediction model of global COVID-19 mortality

Introduction

The COVID-19 pandemic has caused over 6 million deaths worldwide and is a significant cause of mortality. Mortality dynamics vary significantly by country due to pathogen, host, social and environmental factors, in addition to vaccination and treatments. However, there is limited data on the relative contribution of different explanatory variables, which may explain changes in mortality over time. We, therefore, created a predictive model using orthogonal machine learning techniques to attempt to quantify the contribution of static and dynamic variables over time.

Methods

A model was created using Partial Least Squares Regression trained on data from 2020 to rank order the significance and effect size of static variables on mortality per country. This model enables the prediction of mortality levels for countries based on demographics alone. Partial Least Squares Regression was then used to quantify how dynamic variables, including weather and non-pharmaceutical interventions, contributed to the overall mortality in 2020. Finally, mortality levels for the first 60 days of 2021 were predicted using rolling-window Elastic Net regression.

Results

This model allowed prediction of deaths per day and quantification of the degree of influence of included variables, accounting for timing of occurrence or implementation. We found that the most parsimonious model could be reduced to six variables; three policy-related variables - COVID-19 testing policy, canceled public events policy, workplace closing policy; in addition to three environmental variables - maximum temperature per day, minimum temperature per day, and the dewpoint temperature per day.

Conclusion

Country and population-level static and dynamic variables can be used to predict COVID-19 mortality, providing an example of how broad temporal data can inform a preparation and mitigation strategy for both COVID-19 and future pandemics and assist decision-makers by identifying population-level contributors, including interventions, that have the greatest influence in mitigating mortality, and optimizing the health and safety of populations.

Performance of an AI algorithm during the different phases of the COVID pandemics: what can we learn from the AI and vice versa

Background

Artificial intelligence (AI) has proved to be of great value in diagnosing and managing Sars-Cov-2 infection. ALFABETO (ALL-FAster-BEtter-TOgether) is a tool created to support healthcare professionals in the triage, mainly in optimizing hospital admissions.

Methods

The AI was trained during the pandemic's "first wave" (February-April 2020). Our aim was to assess the performance during the "third wave" of the pandemics (February-April 2021) and evaluate its evolution. The neural network proposed behavior (hospitalization vs home care) was compared with what was actually done. If there were discrepancies between ALFABETO's predictions and clinicians' decisions, the disease's progression was monitored. Clinical course was defined as "favorable/mild" if patients could be managed at home or in spoke centers and "unfavorable/severe" if patients need to be managed in a hub center.

Results

ALFABETO showed accuracy of 76%, AUROC of 83%; specificity was 78% and recall 74%. ALFABETO also showed high precision (88%). 81 hospitalized patients were incorrectly predicted to be in "home care" class. Among those "home-cared" by the AI and "hospitalized" by the clinicians, 3 out of 4 misclassified patients (76.5%) showed a favorable/mild clinical course. ALFABETO's performance matched the reports in literature.

Conclusions

The discrepancies mostly occurred when the AI predicted patients could stay at home but clinicians hospitalized them; these cases could be handled in spoke centers rather than hubs, and the discrepancies may aid clinicians in patient selection. The interaction between AI and human experience has the potential to improve both AI performance and our comprehension of pandemic management.

Artificial Intelligence and Machine Learning Applications in Sudden Cardiac Arrest Prediction and Management: A Comprehensive Review

PURPOSE OF REVIEW

This literature review aims to provide a comprehensive overview of the recent advances in prediction models and the deployment of AI and ML in the prediction of cardiopulmonary resuscitation (CPR) success. The objectives are to understand the role of AI and ML in healthcare, specifically in medical diagnosis, statistics, and precision medicine, and to explore their applications in predicting and managing sudden cardiac arrest outcomes, especially in the context of prehospital emergency care.

RECENT FINDINGS

The role of AI and ML in healthcare is expanding, with applications evident in medical diagnosis, statistics, and precision medicine. Deep learning is gaining prominence in radiomics and population health for disease risk prediction. There's a significant focus on the integration of AI and ML in prehospital emergency care, particularly in using ML algorithms for predicting outcomes in COVID-19 patients and enhancing the recognition of out-of-hospital cardiac arrest (OHCA). Furthermore, the combination of AI with automated external defibrillators (AEDs) shows potential in better detecting shockable rhythms during cardiac arrest incidents. AI and ML hold immense promise in revolutionizing the prediction and management of sudden cardiac arrest, hinting at improved survival rates and more efficient healthcare interventions in the future. Sudden cardiac arrest (SCA) continues to be a major global cause of death, with survival rates remaining low despite advanced first responder systems. The ongoing challenge is the prediction and prevention of SCA. However, with the rise in the adoption of AI and ML tools in clinical electrophysiology in recent times, there is optimism about addressing these challenges more effectively.

Machine learning algorithms for predicting determinants of COVID-19 mortality in South Africa

Background

COVID-19 has strained healthcare resources, necessitating efficient prognostication to triage patients effectively. This study quantified COVID-19 risk factors and predicted COVID-19 intensive care unit (ICU) mortality in South Africa based on machine learning algorithms.

Methods

Data for this study were obtained from 392 COVID-19 ICU patients enrolled between 26 March 2020 and 10 February 2021. We used an artificial neural network (ANN) and random forest (RF) to predict mortality among ICU patients and a semi-parametric logistic regression with nine covariates, including a grouping variable based on K-means clustering. Further evaluation of the algorithms was performed using sensitivity, accuracy, specificity, and Cohen's K statistics.

Results

From the semi-parametric logistic regression and ANN variable importance, age, gender, cluster, presence of severe symptoms, being on the ventilator, and comorbidities of asthma significantly contributed to ICU death. In particular, the odds of mortality were six times higher among asthmatic patients than non-asthmatic patients. In univariable and multivariate regression, advanced age, PF1 and 2, FiO2, severe symptoms, asthma, oxygen saturation, and cluster 4 were strongly predictive of mortality. The RF model revealed that intubation status, age, cluster, diabetes, and hypertension were the top five significant predictors of mortality. The ANN performed well with an accuracy of 71%, a precision of 83%, an F1 score of 100%, Matthew's correlation coefficient (MCC) score of 100%, and a recall of 88%. In addition, Cohen's k-value of 0.75 verified the most extreme discriminative power of the ANN. In comparison, the RF model provided a 76% recall, an 87% precision, and a 65% MCC.

Conclusion

Based on the findings, we can conclude that both ANN and RF can predict COVID-19 mortality in the ICU with accuracy. The proposed models accurately predict the prognosis of COVID-19 patients after diagnosis. The models can be used to prioritize COVID-19 patients with a high mortality risk in resource-constrained ICUs.

The prognostic value of estimated glomerular filtration rate on admission for death within 30 days among COVID-19 inpatients using fractional polynomial and spline smoothing

BACKGROUND

The common regression models included estimated glomerular filtration rate (eGFR) in the continuous and categorical form for predicting the mortality in COVID-19 inpatients. However, the relationship may be non-linear, and categorizing implies a loss of information. This study aimed to assess the effect of eGFR on admission on death within 30 days among COVID-19 inpatients using flexible and smooth transformations of eGFR and compare the results against the common models.

METHODS

A retrospective study was conducted on hospitalized COVID-19 patients between April 2019 and July 2019 in Hamadan, Western Iran. The effect of eGFR on the death within 30 days was evaluated using different modeling: categorization, linear, unrestricted cubic spline (USC) with 4 knots, and fractional polynomial (FP). The results adjusted for older age and intensive care unit (ICU) admission. Discrimination power and model performance of the best-fitting model was evaluated using the area under the ROC (AUROC) and Brier score.

RESULTS

In total, 2945 patients (median age 61 years; interquartile range 48-73 years) were included, of whom the mortality rate was 9.23%. The relationship between the eGFR and death within 30 days is non-linear, so the degree-2 FP with powers (- 2, - 1) is the best-fitting model. Using the FP model, the risk increased exponentially in eGFR < 45 and then increased linearly and slowly. The AUROC of the FP model involving eGFR, older age, and ICU admission was 0.92 (95% CI 0.90-0.93) with a Brier score of 0.09.

CONCLUSION

There is a non-linear and asymmetric relationship between eGFR and death within 30 days among COVID-19 inpatients. Kidney function can be measured in COCID-19 patients on admission to know better understanding about prognosis of the patients.

Development and validation of a machine learning model to predict prognosis in HIV-negative cryptococcal meningitis patients: a multicenter study

PURPOSE

To predict prognosis in HIV-negative cryptococcal meningitis (CM) patients by developing and validating a machine learning (ML) model.

METHODS

This study involved 523 HIV-negative CM patients diagnosed between January 1, 1998, and August 31, 2022, by neurologists from 3 tertiary Chinese centers. Prognosis was evaluated at 10 weeks after the initiation of antifungal therapy.

RESULTS

The final prediction model for HIV-negative CM patients comprised 8 variables: Cerebrospinal fluid (CSF) cryptococcal count, CSF white blood cell (WBC), altered mental status, hearing impairment, CSF chloride levels, CSF opening pressure (OP), aspartate aminotransferase levels at admission, and decreased rate of CSF cryptococcal count within 2 weeks after admission. The areas under the curve (AUCs) in the internal, temporal, and external validation sets were 0.87 (95% CI 0.794-0.944), 0.92 (95% CI 0.795-1.000), and 0.86 (95% CI 0.744-0.975), respectively. An artificial intelligence (AI) model was trained to detect and count cryptococci, and the mean average precision (mAP) was 0.993.

CONCLUSION

A ML model for predicting prognosis in HIV-negative CM patients was built and validated, and the model might provide a reference for personalized treatment of HIV-negative CM patients. The change in the CSF cryptococcal count in the early phase of HIV-negative CM treatment can reflect the prognosis of the disease. In addition, utilizing AI to detect and count CSF cryptococci in HIV-negative CM patients can eliminate the interference of human factors in detecting cryptococci in CSF samples and reduce the workload of the examiner.

Machine Learning for Benchmarking Critical Care Outcomes

OBJECTIVES

Enhancing critical care efficacy involves evaluating and improving system functioning. Benchmarking, a retrospective comparison of results against standards, aids risk-adjusted assessment and helps healthcare providers identify areas for improvement based on observed and predicted outcomes. The last two decades have seen the development of several models using machine learning (ML) for clinical outcome prediction. ML is a field of artificial intelligence focused on creating algorithms that enable computers to learn from and make predictions or decisions based on data. This narrative review centers on key discoveries and outcomes to aid clinicians and researchers in selecting the optimal methodology for critical care benchmarking using ML.

METHODS

We used PubMed to search the literature from 2003 to 2023 regarding predictive models utilizing ML for mortality (592 articles), length of stay (143 articles), or mechanical ventilation (195 articles). We supplemented the PubMed search with Google Scholar, making sure relevant articles were included. Given the narrative style, papers in the cohort were manually curated for a comprehensive reader perspective.

RESULTS

Our report presents comparative results for benchmarked outcomes and emphasizes advancements in feature types, preprocessing, model selection, and validation. It showcases instances where ML effectively tackled critical care outcome-prediction challenges, including nonlinear relationships, class imbalances, missing data, and documentation variability, leading to enhanced results.

CONCLUSIONS

Although ML has provided novel tools to improve the benchmarking of critical care outcomes, areas that require further research include class imbalance, fairness, improved calibration, generalizability, and long-term validation of published models.

Improved patient mortality predictions in emergency departments with deep learning data-synthesis and ensemble models

The triage process in emergency departments (EDs) relies on the subjective assessment of medical practitioners, making it unreliable in certain aspects. There is a need for a more accurate and objective algorithm to determine the urgency of patients. This paper explores the application of advanced data-synthesis algorithms, machine learning (ML) algorithms, and ensemble models to predict patient mortality. Patients predicted to be at risk of mortality are in a highly critical condition, signifying an urgent need for immediate medical intervention. This paper aims to determine the most effective method for predicting mortality by enhancing the F1 score while maintaining high area under the receiver operating characteristic curve (AUC) score. This study used a dataset of 7325 patients who visited the Yonsei Severance Hospital's ED, located in Seoul, South Korea. The patients were divided into two groups: patients who deceased in the ED and patients who didn't. Various data-synthesis techniques, such as SMOTE, ADASYN, CTGAN, TVAE, CopulaGAN, and Gaussian Copula, were deployed to generate synthetic patient data. Twenty two ML models were then utilized, including tree-based algorithms like Decision tree, AdaBoost, LightGBM, CatBoost, XGBoost, NGBoost, TabNet, which are deep neural network algorithms, and statistical algorithms such as Support Vector Machine, Logistic Regression, Random Forest, k-nearest neighbors, and Gaussian Naive Bayes, as well as Ensemble Models which use the results from the ML models. Based on 21 patient information features used in the pandemic influenza triage algorithm (PITA), the models explained previously were applied to aim for the prediction of patient mortality. In evaluating ML algorithms using an imbalanced medical dataset, conventional metrics like accuracy scores or AUC can be misleading. This paper emphasizes the importance of using the F1 score as the primary performance measure, focusing on recall and specificity in detecting patient mortality. The highest-ranked model for predicting mortality utilized the Gaussian Copula data-synthesis technique and the CatBoost classifier, achieving an AUC of 0.9731 and an F1 score of 0.7059. These findings highlight the effectiveness of machine learning algorithms and data-synthesis techniques in improving the prediction performance of mortality in EDs.

Prediction of prognosis in COVID-19 patients using machine learning: A systematic review and meta-analysis

BACKGROUND

Accurate prediction of prognostic outcomes in patients with COVID-19 could facilitate clinical decision-making and medical resource allocation. However, little is known about the ability of machine learning (ML) to predict prognosis in COVID-19 patients.

OBJECTIVE

This study aimed to systematically examine the prognostic value of ML in patients with COVID-19.

METHODS

A systematic search was conducted in PubMed, Web of Science, Embase, Cochrane Library, and IEEE Xplore up to December 15, 2021. Studies predicting the prognostic outcomes of COVID-19 patients using ML were eligible for inclusion. Risk of bias was evaluated by a tailored checklist based on Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Pooled sensitivity, specificity, and area under the receiver operating curve (AUC) were calculated to evaluate model performance.

RESULTS

A total of 33 studies that described 35 models were eligible for inclusion, with 27 models presenting mortality, four intensive care unit (ICU) admission, and four use of ventilation. For predicting mortality, ML gave a pooled sensitivity of 0.86 (95% CI, 0.79-0.90), a specificity of 0.87 (95% CI, 0.80-0.92), and an AUC of 0.93 (95% CI, 0.90-0.95). For the prediction of ICU admission, ML had a sensitivity of 0.86 (95% CI, 0.78-0.92), a specificity of 0.81 (95% CI, 0.66-0.91), and an AUC of 0.91 (95% CI, 0.88-0.93). For the prediction of ventilation, ML had a sensitivity of 0.81 (95% CI, 0.68-0.90), a specificity of 0.78 (95% CI, 0.66-0.87), and an AUC of 0.87 (95% CI, 0.83-0.89). Meta-regression analyses indicated that algorithm, population, study design, and source of dataset influenced the pooled estimate.

CONCLUSION

This meta-analysis demonstrated the satisfactory performance of ML in predicting prognostic outcomes in patients with COVID-19, suggesting the potential value of ML to support clinical decision-making. However, improvements to methodology and validation are still necessary before its application in routine clinical practice.

Improving Mortality Risk Prediction with Routine Clinical Data: A Practical Machine Learning Model Based on eICU Patients

Purpose

Mortality risk prediction helps clinicians make better decisions in patient healthcare. However, existing severity scoring systems or algorithms used in intensive care units (ICUs) often rely on laborious manual collection of complex variables and lack sufficient validation in diverse clinical environments, thus limiting their practical applicability. This study aims to evaluate the performance of machine learning models that utilize routinely collected clinical data for short-term mortality risk prediction.

Patients and Methods

Using the eICU Collaborative Research Database, we identified a cohort of 12,393 ICU patients, who were randomly divided into a training group and a validation group at a ratio of 9:1. The models utilized routine variables obtained from regular medical workflows, including age, gender, physiological measurements, and usage of vasoactive medications within a 24-hour period prior to patient discharge. Four different machine learning algorithms, namely logistic regression, random forest, extreme gradient boosting (XGboost), and artificial neural network were employed to develop the mortality risk prediction model. We compared the discrimination and calibration performance of these models in assessing mortality risk within 1-week time window.

Results

Among the tested models, the XGBoost algorithm demonstrated the highest performance, with an area under the receiver operating characteristic curve (AUROC) of 0.9702, an area under precision and recall curves (AUPRC) of 0.8517, and a favorable Brier score of 0.0259 for 24-hour mortality risk prediction. Although the model's performance decreased when considering larger time windows, it still achieved a comparable AUROC of 0.9184 and AUPRC of 0.5519 for 3-day mortality risk prediction.

Conclusion

The findings demonstrate the feasibility of developing a highly accurate and well-calibrated model based on the XGBoost algorithm for short-term mortality risk prediction with easily accessible and interpretative data. These results enhance confidence in the application of the machine learning model to clinical practice.

Improved performance of machine learning models in predicting length of stay, discharge disposition, and inpatient mortality after total knee arthroplasty using patient-specific variables

BACKGROUND

This study aimed to compare the performance of ten predictive models using different machine learning (ML) algorithms and compare the performance of models developed using patient-specific vs. situational variables in predicting select outcomes after primary TKA.

METHODS

Data from 2016 to 2017 from the National Inpatient Sample were used to identify 305,577 discharges undergoing primary TKA, which were included in the training, testing, and validation of 10 ML models. 15 predictive variables consisting of 8 patient-specific and 7 situational variables were utilized to predict length of stay (LOS), discharge disposition, and mortality. Using the best performing algorithms, models trained using either 8 patient-specific and 7 situational variables were then developed and compared.

RESULTS

For models developed using all 15 variables, Linear Support Vector Machine (LSVM) was the most responsive model for predicting LOS. LSVM and XGT Boost Tree were equivalently most responsive for predicting discharge disposition. LSVM and XGT Boost Linear were equivalently most responsive for predicting mortality. Decision List, CHAID, and LSVM were the most reliable models for predicting LOS and discharge disposition, while XGT Boost Tree, Decision List, LSVM, and CHAID were most reliable for mortality. Models developed using the 8 patient-specific variables outperformed those developed using the 7 situational variables, with few exceptions.

CONCLUSION

This study revealed that performance of different models varied, ranging from poor to excellent, and demonstrated that models developed using patient-specific variables were typically better predictive of quality metrics after TKA than those developed employing situational variables.

LEVEL OF EVIDENCE

III.

Predicting intensive care need for COVID-19 patients using deep learning on chest radiography

Purpose

Image-based prediction of coronavirus disease 2019 (COVID-19) severity and resource needs can be an important means to address the COVID-19 pandemic. In this study, we propose an artificial intelligence/machine learning (AI/ML) COVID-19 prognosis method to predict patients' needs for intensive care by analyzing chest X-ray radiography (CXR) images using deep learning.

Approach

The dataset consisted of 8357 CXR exams from 5046 COVID-19-positive patients as confirmed by reverse transcription polymerase chain reaction (RT-PCR) tests for the SARS-CoV-2 virus with a training/validation/test split of 64%/16%/20% on a by patient level. Our model involved a DenseNet121 network with a sequential transfer learning technique employed to train on a sequence of gradually more specific and complex tasks: (1) fine-tuning a model pretrained on ImageNet using a previously established CXR dataset with a broad spectrum of pathologies; (2) refining on another established dataset to detect pneumonia; and (3) fine-tuning using our in-house training/validation datasets to predict patients' needs for intensive care within 24, 48, 72, and 96 h following the CXR exams. The classification performances were evaluated on our independent test set (CXR exams of 1048 patients) using the area under the receiver operating characteristic curve (AUC) as the figure of merit in the task of distinguishing between those COVID-19-positive patients who required intensive care following the imaging exam and those who did not.

Results

Our proposed AI/ML model achieved an AUC (95% confidence interval) of 0.78 (0.74, 0.81) when predicting the need for intensive care 24 h in advance, and at least 0.76 (0.73, 0.80) for 48 h or more in advance using predictions based on the AI prognostic marker derived from CXR images.

Conclusions

This AI/ML prediction model for patients' needs for intensive care has the potential to support both clinical decision-making and resource management.

Multivariate time series short term forecasting using cumulative data of coronavirus

Coronavirus emerged as a highly contagious, pathogenic virus that severely affects the respiratory system of humans. The epidemic-related data is collected regularly, which machine learning algorithms can employ to comprehend and estimate valuable information. The analysis of the gathered data through time series approaches may assist in developing more accurate forecasting models and strategies to combat the disease. This paper focuses on short-term forecasting of cumulative reported incidences and mortality. Forecasting is conducted utilizing state-of-the-art mathematical and deep learning models for multivariate time series forecasting, including extended susceptible-exposed-infected-recovered (SEIR), long-short-term memory (LSTM), and vector autoregression (VAR). The SEIR model has been extended by integrating additional information such as hospitalization, mortality, vaccination, and quarantine incidences. Extensive experiments have been conducted to compare deep learning and mathematical models that enable us to estimate fatalities and incidences more precisely based on mortality in the eight most affected nations during the time of this research. The metrics like mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE) are employed to gauge the model's effectiveness. The deep learning model LSTM outperformed all others in terms of forecasting accuracy. Additionally, the study explores the impact of vaccination on reported epidemics and deaths worldwide. Furthermore, the detrimental effects of ambient temperature and relative humidity on pathogenic virus dissemination have been analyzed.

Predicting 1-year mortality of patients with diabetes mellitus in Kazakhstan based on administrative health data using machine learning

Diabetes mellitus (DM) affects the quality of life and leads to disability, high morbidity, and premature mortality. DM is a risk factor for cardiovascular, neurological, and renal diseases, and places a major burden on healthcare systems globally. Predicting the one-year mortality of patients with DM can considerably help clinicians tailor treatments to patients at risk. In this study, we aimed to show the feasibility of predicting the one-year mortality of DM patients based on administrative health data. We use clinical data for 472,950 patients that were admitted to hospitals across Kazakhstan between mid-2014 to December 2019 and were diagnosed with DM. The data was divided into four yearly-specific cohorts (2016-, 2017-, 2018-, and 2019-cohorts) to predict mortality within a specific year based on clinical and demographic information collected up to the end of the preceding year. We then develop a comprehensive machine learning platform to construct a predictive model of one-year mortality for each year-specific cohort. In particular, the study implements and compares the performance of nine classification rules for predicting the one-year mortality of DM patients. The results show that gradient-boosting ensemble learning methods perform better than other algorithms across all year-specific cohorts while achieving an area under the curve (AUC) between 0.78 and 0.80 on independent test sets. The feature importance analysis conducted by calculating SHAP (SHapley Additive exPlanations) values shows that age, duration of diabetes, hypertension, and sex are the top four most important features for predicting one-year mortality. In conclusion, the results show that it is possible to use machine learning to build accurate predictive models of one-year mortality for DM patients based on administrative health data. In the future, integrating this information with laboratory data or patients' medical history could potentially boost the performance of the predictive models.

A Survey of COVID-19 Diagnosis Using Routine Blood Tests with the Aid of Artificial Intelligence Techniques

Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), causing a disease called COVID-19, is a class of acute respiratory syndrome that has considerably affected the global economy and healthcare system. This virus is diagnosed using a traditional technique known as the Reverse Transcription Polymerase Chain Reaction (RT-PCR) test. However, RT-PCR customarily outputs a lot of false-negative and incorrect results. Current works indicate that COVID-19 can also be diagnosed using imaging resolutions, including CT scans, X-rays, and blood tests. Nevertheless, X-rays and CT scans cannot always be used for patient screening because of high costs, radiation doses, and an insufficient number of devices. Therefore, there is a requirement for a less expensive and faster diagnostic model to recognize the positive and negative cases of COVID-19. Blood tests are easily performed and cost less than RT-PCR and imaging tests. Since biochemical parameters in routine blood tests vary during the COVID-19 infection, they may supply physicians with exact information about the diagnosis of COVID-19. This study reviewed some newly emerging artificial intelligence (AI)-based methods to diagnose COVID-19 using routine blood tests. We gathered information about research resources and inspected 92 articles that were carefully chosen from a variety of publishers, such as IEEE, Springer, Elsevier, and MDPI. Then, these 92 studies are classified into two tables which contain articles that use machine Learning and deep Learning models to diagnose COVID-19 while using routine blood test datasets. In these studies, for diagnosing COVID-19, Random Forest and logistic regression are the most widely used machine learning methods and the most widely used performance metrics are accuracy, sensitivity, specificity, and AUC. Finally, we conclude by discussing and analyzing these studies which use machine learning and deep learning models and routine blood test datasets for COVID-19 detection. This survey can be the starting point for a novice-/beginner-level researcher to perform on COVID-19 classification.

Fusion of imaging and non-imaging data for disease trajectory prediction for coronavirus disease 2019 patients

Purpose

Our study investigates whether graph-based fusion of imaging data with non-imaging electronic health records (EHR) data can improve the prediction of the disease trajectories for patients with coronavirus disease 2019 (COVID-19) beyond the prediction performance of only imaging or non-imaging EHR data.

Approach

We present a fusion framework for fine-grained clinical outcome prediction [discharge, intensive care unit (ICU) admission, or death] that fuses imaging and non-imaging information using a similarity-based graph structure. Node features are represented by image embedding, and edges are encoded with clinical or demographic similarity.

Results

Experiments on data collected from the Emory Healthcare Network indicate that our fusion modeling scheme performs consistently better than predictive models developed using only imaging or non-imaging features, with area under the receiver operating characteristics curve of 0.76, 0.90, and 0.75 for discharge from hospital, mortality, and ICU admission, respectively. External validation was performed on data collected from the Mayo Clinic. Our scheme highlights known biases in the model prediction, such as bias against patients with alcohol abuse history and bias based on insurance status.

Conclusions

Our study signifies the importance of the fusion of multiple data modalities for the accurate prediction of clinical trajectories. The proposed graph structure can model relationships between patients based on non-imaging EHR data, and graph convolutional networks can fuse this relationship information with imaging data to effectively predict future disease trajectory more effectively than models employing only imaging or non-imaging data. Our graph-based fusion modeling frameworks can be easily extended to other prediction tasks to efficiently combine imaging data with non-imaging clinical data.

Predicting the negative conversion time of nonsevere COVID-19 patients using machine learning methods

Based on the patient's clinical characteristics and laboratory indicators, different machine-learning methods were used to develop models for predicting the negative conversion time of nonsevere coronavirus disease 2019 (COVID-19) patients. A retrospective analysis was performed on 376 nonsevere COVID-19 patients admitted to Wuxi Fifth People's Hospital from May 2, 2022, to May 14, 2022. The patients were divided into training set (n = 309) and test set (n = 67). The clinical features and laboratory parameters of the patients were collected. In the training set, the least absolute shrinkage and selection operator (LASSO) was used to select predictive features and train six machine learning models: multiple linear regression (MLR), K-Nearest Neighbors Regression (KNNR), random forest regression (RFR), support vector machine regression (SVR), XGBoost regression (XGBR), and multilayer perceptron regression (MLPR). Seven best predictive features selected by LASSO included: age, gender, vaccination status, IgG, lymphocyte ratio, monocyte ratio, and lymphocyte count. The predictive performance of the models in the test set was MLPR > SVR > MLR > KNNR > XGBR > RFR, and MLPR had the strongest generalization performance, which is significantly better than SVR and MLR. In the MLPR model, vaccination status, IgG, lymphocyte count, and lymphocyte ratio were protective factors for negative conversion time; male gender, age, and monocyte ratio were risk factors. The top three features with the highest weights were vaccination status, gender, and IgG. Machine learning methods (especially MLPR) can effectively predict the negative conversion time of non-severe COVID-19 patients. It can help to rationally allocate limited medical resources and prevent disease transmission, especially during the Omicron pandemic.

Prognostic models in COVID-19 infection that predict severity: a systematic review

Current evidence on COVID-19 prognostic models is inconsistent and clinical applicability remains controversial. We performed a systematic review to summarize and critically appraise the available studies that have developed, assessed and/or validated prognostic models of COVID-19 predicting health outcomes. We searched six bibliographic databases to identify published articles that investigated univariable and multivariable prognostic models predicting adverse outcomes in adult COVID-19 patients, including intensive care unit (ICU) admission, intubation, high-flow nasal therapy (HFNT), extracorporeal membrane oxygenation (ECMO) and mortality. We identified and assessed 314 eligible articles from more than 40 countries, with 152 of these studies presenting mortality, 66 progression to severe or critical illness, 35 mortality and ICU admission combined, 17 ICU admission only, while the remaining 44 studies reported prediction models for mechanical ventilation (MV) or a combination of multiple outcomes. The sample size of included studies varied from 11 to 7,704,171 participants, with a mean age ranging from 18 to 93 years. There were 353 prognostic models investigated, with area under the curve (AUC) ranging from 0.44 to 0.99. A great proportion of studies (61.5%, 193 out of 314) performed internal or external validation or replication. In 312 (99.4%) studies, prognostic models were reported to be at high risk of bias due to uncertainties and challenges surrounding methodological rigor, sampling, handling of missing data, failure to deal with overfitting and heterogeneous definitions of COVID-19 and severity outcomes. While several clinical prognostic models for COVID-19 have been described in the literature, they are limited in generalizability and/or applicability due to deficiencies in addressing fundamental statistical and methodological concerns. Future large, multi-centric and well-designed prognostic prospective studies are needed to clarify remaining uncertainties.

Predicting the mortality of patients with Covid-19: A machine learning approach

Background and Aims

Infection with Covid-19 disease can lead to mortality in a short time. Early prediction of the mortality during an epidemic disease can save patients' lives through taking timely and necessary care interventions. Therefore, predicting the mortality of patients with Covid-19 using machine learning techniques can be effective in reducing mortality rate in Covid-19. The aim of this study is to compare four machine-learning algorithm for predicting mortality in Covid-19 disease.

Methods

The data of this study were collected from hospitalized patients with COVID-19 in five hospitals settings in Tehran (Iran). Database contained 4120 records, about 25% of which belonged to patients who died due to Covid-19. Each record contained 38 variables. Four machine-learning techniques, including random forest (RF), regression logistic (RL), gradient boosting tree (GBT), and support vector machine (SVM) were used in modeling.

Results

GBT model presented higher performance compared to other models (accuracy 70%, sensitivity 77%, specificity 69%, and the ROC area under the curve 0.857). RF, RL, and SVM models with the ROC area under curve 0.836, 0.818, and 0.794 were in the second and third places.

Conclusion

Considering the combination of multiple influential factors affecting death Covid-19 can help in early prediction and providing a better care plan. In addition, using different modeling on data can be useful for physician in providing appropriate care.

A machine learning analysis of correlates of mortality among patients hospitalized with COVID-19

It is vital to determine how patient characteristics that precede COVID-19 illness relate to COVID-19 mortality. This is a retrospective cohort study of patients hospitalized with COVID-19 across 21 healthcare systems in the US. All patients (N = 145,944) had COVID-19 diagnoses and/or positive PCR tests and completed their hospital stays from February 1, 2020 through January 31, 2022. Machine learning analyses revealed that age, hypertension, insurance status, and healthcare system (hospital site) were especially predictive of mortality across the full sample. However, multiple variables were especially predictive in subgroups of patients. The nested effects of risk factors such as age, hypertension, vaccination, site, and race accounted for large differences in mortality likelihood with rates ranging from about 2-30%. Subgroups of patients are at heightened risk of COVID-19 mortality due to combinations of preadmission risk factors; a finding of potential relevance to outreach and preventive actions.

Novel architecture for gated recurrent unit autoencoder trained on time series from electronic health records enables detection of ICU patient subgroups

Electronic health records (EHRs) are used in hospitals to store diagnoses, clinician notes, examinations, lab results, and interventions for each patient. Grouping patients into distinct subsets, for example, via clustering, may enable the discovery of unknown disease patterns or comorbidities, which could eventually lead to better treatment through personalized medicine. Patient data derived from EHRs is heterogeneous and temporally irregular. Therefore, traditional machine learning methods like PCA are ill-suited for analysis of EHR-derived patient data. We propose to address these issues with a new methodology based on training a gated recurrent unit (GRU) autoencoder directly on health record data. Our method learns a low-dimensional feature space by training on patient data time series, where the time of each data point is expressed explicitly. We use positional encodings for time, allowing our model to better handle the temporal irregularity of the data. We apply our method to data from the Medical Information Mart for Intensive Care (MIMIC-III). Using our data-derived feature space, we can cluster patients into groups representing major classes of disease patterns. Additionally, we show that our feature space exhibits a rich substructure at multiple scales.

The race to understand immunopathology in COVID-19: Perspectives on the impact of quantitative approaches to understand within-host interactions

The COVID-19 pandemic has revealed the need for the increased integration of modelling and data analysis to public health, experimental, and clinical studies. Throughout the first two years of the pandemic, there has been a concerted effort to improve our understanding of the within-host immune response to the SARS-CoV-2 virus to provide better predictions of COVID-19 severity, treatment and vaccine development questions, and insights into viral evolution and the impacts of variants on immunopathology. Here we provide perspectives on what has been accomplished using quantitative methods, including predictive modelling, population genetics, machine learning, and dimensionality reduction techniques, in the first 26 months of the COVID-19 pandemic approaches, and where we go from here to improve our responses to this and future pandemics.

Machine Learning-Based Integration of Metabolomics Characterisation Predicts Progression of Myopic Retinopathy in Children and Adolescents

Myopic retinopathy is an important cause of irreversible vision loss and blindness. As metabolomics has recently been successfully applied in myopia research, this study sought to characterize the serum metabolic profile of myopic retinopathy in children and adolescents (4-18 years) and to develop a diagnostic model that combines clinical and metabolic features. We selected clinical and serum metabolic data from children and adolescents at different time points as the training set (n = 516) and the validation set (n = 60). All participants underwent an ophthalmologic examination. Untargeted metabolomics analysis of serum was performed. Three machine learning (ML) models were trained by combining metabolic features and conventional clinical factors that were screened for significance in discrimination. The better-performing model was validated in an independent point-in-time cohort and risk nomograms were developed. Retinopathy was present in 34.2% of participants (n = 185) in the training set, including 109 (28.61%) with mild to moderate myopia. A total of 27 metabolites showed significant variation between groups. After combining Lasso and random forest (RF), 12 modelled metabolites (mainly those involved in energy metabolism) were screened. Both the logistic regression and extreme Gradient Boosting (XGBoost) algorithms showed good discriminatory ability. In the time-validation cohort, logistic regression (AUC 0.842, 95% CI 0.724-0.96) and XGBoost (AUC 0.897, 95% CI 0.807-0.986) also showed good prediction accuracy and had well-fitted calibration curves. Three clinical characteristic coefficients remained significant in the multivariate joint model (p < 0.05), as did 8/12 metabolic characteristic coefficients. Myopic retinopathy may have abnormal energy metabolism. Machine learning models based on metabolic profiles and clinical data demonstrate good predictive performance and facilitate the development of individual interventions for myopia in children and adolescents.

Severity of Illness Scores and Biomarkers for Prognosis of Patients with Coronavirus Disease 2019

The spectrum of disease severity and the insidiousness of clinical presentation make it difficult to recognize patients with coronavirus disease 2019 (COVID-19) at higher risk of worse outcomes or death when they are seen in the early phases of the disease. There are now well-established risk factors for worse outcomes in patients with COVID-19. These should be factored in when assessing the prognosis of these patients. However, a more precise prognostic assessment in an individual patient may warrant the use of predictive tools. In this manuscript, we conduct a literature review on the severity of illness scores and biomarkers for the prognosis of patients with COVID-19. Several COVID-19-specific scores have been developed since the onset of the pandemic. Some of them are promising and can be integrated into the assessment of these patients. We also found that the well-known pneumonia severity index (PSI) and CURB-65 (confusion, uremia, respiratory rate, BP, age ≥ 65 years) are good predictors of mortality in hospitalized patients with COVID-19. While neither the PSI nor the CURB-65 should be used for the triage of outpatient versus inpatient treatment, they can be integrated by a clinician into the assessment of disease severity and can be used in epidemiological studies to determine the severity of illness in patient populations. Biomarkers also provide valuable prognostic information and, importantly, may depict the main physiological derangements in severe disease. We, however, do not advocate the isolated use of severity of illness scores or biomarkers for decision-making in an individual patient. Instead, we suggest the use of these tools on a case-by-case basis with the goal of enhancing clinician judgment.

Data quality and autism: Issues and potential impacts

INTRODUCTION

Large healthcare datasets can provide insight that has the potential to improve outcomes for patients. However, it is important to understand the strengths and limitations of such datasets so that the insights they provide are accurate and useful. The aim of this study was to identify data inconsistencies within the Hospital Episodes Statistics (HES) dataset for autistic patients and assess potential biases introduced through these inconsistencies and their impact on patient outcomes. The study can only identify inconsistencies in recording of autism diagnosis and not whether the inclusion or exclusion of the autism diagnosis is the error.

METHODS

Data were extracted from the HES database for the period 1st April 2013 to 31st March 2021 for patients with a diagnosis of autism. First spells in hospital during the study period were identified for each patient and these were linked to any subsequent spell in hospital for the same patient. Data inconsistencies were recorded where autism was not recorded as a diagnosis in a subsequent spell. Features associated with data inconsistencies were identified using a random forest classifiers and regression modelling.

RESULTS

Data were available for 172,324 unique patients who had been recorded as having an autism diagnosis on first admission. In total, 43.7 % of subsequent spells were found to have inconsistencies. The features most strongly associated with inconsistencies included greater age, greater deprivation, longer time since the first spell, change in provider, shorter length of stay, being female and a change in the main specialty description. The random forest algorithm had an area under the receiver operating characteristic curve of 0.864 (95 % CI [0.862 - 0.866]) in predicting a data inconsistency. For patients who died in hospital, inconsistencies in their final spell were significantly associated with being 80 years and over, being female, greater deprivation and use of a palliative care code in the death spell.

CONCLUSIONS

Data inconsistencies in the HES database were relatively common in autistic patients and were associated a number of patient and hospital admission characteristics. Such inconsistencies have the potential to distort our understanding of service use in key demographic groups.

An Interplay between Oxidative Stress (Lactate Dehydrogenase) and Inflammation (Anisocytosis) Mediates COVID-19 Severity Defined by Routine Clinical Markers

Viral infections activate the innate immune response and the secretion of inflammatory cytokines. They also alter oxidative stress markers, which potentially can have an involvement in the pathogenesis of the disease. The aim of this research was to study the role of the oxidative stress process assessed through lactate dehydrogenase (LDH) on the severity of COVID-19 measured by oxygen saturation (SaO2) and the putative interaction with inflammation. The investigation enrolled 1808 patients (mean age of 68 and 60% male) with COVID-19 from the HM Hospitals database. To explore interactions, a regression model and mediation analyses were performed. The patients with lower SaO2 presented lymphopenia and higher values of neutrophils-to-lymphocytes ratio and on the anisocytosis coefficient. The regression model showed an interaction between LDH and anisocytosis, suggesting that high levels of LDH (>544 U/L) and an anisocytosis coefficient higher than 10% can impact SaO2 in COVID-19 patients. Moreover, analysis revealed that LDH mediated 41% (p value = 0.001) of the effect of anisocytosis on SaO2 in this cohort. This investigation revealed that the oxidative stress marker LDH and the interaction with anisocytosis have an important role in the severity of COVID-19 infection and should be considered for the management and treatment of the oxidative phenomena concerning this within a precision medicine strategy.

Prediction of mortality risk and duration of hospitalization of COVID-19 patients with chronic comorbidities based on machine learning algorithms

Background

The severity of coronavirus (COVID-19) in patients with chronic comorbidities is much higher than in other patients, which can lead to their death. Machine learning (ML) algorithms as a potential solution for rapid and early clinical evaluation of the severity of the disease can help in allocating and prioritizing resources to reduce mortality.

Objective

The objective of this study was to predict the mortality risk and length of stay (LoS) of patients with COVID-19 and history of chronic comorbidities using ML algorithms.

Methods

This retrospective study was conducted by reviewing the medical records of COVID-19 patients with a history of chronic comorbidities from March 2020 to January 2021 in Afzalipour Hospital in Kerman, Iran. The outcome of patients, hospitalization was recorded as discharge or death. The filtering technique used to score the features and well-known ML algorithms were applied to predict the risk of mortality and LoS of patients. Ensemble Learning methods is also used. To evaluate the performance of the models, different measures including F1, precision, recall, and accuracy were calculated. The TRIPOD guideline assessed transparent reporting.

Results

This study was performed on 1291 patients, including 900 alive and 391 dead patients. Shortness of breath (53.6%), fever (30.1%), and cough (25.3%) were the three most common symptoms in patients. Diabetes mellitus(DM) (31.3%), hypertension (HTN) (27.3%), and ischemic heart disease (IHD) (14.2%) were the three most common chronic comorbidities of patients. Twenty-six important factors were extracted from each patient's record. Gradient boosting model with 84.15% accuracy was the best model for predicting mortality risk and multilayer perceptron (MLP) with rectified linear unit function (MSE = 38.96) was the best model for predicting the LoS. The most common chronic comorbidities among these patients were DM (31.3%), HTN (27.3%), and IHD (14.2%). The most important factors in predicting the risk of mortality were hyperlipidemia, diabetes, asthma, and cancer, and in predicting LoS was shortness of breath.

Conclusion

The results of this study showed that the use of ML algorithms can be a good tool to predict the risk of mortality and LoS of patients with COVID-19 and chronic comorbidities based on physiological conditions, symptoms, and demographic information of patients. The Gradient boosting and MLP algorithms can quickly identify patients at risk of death or long-term hospitalization and notify physicians to do appropriate interventions.

Interpretable generalized neural additive models for mortality prediction of COVID-19 hospitalized patients in Hamadan, Iran

BACKGROUND

The high number of COVID-19 deaths is a serious threat to the world. Demographic and clinical biomarkers are significantly associated with the mortality risk of this disease. This study aimed to implement Generalized Neural Additive Model (GNAM) as an interpretable machine learning method to predict the COVID-19 mortality of patients.

METHODS

This cohort study included 2181 COVID-19 patients admitted from February 2020 to July 2021 in Sina and Besat hospitals in Hamadan, west of Iran. A total of 22 baseline features including patients' demographic information and clinical biomarkers were collected. Four strategies including removing missing values, mean, K-Nearest Neighbor (KNN), and Multivariate Imputation by Chained Equations (MICE) imputation methods were used to deal with missing data. Firstly, the important features for predicting binary outcome (1: death, 0: recovery) were selected using the Random Forest (RF) method. Also, synthetic minority over-sampling technique (SMOTE) method was used for handling imbalanced data. Next, considering the selected features, the predictive performance of GNAM for predicting mortality outcome was compared with logistic regression, RF, generalized additive model (GAMs), gradient boosting decision tree (GBDT), and deep neural networks (DNNs) classification models. Each model trained on fifty different subsets of a train-test dataset to ensure a model performance. The average accuracy, F1-score and area under the curve (AUC) evaluation indices were used for comparison of the predictive performance of the models.

RESULTS

Out of the 2181 COVID-19 patients, 624 died during hospitalization and 1557 recovered. The missing rate was 3 percent for each patient. The mean age of dead patients (71.17 ± 14.44 years) was statistically significant higher than recovered patients (58.25 ± 16.52 years). Based on RF, 10 features with the highest relative importance were selected as the best influential features; including blood urea nitrogen (BUN), lymphocytes (Lym), age, blood sugar (BS), serum glutamic-oxaloacetic transaminase (SGOT), monocytes (Mono), blood creatinine (CR), neutrophils (NUT), alkaline phosphatase (ALP) and hematocrit (HCT). The results of predictive performance comparisons showed GNAM with the mean accuracy, F1-score, and mean AUC in the test dataset of 0.847, 0.691, and 0.774, respectively, had the best performance. The smooth function graphs learned from the GNAM were descending for the Lym and ascending for the other important features.

CONCLUSIONS

Interpretable GNAM can perform well in predicting the mortality of COVID-19 patients. Therefore, the use of such a reliable model can help physicians to prioritize some important demographic and clinical biomarkers by identifying the effective features and the type of predictive trend in disease progression.

Frailty, Comorbidity, and Associations With In-Hospital Mortality in Older COVID-19 Patients: Exploratory Study of Administrative Data

BACKGROUND

Older adults have worse outcomes following hospitalization with COVID-19, but within this group there is substantial variation. Although frailty and comorbidity are key determinants of mortality, it is less clear which specific manifestations of frailty and comorbidity are associated with the worst outcomes.

OBJECTIVE

We aimed to identify the key comorbidities and domains of frailty that were associated with in-hospital mortality in older patients with COVID-19 using models developed for machine learning algorithms.

METHODS

This was a retrospective study that used the Hospital Episode Statistics administrative data set from March 1, 2020, to February 28, 2021, for hospitalized patients in England aged 65 years or older. The data set was split into separate training (70%), test (15%), and validation (15%) data sets during model development. Global frailty was assessed using the Hospital Frailty Risk Score (HFRS) and specific domains of frailty were identified using the Global Frailty Scale (GFS). Comorbidity was assessed using the Charlson Comorbidity Index (CCI). Additional features employed in the random forest algorithms included age, sex, deprivation, ethnicity, discharge month and year, geographical region, hospital trust, disease severity, and International Statistical Classification of Disease, 10th Edition codes recorded during the admission. Features were selected, preprocessed, and input into a series of random forest classification algorithms developed to identify factors strongly associated with in-hospital mortality. Two models were developed; the first model included the demographic, hospital-related, and disease-related items described above, as well as individual GFS domains and CCI items. The second model was similar to the first but replaced the GFS domains and CCI items with the HFRS as a global measure of frailty. Model performance was assessed using the area under the receiver operating characteristic (AUROC) curve and measures of model accuracy.

RESULTS

In total, 215,831 patients were included. The model using the individual GFS domains and CCI items had an AUROC curve for in-hospital mortality of 90% and a predictive accuracy of 83%. The model using the HFRS had similar performance (AUROC curve 90%, predictive accuracy 82%). The most important frailty items in the GFS were dementia/delirium, falls/fractures, and pressure ulcers/weight loss. The most important comorbidity items in the CCI were cancer, heart failure, and renal disease.

CONCLUSIONS

The physical manifestations of frailty and comorbidity, particularly a history of cognitive impairment and falls, may be useful in identification of patients who need additional support during hospitalization with COVID-19.

Clinical utility of automatic phenotype annotation in unstructured clinical notes: intensive care unit use

OBJECTIVE

Clinical notes contain information that has not been documented elsewhere, including responses to treatment and clinical findings, which are crucial for predicting key outcomes in patients in acute care. In this study, we propose the automatic annotation of phenotypes from clinical notes as a method to capture essential information to predict outcomes in the intensive care unit (ICU). This information is complementary to typically used vital signs and laboratory test results.

METHODS

In this study, we developed a novel phenotype annotation model to extract the phenotypical features of patients, which were then used as input features of predictive models to predict ICU patient outcomes. We demonstrated and validated this approach by conducting experiments on three ICU prediction tasks, including in-hospital mortality, physiological decompensation and length of stay (LOS) for over 24 000 patients using the Medical Information Mart for Intensive Care (MIMIC-III) dataset.

RESULTS

The predictive models incorporating phenotypical information achieved 0.845 (area under the curve-receiver operating characteristic (AUC-ROC)) for in-hospital mortality, 0.839 (AUC-ROC) for physiological decompensation and 0.430 (kappa) for LOS, all of which consistently outperformed the baseline models using only vital signs and laboratory test results. Moreover, we conducted a thorough interpretability study showing that phenotypes provide valuable insights at both the patient and cohort levels.

CONCLUSION

The proposed approach demonstrates that phenotypical information complements traditionally used vital signs and laboratory test results and significantly improves the accuracy of outcome prediction in the ICU.

Homogeneous ensemble models for predicting infection levels and mortality of COVID-19 patients: Evidence from China

Background

Persistence of long-term COVID-19 pandemic is putting high pressure on healthcare services worldwide for several years. This article aims to establish models to predict infection levels and mortality of COVID-19 patients in China.

Methods

Machine learning models and deep learning models have been built based on the clinical features of COVID-19 patients. The best models are selected by area under the receiver operating characteristic curve (AUC) scores to construct two homogeneous ensemble models for predicting infection levels and mortality, respectively. The first-hand clinical data of 760 patients are collected from Zhongnan Hospital of Wuhan University between 3 January and 8 March 2020. We preprocess data with cleaning, imputation, and normalization.

Results

Our models obtain AUC = 0.7059 and Recall (Weighted avg) = 0.7248 in predicting infection level, while AUC=0.8436 and Recall (Weighted avg) = 0.8486 in predicting mortality ratio. This study also identifies two sets of essential clinical features. One is C-reactive protein (CRP) or high sensitivity C-reactive protein (hs-CRP) and the other is chest tightness, age, and pleural effusion.

Conclusions

Two homogeneous ensemble models are proposed to predict infection levels and mortality of COVID-19 patients in China. New findings of clinical features for benefiting the machine learning models are reported. The evaluation of an actual dataset collected from January 3 to March 8, 2020 demonstrates the effectiveness of the models by comparing them with state-of-the-art models in prediction.

Status Forecasting Based on the Baseline Information Using Logistic Regression

In the status forecasting problem, classification models such as logistic regression with input variables such as physiological, diagnostic, and treatment variables are typical ways of modeling. However, the parameter value and model performance differ among individuals with different baseline information. To cope with these difficulties, a subgroup analysis is conducted, in which models' ANOVA and rpart are proposed to explore the influence of baseline information on the parameters and model performance. The results show that the logistic regression model achieves satisfactory performance, which is generally higher than 0.95 in AUC and around 0.9 in F1 and balanced accuracy. The subgroup analysis presents the prior parameter values for monitoring variables including SpO2, milrinone, non-opioid analgesics and dobutamine. The proposed method can be used to explore variables that are and are not medically related to the baseline variables.

Using machine learning in prediction of ICU admission, mortality, and length of stay in the early stage of admission of COVID-19 patients

The recent COVID-19 pandemic has affected health systems across the world. Especially, Intensive Care Units (ICUs) have played a pivotal role in the treatment of critically-ill patients. At the same time however, the increasing number of admissions due to the vast prevalence of the virus have caused several problems for ICU wards such as overburdening of staff and shortages of medical resources. These issues might have affected the quality of healthcare services provided directly impacting a patient's survival. The objective of this research is to leverage Machine Learning (ML) on hospital data in order to support hospital managers and practitioners with the treatment of COVID-19 patients. This is accomplished by providing more detailed inference about a patient's likelihood of ICU admission, mortality and in case of hospitalization the length of stay (LOS). In this pursuit, the outcome variables are in three separate models predicted by five different ML algorithms: eXtreme Gradient Boosting (XGB), K-Nearest Neighbor (KNN), Random Forest (RF), bagged-CART (b-CART), and LogitBoost (LB). With the exception of KNN, the studied models show good predictive capabilities when evaluating relevant accuracy scores, such as area under the curve. By implementing an ensemble stacking approach (either a Neural Net or a General Linear Model) on top of the aforementioned ML algorithms the performance is further boosted. Ultimately, for the prediction of admission to the ICU, the ensemble stacking via a Neural Net achieved the best result with an accuracy of over 95%. For mortality at the ICU, the vanilla XGB performed slightly better (1% difference with the meta-model). To predict large length of stays both ensemble stacking approaches yield comparable results. Besides it direct implications for managing COVID-19 patients, the approach presented serves as an example how data can be employed in future pandemics or crises.

Prediction of hospital mortality in intensive care unit patients from clinical and laboratory data: A machine learning approach

BACKGROUND

Intensive care unit (ICU) patients demand continuous monitoring of several clinical and laboratory parameters that directly influence their medical progress and the staff's decision-making. Those data are vital in the assistance of these patients, being already used by several scoring systems. In this context, machine learning approaches have been used for medical predictions based on clinical data, which includes patient outcomes.

AIM

To develop a binary classifier for the outcome of death in ICU patients based on clinical and laboratory parameters, a set formed by 1087 instances and 50 variables from ICU patients admitted to the emergency department was obtained in the "WiDS (Women in Data Science) Datathon 2020: ICU Mortality Prediction" dataset.

METHODS

For categorical variables, frequencies and risk ratios were calculated. Numerical variables were computed as means and standard deviations and Mann-Whitney U tests were performed. We then divided the data into a training (80%) and test (20%) set. The training set was used to train a predictive model based on the Random Forest algorithm and the test set was used to evaluate the predictive effectiveness of the model.

RESULTS

A statistically significant association was identified between need for intubation, as well predominant systemic cardiovascular involvement, and hospital death. A number of the numerical variables analyzed (for instance Glasgow Coma Score punctuations, mean arterial pressure, temperature, pH, and lactate, creatinine, albumin and bilirubin values) were also significantly associated with death outcome. The proposed binary Random Forest classifier obtained on the test set (n = 218) had an accuracy of 80.28%, sensitivity of 81.82%, specificity of 79.43%, positive predictive value of 73.26%, negative predictive value of 84.85%, F1 score of 0.74, and area under the curve score of 0.85. The predictive variables of the greatest importance were the maximum and minimum lactate values, adding up to a predictive importance of 15.54%.

CONCLUSION

We demonstrated the efficacy of a Random Forest machine learning algorithm for handling clinical and laboratory data from patients under intensive monitoring. Therefore, we endorse the emerging notion that machine learning has great potential to provide us support to critically question existing methodologies, allowing improvements that reduce mortality.

Optical coherence tomography and convolutional neural networks can differentiate colorectal liver metastases from liver parenchyma ex vivo

PURPOSE

Optical coherence tomography (OCT) is an imaging technology based on low-coherence interferometry, which provides non-invasive, high-resolution cross-sectional images of biological tissues. A potential clinical application is the intraoperative examination of resection margins, as a real-time adjunct to histological examination. In this ex vivo study, we investigated the ability of OCT to differentiate colorectal liver metastases (CRLM) from healthy liver parenchyma, when combined with convolutional neural networks (CNN).

METHODS

Between June and August 2020, consecutive adult patients undergoing elective liver resections for CRLM were included in this study. Fresh resection specimens were scanned ex vivo, before fixation in formalin, using a table-top OCT device at 1310 nm wavelength. Scanned areas were marked and histologically examined. A pre-trained CNN (Xception) was used to match OCT scans to their corresponding histological diagnoses. To validate the results, a stratified k-fold cross-validation (CV) was carried out.

RESULTS

A total of 26 scans (containing approx. 26,500 images in total) were obtained from 15 patients. Of these, 13 were of normal liver parenchyma and 13 of CRLM. The CNN distinguished CRLM from healthy liver parenchyma with an F1-score of 0.93 (0.03), and a sensitivity and specificity of 0.94 (0.04) and 0.93 (0.04), respectively.

CONCLUSION

Optical coherence tomography combined with CNN can distinguish between healthy liver and CRLM with great accuracy ex vivo. Further studies are needed to improve upon these results and develop in vivo diagnostic technologies, such as intraoperative scanning of resection margins.