Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Wave-aware mortality prediction in COVID-19: a multi-stage feature selection and explainable machine-learning

Elaheh Fereidouni¹, Shadi Shafaghi², Hamidreza Jamaati³

¹Department of Biostatistics, School of Public Health, Iran University of Medical Sciences, Tehran, Iran.

BMC Medical Informatics and Decision Making

|June 10, 2026

Summary

Related Concept Videos

Steps in Outbreak Investigation

Steps in Outbreak Investigation

In the ever-evolving field of public health, statistical analysis serves as a cornerstone for understanding and managing disease outbreaks. By leveraging various statistical tools, health professionals can predict potential outbreaks, analyze ongoing situations, and devise effective responses to mitigate impact. For that to happen, there are a few possible stages of the analysis:

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Translating Data Into Clinical Tools: An Integrative Strategy for Precision Biomarker Identification in Soft Tissue Sarcoma Diagnosis and Prognosis.

Cancer informatics·2026

Same author

Knowledge, attitudes, and practices toward artificial intelligence in medicine among iranian physicians and medical students: a cross-sectional survey.

BMC research notes·2026

Same author

Insulin resistance and lower insulin sensitivity: significant contributors to weight regain in non-menopausal Iranian women.

BMC nutrition·2026

Same author

Effects of Time-Restricted Eating With Calorie Restriction Versus Calorie Restriction on Health Outcomes in Patients With Weight Regain Post-Sleeve Surgery: A Randomized Controlled Trial.

Obesity surgery·2026

Same author

The Effect of Intermittent Use of Oxygen Therapy via High-Flow Nasal Cannula Alternating with Noninvasive Ventilation on Critically Ill Patients.

Tanaffos·2026

Same author

New Insights into Pulmonary Arterial Hypertension Treatment with Non-Coding RNAs: What is the Best Option, microRNAs or Small-Interfering RNAs?

Tanaffos·2026

Same journal

Data analysis tool for identifying multidimensional health profiles associated with frailty in older adults.

BMC medical informatics and decision making·2026

Same journal

Development and external validation of an interpretable machine learning model for predicting prolonged postoperative ICU length of stay in coronary artery bypass grafting patients using MIMIC-IV 3.1 and eICU-CRD 2.0.

BMC medical informatics and decision making·2026

Same journal

Construction and validation of a predictive nomogram for 28-day mortality in critically ill patients with toxic encephalopathy.

BMC medical informatics and decision making·2026

Same journal

Towards an information-centric architecture framework for health information logistics: a design science research study.

BMC medical informatics and decision making·2026

Same journal

Machine learning for predicting institutionalization and mortality risks among older home care recipients with routinely collected need assessment data: explainable AI for long-term care.

BMC medical informatics and decision making·2026

Same journal

An integrated evaluation framework for synthetic clinical data in severely imbalanced settings: fidelity, privacy-risk profiling, and diagnostic utility.

BMC medical informatics and decision making·2026

See all related articles

This summary is machine-generated.

COVID-19 mortality prediction models degrade over time due to evolving epidemic conditions. A wave-aware machine learning approach is crucial for maintaining clinical utility by accounting for temporal drift in risk factors.

Area of Science:

Machine learning applications in public health
Epidemiology and disease modeling
Computational biology and bioinformatics

Background:

Hospitalized COVID-19 patient mortality varied significantly across epidemic waves due to shifting variants, population immunity, and healthcare system pressures.
Concerns exist regarding the reliability of early-phase COVID-19 prediction models in later, more heterogeneous epidemic contexts.

Purpose of the Study:

To develop a machine learning framework that is aware of epidemic waves.
To differentiate between stable physiological predictors of mortality and context-dependent predictors whose importance changes across waves.

Main Methods:

Analysis of 732,654 adult hospitalizations from Iran's national COVID-19 registry across Waves 2-5 and the post-Wave-5 period.
Independent preprocessing, feature selection, and model training within each wave to preserve temporal structure.

Keywords:

COVID-19 Concept drift Deep neural network Elastic net Explainable AI Feature selection In-hospital mortality Machine learning Random forest SHAP Temporal modeling

Related Experiment Videos

A three-stage feature selection approach (Elastic Net, Random Forest, Variance Inflation Factor) to identify stable and time-varying predictors.

Main Results:

A core set of physiological predictors (age, hypoxemia, chronic comorbidities) remained consistent across waves.
Nonlinear models, particularly Random Forest, outperformed Logistic Regression in within-wave testing (AUC up to 0.94, F1 up to 0.68).
Early-wave models showed significant performance degradation when applied to later waves, with F1 scores collapsing due to miscalibration.

Conclusions:

COVID-19 mortality is influenced by a stable physiological core and dynamic, wave-specific factors.
Static prediction models are susceptible to temporal drift, leading to degraded clinical utility.
Drift-aware approaches, such as recalibration or adaptive thresholding, are essential for maintaining the clinical utility of prediction models in evolving infectious disease environments.