Related Experiment Video
Updated: Jul 9, 2025
07:31
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
7.1K
Using Multi-Modal Electronic Health Record Data for the Development and Validation of Risk Prediction Models for Long
Weijia Jin1,2, Wei Hao1,2, Xu Shi1
1Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, MI 48109, USA.
Journal of Clinical Medicine
|December 9, 2023
Summary
Predicting Post-Acute Sequelae of COVID-19 (PASC) is crucial. Overweight/obesity, prior statin use, and respiratory symptoms during infection are key predictors. A composite risk score aids in identifying individuals at higher risk for PASC.
Area of Science:
- Medical research
- Public health
- Data science
Background:
- Post-Acute Sequelae of COVID-19 (PASC) present a significant global health challenge.
- Identifying predictive factors for PASC is essential for proactive healthcare management.
Purpose of the Study:
- To uncover predictive factors for PASC using multi-modal data.
- To develop and evaluate a predictive model for PASC diagnoses.
Main Methods:
- Analysis of electronic health records from 92,301 COVID-19 patients.
- Utilized a Super Learner-based prediction approach for factor identification.
- Integrated model outputs into individual and composite risk scores for performance evaluation.
Main Results:
- Identified overweight/obesity, pre-infection HMG CoA reductase inhibitor use, and respiratory symptoms during COVID-19 as predictive factors for PASC.
- Developed a composite risk score with moderate discriminatory ability (AUC: 0.66).
- The composite risk score identified 10% of individuals with a 2.2-fold increased risk for PASC.
Conclusions:
- Several factors predictive of PASC diagnoses were identified.
- A composite risk score was developed to aid in PASC prediction.
- This risk score can help identify high-risk individuals and inform preventive strategies.
