Related Experiment Video
Updated: Jun 1, 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
7.0K
Predicting work ability impairment in post COVID-19 patients: a machine learning model based on clinical parameters.
Tarek Jebrini1, Michael Ruzicka2,3, Felix Völk4
1Department of Psychiatry and Psychotherapy, Ludwig Maximilian University (LMU) University Hospital, LMU Munich, Munich, Germany.
Infection
|January 17, 2025
Summary
Post COVID-19 condition (PCC) can cause inability to work (ITW). This study identified factors linked to ITW and developed a predictive model to identify patients at risk of prolonged work absence.
Area of Science:
- Medical research
- Public health
- Long COVID studies
Background:
- Post COVID-19 condition (PCC) significantly impacts daily functioning, notably patients' ability to work (ITW).
- Understanding factors contributing to ITW is crucial for effective patient management and rehabilitation strategies.
- Previous research has highlighted the multifaceted nature of PCC, but specific predictors for sustained ITW require further investigation.
Purpose of the Study:
- To identify clinical and patient-reported factors associated with inability to work (ITW) in Post COVID-19 condition (PCC) patients.
- To develop and validate a machine learning model for predicting ITW twelve months post-baseline.
- To provide insights for clinical decision-making and long-term treatment planning for individuals with PCC.
Main Methods:
- Analysis of 259 PCC patients from the PCC-study, comparing those with and without ITW.
- Utilized nine clinical parameters including hospitalization, infection severity, comorbidities, and Karnofsky index for model construction.
- Employed TensorFlow Decision Forests for model training and validated performance using cross-validation and an independent testing dataset.
Main Results:
- ITW was significantly associated with dyslipidemia, poorer patient-reported outcomes (FSS, WHOQOL-BREF, PHQ-9), pre-existing psychiatric conditions, and extensive medical work-up.
- The machine learning model achieved a mean AUC of 0.83 during 10-fold cross-validation.
- The model demonstrated an AUC of 0.76 on the independent testing dataset, indicating strong predictive capability.
Conclusions:
- Several key factors predict inability to work in Post COVID-19 condition patients.
- The developed machine learning model shows high performance in predicting extended ITW.
- This predictive tool can aid clinicians in identifying at-risk patients, guiding management and setting realistic treatment goals.
