Artificial intelligence machine learning models, trained using health data, can predict a patient's risk of being hospitalized with COVID-19, according to a Nov. 15 study published in the Journal of Medical Internet Research.
Researchers utilized data from the COVID-19 Research Data Commons, a statewide health information exchange from 23 health systems and 93 hospitals in Indiana, to train machine learning models to identify patient-level need for healthcare.
The decision model used patients' age, chronic obstructive pulmonary disease status, smoking, diabetes, indication of neurological diseases, mental disorders, residence type and income level for its predictions.
The study found that out of the 92,026 COVID-19 patients, 18,694 were hospitalized during their first week of testing positive for COVID-19, and 22,678 were hospitalized during the first six weeks of testing positive.
"Such utilization prediction models may be used for population health management programs in health systems, to identify high-risk populations to monitor or screen, as well as predicting resource needs in crisis situations, such as future spikes in pandemic activity or outbreaks," the study said.
The model performed well when identifying which patients were at risk for hospitalization during their first week of being diagnosed with COVID-19, but the model did show biases. For example, it showed stronger predictive performance when used on men or people who lived in urban areas.