This study successfully developed and validated a deep learning-based continuous in-hospital deterioration prediction model using data collected from wearable chest-worn monitors. The researchers piloted two different Continuous Monitoring (CM) devices in 888 non-ICU inpatients across four hospitals within the Northwell Health system, gathering over 2,897 patient days of continuous vital sign data between March 2020 and November 2022.

By comparing the continuously measured vitals against traditional episodically measured vitals from the Electronic Health Record (EHR), the team demonstrated that CM data can generate more frequent and earlier clinical alerts related to patient deterioration.

The core achievement of the study is the deep learning algorithm, which integrates both CM and EHR data to predict clinical alerts up to 24 hours in advance.

Crucially, the model maintained consistent high performance across different patient cohorts, hospital environments, and the two distinct CM devices used in the pilot.

This robust algorithm shows significant promise in predicting “hard outcomes” such as ICU transfer and death.

To the authors’ knowledge, this is the largest study of its kind to propose a comprehensive inpatient CM framework tested on hard clinical outcomes and validated against real-time deployment scenarios, marking a major step toward proactive, wearable-driven hospital care.

Read More