Predicting Adverse Events During Six-Minute Walk Test Using Continuous Physiological Signals.

Area of Science:

• Cardiology
• Pulmonology
• Biomedical Engineering

Background:

• The six-minute walk test (6MWT) is a standard functional assessment.
• Adverse events during 6MWT can lead to severe health consequences and reduced quality of life.

Purpose of the Study:

• To predict adverse events during the 6MWT.
• To utilize continuous physiological parameters and demographic data for prediction.

Main Methods:

• 578 patients with respiratory diseases undergoing 6MWT with wearable devices were analyzed.
• ECG, respiratory signals, acceleration, oxygen saturation, and demographics were collected.
• Machine learning models (LightGBM, Logistic Regression) were trained and validated using 5-fold cross-validation.

Main Results:

• LightGBM achieved the highest AUC of 0.874 ± 0.063; Logistic Regression achieved 0.869 ± 0.067.
• Features related to blood oxygen were most important; heart rate features were most numerous.
• Traditional scales (mMRC, Borg) showed lower predictive performance (AUC 0.733 and 0.656).

Conclusions:

• Predicting 6MWT adverse events using continuous physiological data and demographics is feasible.
• Wearable sensors offer a valuable tool for continuous monitoring and enhancing patient safety during 6MWT.