Predicting post-experiment fatigue among healthy young adults: Random forest regression analysis

Area of Science:

• Psychology
• Health Sciences
• Data Science

Background:

• Fatigue is a common experience following stressful events.
• Understanding predictors of post-experiment fatigue is crucial for managing well-being.
• Previous research has explored various factors contributing to fatigue, but complex interactions remain under investigation.

Purpose of the Study:

• To predict post-experiment fatigue using a machine learning approach.
• To identify key predictors of fatigue following a mildly stressful experiment.
• To explore the utility of random forest regression in analyzing complex health-related data.

Main Methods:

• A random forest regression analysis was employed.
• The study included 212 healthy participants aged 18-30.
• Thirty features encompassing demographics, lifestyle, psychological state, and physiological indicators were utilized.

Main Results:

• The model accurately predicted post-experiment fatigue (R equivalent = 0.93), with an average out-of-bag R of 0.52.
• Self-reported pre-experiment fatigue was the strongest predictor (54%).
• State anxiety, blood pressure, heart rate variability (HRV), alcohol-related problems, and sleep quality also contributed to fatigue prediction.

Conclusions:

• Complex interactions across multiple systems influence fatigue.
• Random forest regression is an effective tool for identifying predictors and understanding relationships in health behavior research.
• These findings highlight the multifaceted nature of fatigue and its underlying regulatory mechanisms.