Personalized weight loss management through wearable devices and artificial intelligence.

Area of Science:

• Biomedical Engineering
• Health Informatics
• Artificial Intelligence in Healthcare

Background:

• Early detection of chronic diseases is vital for effective management.
• Weight management in overweight and obese individuals is a significant public health concern.
• Personalized interventions require accurate prediction of individual responses.

Purpose of the Study:

• To investigate the use of wearable device data and AI for predicting weight loss in overweight/obese individuals.
• To identify key physiological and behavioral features influencing weight loss.
• To assess the performance of AI models in weight loss prediction.

Main Methods:

• Utilized a 1-month wearable data collection from ~100 subjects in the AI4FoodDB database.
• Extracted physiological (e.g., heart rate variability) and behavioral (e.g., sleep, physical activity) features.
• Employed feature selection and classification algorithms (e.g., Gradient Boosting) for prediction.

Main Results:

• Identified significant differences in physiological and behavioral patterns between weight loss achievers and non-achievers.
• Heart rate variability, sleep duration, and physical activity were key predictors of weight loss.
• Gradient Boosting classifier achieved an 84.44% area under the curve (AUC).

Conclusions:

• Wearable devices combined with AI can effectively predict weight loss in overweight/obese individuals.
• Integration of diverse data sources (vital signs, activity, sleep) improves prediction accuracy.
• This approach holds promise for personalized healthcare and early disease intervention.