Unveiling overeating patterns within digital longitudinal data on eating behaviors and contexts

Area of Science:

• Behavioral Science
• Obesity Research
• Digital Health

Background:

• Overeating is a significant contributor to obesity, a pressing public health issue.
• Understanding the nuances of overeating behaviors is crucial for developing effective interventions.

Purpose of the Study:

• To investigate the behavioral, psychological, and contextual factors contributing to overeating in individuals with obesity.
• To identify distinct overeating phenotypes using passive sensing and Ecological Momentary Assessments (EMAs).

Main Methods:

• The SenseWhy study monitored 65 individuals with obesity in free-living settings.
• Data collection involved wearable cameras, mobile apps, dietary recalls, and EMAs over 657 days.
• Semi-supervised learning was applied to EMA-derived features to identify overeating phenotypes.

Main Results:

• Overeating episodes were predicted with high accuracy (mean AUROC=0.86, mean AUPRC=0.84) using EMA and passive sensing data.
• Five distinct overeating phenotypes were identified: "Take-out Feasting," "Evening Restaurant Reveling," "Evening Craving," "Uncontrolled Pleasure Eating," and "Stress-driven Evening Nibbling."
• These phenotypes highlight the complex interplay of factors influencing overeating.

Conclusions:

• Overeating is multifaceted, influenced by a combination of behavioral, psychological, and environmental factors.
• The identified overeating phenotypes provide a basis for developing personalized interventions to combat obesity.
• Passive sensing and EMA data offer valuable insights into eating behaviors in real-world settings.