Background: Digital healthcare aims to improve user engagement and health outcomes by analyzing various digital behaviors and traits.

Objective: This study examines how well different characteristics predict engagement and health outcomes in a digital cognitive behavioral therapy (CBT) program.

Methods: 45 participants used a CBT mobile app for 8 weeks. Researchers used machine learning to analyze both traditional psychological questionnaires and digital behaviors recorded by the app.

Results:

• Higher engagement was linked to greater weight loss at 8 and 24 weeks.

• Key factors included:

o Lower self-esteem (traditional measure).

o Higher motivation (digital measure).

• Specific digital behaviors, like eating fewer high-calorie foods and frequent mentor interactions, predicted engagement.

• Eating fewer carbs and more low-calorie foods predicted short- and long-term weight changes.

Conclusions: The study used machine learning to show that multiple psychological and behavioral factors can explain the effectiveness of digital health programs, leading to better, personalized treatments in the future.