AI-Driven Personalized Weight Loss Strategies and Behavioral Patterns Among Obese Adults

Abstract

Background: Obesity is a multifactorial chronic disease in which conventional lifestyle programs often produce heterogeneous outcomes due to limited personalization and suboptimal adherence. Artificial intelligence (AI)–enabled platforms can adapt dietary, activity, and behavioral recommendations using continuous user data, potentially improving engagement and clinical response. Objective: To evaluate the effectiveness of an AI-driven personalized weight management intervention versus standard counseling-based weight management in improving weight loss and adherence-related behaviors among obese adults in South Punjab, Pakistan. Methods: In this parallel-group randomized controlled trial, 180 adults aged 25–55 years with BMI 30.0–39.9 kg/m² were randomized 1:1 to an AI-assisted mobile program integrating self-monitoring inputs and wearable-derived activity/sleep metrics or to standard biweekly counseling without algorithmic personalization. Assessments at baseline, 8 weeks, and 16 weeks included anthropometry and validated behavioral measures (IPAQ; WELQ). Analyses followed intention-to-treat principles with repeated-measures testing and effect size estimation. Results: The AI group achieved greater mean weight loss than controls (−8.9 kg [95% CI −9.6 to −8.2] vs −4.2 kg [−4.9 to −3.5]), with a between-group difference of −4.7 kg (−5.6 to −3.8; p<0.001; d=1.59). BMI reduction was larger in the AI group (−3.2 vs −1.6 kg/m²; p<0.001), and waist circumference declined more (−8.4 vs −4.1 cm; p<0.001). The AI group showed higher physical activity (2860±520 vs 2210±480 MET-min/week), dietary adherence (84.5±6.1% vs 69.8±8.0%), and self-monitoring (5.6±1.0 vs 3.1±1.2 days/week) (all p<0.001). Conclusion: AI-driven personalized lifestyle intervention produced clinically and statistically superior short-term weight loss and adherence-related behavioral improvements compared with standard counseling, supporting its potential as a scalable adjunct for obesity management in resource-constrained settings.