Development of a Risk Scoring Model for Predicting Obesity in Adolescents: A Study in Palembang, Indonesia

Adolescent obesity is a growing public health concern in Indonesia, including Palembang. Existing Western-derived risk models lack ethnicity, parental genetics, and socioeconomic status (SES). This study aimed to develop a simple integer-based risk-scoring model for obesity among adolescents in Palembang, using both modifiable and non-modifiable predictors. A cross-sectional study from March to November 2024 enrolled 332 adolescents (176 boys, 156 girls) from public secondary schools. Anthropometric measurements (height, weight, body mass index (BMI) z-score), International Physical Activity Questionnaire-Short Form (IPAQ-SF), dietary intake, Child Eating Behavior Questionnaire (CEBQ), parental genetics, and SES were collected. Logistic regression identified predictors, and a scoring system was derived from the β-coefficients and evaluated using the area under the curve (AUC). The study revealed that the prevalence of obesity was 46.98%. Significant predictors included age (OR 31.80, p=0.017), high-calorie intake (OR 52.98, p=0.042), high-fat diet (OR 56.04, p=0.031), high food approach (OR 43.96, p=0.023), and low food avoidance (OR 15.85, p=0.048). A high-fat diet was the dominant predictor. It is concluded that the developed integer-based risk scoring model, incorporating age, high-calorie intake, high-fat dietary pattern, high food approach, and low food avoidance, demonstrates excellent discrimination (AUC 0.9406) and serves as a practical, non-invasive tool for early obesity risk stratification among adolescents in Palembang.