NutriTrack: AI-Assisted Nutritional Tracking in the Obesity Clinic

Overview

NutriTrack is a digital health application designed to support nutritional and behavioral tracking in patients with obesity followed in an outpatient obesity clinic. The application allows patients to record food intake using food photographs, barcode scanning, or manual search, and to register behavioral variables related to eating episodes. This prospective, single-center, observational pilot study will evaluate the feasibility and usability of NutriTrack in 20 to 50 adult patients with obesity or overweight with comorbidities followed at the Obesity Clinic of Hospital Clínico San Carlos. Participants will use the application for 4 weeks as a complementary tool. The information generated by NutriTrack will be available to healthcare professionals as supportive information and will not replace clinical judgment or modify usual care decisions. The main outcome is usability measured using the System Usability Scale. Secondary and exploratory outcomes include agreement between artificial intelligence-based nutritional estimates and standard dietitian assessment, adherence to daily food logging, professional perceived clinical utility, changes in eating behavior and emotional regulation scales, and technical feasibility of data export.

NutriTrack is a digital health application developed within Hospital Clínico San Carlos and Universidad Complutense de Madrid to support nutritional and behavioral tracking in patients with obesity. The application integrates artificial intelligence-based food image recognition, the Spanish BEDCA nutritional database, and a rule-based engine for the detection of clinically relevant eating patterns. This is a prospective, single-center, observational pilot study with complementary clinical use. The application will be used by adult patients followed at the Obesity Clinic for 4 weeks. Participants will record daily food intake and behavioral variables related to eating episodes, including hunger, satiety, emotional state, eating context, eating speed, perceived control, and cravings. Healthcare professionals may review the NutriTrack clinical panel as supportive information. The application output will not replace clinical judgment, will not trigger automated clinical decisions, will not modify usual care, and will not be integrated into the hospital electronic health record during the pilot phase. The study will assess usability, nutritional estimation accuracy, adherence to food logging, professional perceived clinical utility, exploratory changes in emotional regulation and eating behavior, and technical feasibility of data export. The study is intended to generate preliminary evidence to support future larger-scale validation.