Fts App: Towards Explainable AI-Driven Adaptive Weight Tracking

We present the FTS app, an adaptive weight management tracking application that integrates an explainable artificial intelligence (XAI) model, specifically designed for normal weight individuals who aim to maintain their current weight status. Weight maintenance requires consistent monitoring of calories consumed from food and those expended through exercise, a challenging task with conventional apps available on Play Store or App Store. The FTS app uniquely analyzes both food logs and daily steps to predict users' current weight. A key feature is its explainable AI framework that helps users understand why certain foods might trigger weight fluctuations and offer healthier alternatives. The system integrates two complementary models: (1) a ChatGPT-4 model and (2) the proprietary FTS- XAI model, both designed to classify foods as healthy or unhealthy while calculating caloric content to estimate the user's current weight. The app includes a built-in step counting module that tracks physical activity before meals, eliminating the need for expensive wearable devices and reducing privacy concerns related to thirdparty access. By combining these calorie metrics, the app automatically calculates the user's current weight and keeps them informed. This paper presents our design framework to guide the research community in addressing the needs of this predominantly overlooked population with weight maintenance goals.