Obesity among adolescents is increasingly recognized as a major health issue, prompting the development of a new framework for identifying obesity through deep learning techniques.
This study utilizes smartphone sensors to analyze gait patterns and employs three types of deep learning models—CNNs, LSTMs, and a hybrid model—demonstrating high accuracy in distinguishing between normal and obese individuals.
While the hybrid model achieved the best accuracy of 97%, the research faces limitations like a small sample size and insufficient representation of individuals with abnormal gait, leading to future plans for more inclusive models.