TAPE–TAFTE: Redefining Mental Health Detection Through Adaptive Temporal Fusion in 6G-Enabled Health Systems

Mental disorder identification has turned into a substantial element in the 6G-powered healthcare setup of tomorrow that makes use of the characteristics such as edge intelligence, ultra-low latency, and extensive data connectivity to render the mental health monitoring a smooth, uninterrupted, and customizable process. Nevertheless, these pieces of research in their present state encounter various obstacles like the restricted multimodal integration, lack of temporal and context-awareness, high computational overheads, and reduced interpretability of mental disorder classification. Moreover, the conventional deep learning models are not capable of generalizing across different datasets and might overlook even more minute behavioral and linguistic aspects that are highly instrumental in the early stage of diagnosis. This paper devises a smart mechanism with Tabular Adaptive Projection and Embedding (TAPE) and Tabular-Aware Field-Time Encoder (TAFTE) modules that help in a quick identification and accurate localization of mental disorders in clinical and social data at large scale. The TAPE module enables the feature mapping through the embedding of the multi-dimensional mental health metrics into one single latent representation, whereas TAFTE automatically captures the contextual and temporal relations thus facilitating self-learning and interpretability. The research framework utilized Reddit and Kaggle Mental Health datasets that are open sources of real and varied emotional, linguistic, and behavioral data distributions. The experimental results reveal that the proposed TAPE + TAFTE model achieves the topmost performance with the following measures: accuracy of 99.12%, precision of 98.97%, recall of 99.03%, F1-score of 98.99% and time of 1.5s.