DEEP MULTI-TASK LEARNING FRAMEWORK FOR SIMULTANEOUS AUTO IMMUNE DISEASE PREDICTION AND SEVERITY ANALYSIS

Autoimmune diseases affect nearly 10% of the global population and pose significant challenges in early diagnosis and disease management. Accurate prediction of disease occurrence along with severity assessment is essential for effective treatment planning and improved patient outcomes. However, existing computational methods often handle disease prediction and severity analysis as separate tasks, limiting the ability to capture their underlying clinical relationships. To address this limitation, this study proposes a Deep Multi-Task Learning Framework for
simultaneous autoimmune disease prediction and severity analysis. The proposed architecture utilizes a shared
deep learning encoder to learn meaningful representations from multi-modal clinical data, followed by taskspecific layers for disease classification and severity estimation. An uncertainty-weighted composite loss function
is employed to optimize both tasks jointly and improve model generalization. Experimental evaluation
demonstrates that the proposed framework outperforms conventional single-task models. The model achieved a
classification accuracy of 94.2% with an Area Under the Curve (AUC) of 0.96 for disease prediction, while also
obtaining a low Mean Absolute Error (MAE) of 0.12 for severity estimation. The results highlight the
effectiveness of multi-task deep learning in healthcare applications by enabling integrated disease prediction and
severity assessment within a unified framework. This approach can support clinicians in early diagnosis,
personalized treatment planning, and efficient disease monitoring.