A Distributed AI-based Disease Classification Approach

This paper presents a distributed learning approach designed to assist physicians in diagnosing diseases based on a patient’s current health status, clinical history, and test results. The methodology relies on neural network models and super-vised classifiers to identify illnesses. It utilizes word embedding to capture semantic relationships among hospital admissions, symptoms, and diagnoses. Additionally, it introduces a method to evaluate the connection between different diagnoses based on symptom similarity, aiding in prediction tasks. Experimental results on a real-world Electronic Health Records (EHR) dataset showcase the effectiveness and accuracy of the proposed technique, providing clinically relevant interpretations. These findings suggest promising avenues for future enhancements of the framework as a valuable diagnostic tool.