Data Driven Predictive AI Systems For Medical Diseases

Abstract: This paper presents an intelligent, data‑driven predictive artificial intelligence system for early diagnosis of four major medical conditions: vitamin deficiency, heart disease, stroke, and diabetes. The system integrates image‑based and structured‑data modes of analysis. A convolutional neural network (CNN) processes clinical images—such as ocular, nail, or lingual photographs—to detect signs of vitamin deficiencies. Meanwhile, decision tree and random forest classifiers are trained on structured patient data to estimate the probability of heart disease, stroke, or diabetes. The architecture features a dual‑interface design: a Flask‑based web API handles model inference, data ingestion, and prediction delivery, while a C# Windows Forms application serves as a secure admin console for user authentication, message management (text and multimedia), and integration with the predictive engine. Results indicate the system’s potential for accelerated, non‑invasive screening support.

Keywords: Deep learning, convolutional neural network (CNN), machine learning, decision tree, random forest, Windows Forms application.