Decoding Human Diseases through Symptom Encoding and Support Vector Machines

Abstract

Early disease diagnosis improves treatment outcomes and reduces healthcare burdens. This study develops a predictive model using symptom encoding and the Support Vector Machine (SVM) algorithm. Symptoms are encoded into a structured format using one-hot encoding, and SVM is trained with different kernel functions (linear, polynomial, RBF) to optimize accuracy.
Preprocessing techniques like dimensionality reduction and feature scaling enhance model performance. Evaluation metrics, including accuracy, precision, recall, F1-score, and ROC-AUC, confirm the model's reliability in disease prediction. This research highlights machine learning’s role in healthcare, offering a scalable diagnostic tool. Future improvements may integrate additional datasets, ensemble learning, and real-time symptom data.