Artificial Intelligence Driven Predictive Analytics for Early Disease Detection Using Machine Learning Techniques

Abstract

As technology has advanced rapidly in the healthcare industry, the use of machine learning techniques has become vital for disease diagnosis and clinical decision-making. Smart solutions are also employed in hospitals and healthcare companies to support disease detection at an early stage by making use of medical records and clinical information of patients. The research looks at a CNN-BiLSTM model that combines CNN and BiLSTM to find the disease early on. However, the UCI Heart Disease dataset has problems finding the disease. Training and testing use a 70:30 split of data, and the technique includes data pre-processing steps such outlier removal, Min-Max normalisation, and missing value imputation. Improve prediction accuracy with the help of the suggested CNN-BiLSTM model's spatial feature extraction and forward and backward learning sequence capabilities. Based on the experimental findings, the suggested model achieves better performance than the traditional ML models, such as MLP (0.99), AdaBoost (0.99), Logistic Regression (0.97), and XGBoost (0.96). It achieves an ACC of 99.62%, PRE of 99.95%, REC of 99.94%, and an F1 score of 99.98%. These results suggest that the hybrid approach is a comprehensive and reliable tool for early disease detection that allows timely clinical interventions and better patient management.