Applying Artificial Intelligence for Enhanced Precision in Early Disease Diagnosis from Healthcare Dataset Analytics

Background: Accurate illness diagnosis is critical for effective treatment planning and patient well-being. Diabetes, in particular, demands early detection to manage and prevent complications. Artificial Intelligence (AI) and Machine Learning (ML) have shown promise in enhancing diagnostic accuracy in healthcare. Methodology: This study utilizes the Pima Indian Diabetes Dataset. The data undergoes preprocessing and normalization. To address class imbalance, SMOTE is applied. For this work proposed Bi-LSTM model and compared with other existing ML models CNN, KNN, and Naive Bayes for diabetes diagnosis. Results: It is possible to measure the efficacy of a model using measures like F1-score, recall, accuracy, and precision. The Bi-LSTM model outperforms others, achieving 98.85% accuracy, 98% precision, 94% recall, and a 96% F1-score, demonstrating superior diagnostic capabilities. Conclusion: The study highlights the effectiveness of ML, especially Bi-LSTM, in early diabetes detection, emphasizing AI's role in improving healthcare delivery and patient outcomes.