A Sophisticated Onscreen Smart Framework for Predicting Diabetes in Remote Healthcare

Area of Science:

• Artificial Intelligence in Healthcare
• Machine Learning for Disease Prediction
• Deep Learning in Medical Diagnosis

Background:

• Diabetes is a prevalent metabolic disease characterized by prolonged high blood sugar.
• Early diabetes detection is crucial for reducing disease severity and associated risks.
• Existing AI models for diabetes prediction often lack robustness and precision.

Purpose of the Study:

• To design and develop an automated tool for chronic disease diagnosis using a novel AI methodology.
• To enhance the robustness, dependability, and precision of AI-based diabetes prediction.
• To introduce the Brass Optimized Learning-Based Diabetes Prediction (BOLD) model for remote healthcare applications.

Main Methods:

• The BOLD model utilizes optimization-integrated deep learning for enhanced performance.
• Data preprocessing includes splitting, normalization, and cleaning of the diabetes dataset.
• Feature selection is performed using Brassy Pelican Optimization (BPO).
• Classification is conducted using Hunting Optimized Recurrent Neural Network-Long Short-Term Memory (RNN-LSTM) with Deer Hunting Optimization (DHO) for hyperparameter tuning.

Main Results:

• The BOLD framework achieved high performance metrics: RMSE of 0.015, Cohen's Kappa of 0.99, precision of 0.991, recall of 0.99, accuracy of 0.996, and AUC of 0.99.
• The model demonstrated effectiveness across multiple datasets, including PIDD, Indonesia diabetic database, and kidney disease dataset.
• Validation confirmed the framework's ability to accurately predict diabetes.

Conclusions:

• The BOLD model offers a highly accurate and efficient solution for diabetes prediction.
• The framework achieves excellent performance in a remarkably short processing time of 0.8 seconds.
• This combination of speed and accuracy makes the BOLD model a valuable tool for remote healthcare and early disease intervention.