A Hybrid Recommendation System for Diabetes Prediction using Leabra-Based Multi-Head Self-Attention

Abstract

The escalating prevalence of diabetes presents an urgent need for innovative predictive solutions. This study proposes a novel hybrid recommendation system that integrates Leabra-based multi-head self-attention to enhance diabetes prediction accuracy. Diabetes prediction is a complex task due to multifaceted interdependencies among various factors. Conventional methods often fall short in capturing these intricate relationships. To address this, our Prediction of Diabetes Hybrid Recommendation System (PDHRS) leverages the power of Leabra-based multi-head self-attention (L-MHSA). This framework analyzes diverse patient attributes, health indicators, and lifestyle factors, generating accurate predictions. The system adapts a multi-head self-attention mechanism inspired by cognitive neuroscience principles, allowing for nuanced feature extraction and comprehensive pattern recognition. Through this innovative approach, PDHRS offers a new dimension of accuracy and interpretability in diabetes prediction, potentially revolutionizing proactive healthcare interventions.