Non-invasive Diabetes Risk Assessment Model Using Deep Learning

Abstract

This Current Diabetes Detection Methods Predominantly Rely On Invasive Blood Tests, Posing Significant Barriers To Widespread, Proactive Screening. This Paper Proposes The Development Of A Highly Precise And Accurate Non-invasive Predictive Model For Assessing Diabetes Risk. Leveraging A Comprehensive Set Of Readily Observable Symptoms, Anthropometric Measurements, Lifestyle Habits, And Medical History, This Deep Learning/Neural Network Model Aims To Identify Individuals At Risk Without Requiring Blood Samples. Experimental Results Demonstrate The Model's Effective Learning, Convergence, And Strong Predictive Capabilities, With Key Features Like Genetic Risk And Physical Activity Significantly Influencing The Predictions. The Insights Gained From This Research Also Highlight Crucial Non-invasive Pathways For Diabetes Prevention And Reduction. The Proposed Solution Seeks To Serve As An Accessible Initial Screening Tool, A Proactive Awareness Generator, And A Valuable Resource For Public Health Initiatives, Ultimately Reducing The Burden Of Diabetes And Its Associated Complications Through Early Detection And Timely Intervention.

Keywords