An Empirical Model to Predict the Diabetic Positive Using Stacked Ensemble Approach

Area of Science:

• Medical Informatics
• Computational Biology
• Public Health

Background:

• Diabetes mellitus is a global health concern with rising incidence, necessitating improved diagnostic methods.
• Current diabetes detection relies on manual processes, highlighting the need for automated solutions.
• Single predictive algorithms often fall short for complex diseases like diabetes.

Purpose of the Study:

• To develop and evaluate a novel stacked ensemble model for automated diabetes prediction.
• To compare the performance of the proposed heterogeneous ensemble model against traditional single-algorithm approaches.
• To enhance the accuracy and reliability of diabetes diagnosis in healthcare systems.

Main Methods:

• Implementation of a heterogeneous ensemble model, specifically a stacked ensemble, for diabetes prediction.
• Utilizing multiple algorithms within the ensemble to leverage diverse predictive strengths.
• Comparative analysis against established methods including logistic regression, Naïve Bayes, and Linear Discriminant Analysis (LDA).

Main Results:

• The proposed stacked ensemble model achieved a prediction accuracy of 93.1% for diabetes.
• This accuracy significantly surpasses existing methods: logistic regression (72%), Naïve Bayes (74.4%), and LDA (81%).
• The stacked ensemble approach demonstrates superior performance in identifying diabetes cases.

Conclusions:

• Heterogeneous ensemble models, particularly stacked ensembles, offer a powerful and accurate solution for automated diabetes detection.
• The developed model provides a significant advancement over traditional single-algorithm predictive techniques.
• This research supports the integration of advanced machine learning for more effective diabetes management and diagnosis.