An explainable analysis of diabetes mellitus using statistical and artificial intelligence techniques.

Area of Science:

• Computational biology
• Medical informatics
• Data science

Background:

• Diabetes mellitus (DM) is a global chronic disease requiring advanced analytical methods for early detection and management.
• Sociodemographic and clinical data are crucial for understanding and predicting DM.
• Multifaceted approaches are needed to improve patient outcomes and reduce mortality rates.

Purpose of the Study:

• To develop an explainable analysis of DM by integrating sociodemographic and clinical data.
• To compare the predictive performance of various statistical and artificial intelligence (AI) techniques for DM classification.
• To assess the risk factors associated with DM using explainable AI methods.

Main Methods:

• Utilized a dataset comprising sociodemographic and clinical profiles of diabetic and non-diabetic individuals.
• Applied statistical tests including Student's t-test and Chi-square.
• Employed AI techniques such as fuzzy cognitive maps (FCM), artificial neural networks (ANN), support vector machines (SVM), and XGBoost.

Main Results:

• Statistical models identified significant variable associations.
• AI models demonstrated high efficacy in DM classification, with XGBoost achieving perfect accuracy, sensitivity, and specificity.
• FCM provided explainability through scenario-based simulations, highlighting key predictive variables.

Conclusions:

• An integrated analytical approach combining diverse methodologies is essential for timely DM detection.
• Informed clinical decision-making can be enhanced through explainable AI and statistical analysis.
• This research underscores the importance of combining predictive power with interpretability in disease analysis.