Deep learning approach for diabetes prediction using PIMA Indian dataset.

Area of Science:

• Computational biology and bioinformatics
• Medical informatics and health data science
• Machine learning applications in healthcare

Background:

• Diabetes mellitus is a growing global health crisis, projected to affect 629 million people by 2045.
• Early detection and intervention are vital for preventing severe complications like heart disease and stroke.
• Healthcare generates vast datasets, but extracting actionable insights for early diagnosis remains challenging.

Purpose of the Study:

• To develop an effective prognostic tool for early diabetes detection.
• To leverage machine learning algorithms for identifying patterns indicative of diabetes onset.
• To provide a data-driven approach for recommending lifestyle changes to prevent disease progression.

Main Methods:

• Utilized the PIMA dataset for training and evaluating diverse machine learning algorithms.
• Implemented and compared Artificial Neural Network (ANN), Naive Bayes (NB), Decision Tree (DT), and Deep Learning (DL) models.
• Focused on extracting hidden patterns within electronic health records and other omics data for prediction.

Main Results:

• All evaluated classifiers achieved high accuracy, ranging from 90% to 98%.
• Deep Learning (DL) demonstrated superior performance, achieving an accuracy rate of 98.07% on the PIMA dataset.
• The proposed system effectively functions as a prognostic tool for healthcare professionals.

Conclusions:

• Deep Learning (DL) models provide the most accurate predictions for diabetes onset using the PIMA dataset.
• The 98.07% accuracy of DL indicates its potential for developing an automated prognosis tool.
• Future enhancements could involve integrating omics data to further improve DL model accuracy for disease prediction.