Machine Learning Models for Analyzing Diabetes Datasets to Predict Disease Progression and Patient Outcomes

Abstract

Machine learning is central in analyzing numerical diabetes datasets to predict disease progression and patient outcomes. The primary objective is to develop a model that accurately forecasts the patient's future health status based on historical and current medical data. Predictive analytics is essential for early intervention and personalized patient care. The chapter focuses on the pressing healthcare challenge of diabetes, a pervasive condition characterized by significant complexity and requiring precise diagnostic measures. Recognizing the ample data available on diabetes and the severe complications associated with the disease, there is an imperative demand for enhancing the accuracy of its diagnosis. The dataset utilized in this chapter was meticulously compiled from patient records within the Iraqi healthcare system, specifically sourced from the Medical City Hospital's laboratory and the Specialized Center for Endocrinology and Diabetes at Al-Kindy Teaching Hospital. To advance the diagnosis and predictive modeling of diabetes, a comprehensive dataset was curated, drawing from a wide range of medical and laboratory analysis records. This dataset was pivotal in fostering a deeper understanding of the multifactorial nature of diabetes within Iraqi society. Support Vector Regression (SVR), Random Forest Regressor, and K-Nearest Neighbors Regressor (KNN) were applied to the tested dataset. In the experimental results, the SVR model showed the most significant improvement postoptimization, with substantial decreases in Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE), and a notable increase in R2, indicating a substantial enhancement in both the model's accuracy and its explanatory power.

Ähnliche Arbeiten