Year: 2024
Type: conference-paper
Abstract: Diabetes is a chronic and persistent health issue. Detecting it in the prediabetes phase in a timely manner and reducing the potential complications is the best option. Therefore, it is very necessary to predict whether a patient has diabetes. This article takes the dataset from the Kaggle platform as the research object, conducts sampling data preprocessing on the diabetes dataset; selects the XGBoost model, and introduces the genetic algorithm to optimize its hyperparameters, forming an improved GA-XGBoost model based on the genetic algorithm; analyzes the SHAP values of each feature, and uses the features with greater impact as typical characteristics for whether the patient is ill; through comparative experiments, the improved model shows a significant enhancement in predictive effect compared to the traditional model.
Institution: Sichuan University of Science and Engineering
Language: English
Subfield: Health Information Management
Field: Health Professions
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W4410740123