پیشبینی خطر ابتلا به دیابت نوع ۲ با استفاده از رویکرد جنگل تصادفی
کد: G-1007
نویسندگان: Amir Mohammadi ℗, محمد رجبی *
زمان بندی: زمان بندی نشده!
برچسب: سیستم های تصمیم یار بالینی
دانلود: دانلود پوستر
خلاصه مقاله:
خلاصه مقاله
Background and Aims: Type 2 Diabetes Mellitus (T2DM) is a major risk factor for cardiovascular diseases and is widely recognized as a significant clinical and public health issue. Early diagnosis can help mitigate the impact of the disease. This study aimed to present a simple model for predicting T2DM using the Random Forest method, which can be utilized in telemedicine devices and mobile applications. Method: In this study, we selected the Random Forest classifier as our predictive model. The Random Forest algorithm is an ensemble learning technique that builds multiple decision trees during training and makes predictions by selecting the most frequent class for classification or averaging the predictions of individual trees for regression. Using this model, we investigated various health-related factors and their interactions to classify diabetes accurately. We identified the most important features for predicting diabetes using the Random Forest model. The importance of each feature was measured by how much it reduced uncertainty in the decision-making process across all trees in the model. These features included factors such as age, gender, Body Mass Index (BMI), HbA1c level (a measure of average blood sugar over time), and blood glucose level. Additionally, other relevant features included family history of diabetes, blood pressure, cholesterol levels, lifestyle factors such as physical activity and diet, and any other medical conditions or biomarkers that may influence the likelihood of developing diabetes. Results: The model achieved an accuracy of approximately 95%, with a precision of 0.98 for class A (non-diabetic) and 0.7 for class B (diabetic). Additionally, it correctly recalled 96% of non-diabetic cases and 80% of diabetic cases. The relatively high accuracy, along with balanced performance across both classes, suggests that the model is well-calibrated and robust. Conclusion: The results indicate that T2DM can be predicted using a decision tree model without the need for laboratory tests. Therefore, this model could be applied in pre-clinical and public health screening programs.
کلمات کلیدی
Classification, Desicion Tree, Diabetes, Random Forest
