Research Area:  Machine Learning

Abstract:

Diabetes is a metabolic disorder in the world today. The rate of production of diabetic patients is rising day by day. Diabetic disease occurs when the blood glucose level gets high, leading inevitably to other health conditions such as heart disease, kidney disease, etc. Symptoms of diabetes are increased appetite and urination, increased hunger, fatigue, blurred vision, sores that do not heal, unexplained weight loss. People with diabetes are at high risk for diseases such as eye problems, nerve damage, etc. In this paper, we proposed a diabetes prediction model for better diabetes classification that includes a model of a few external diabetes factors along with normal factors such as glucose, age, gender, Blood Pressure, Sugar, Red Blood Cells, Hemoglobin, Blood Urea, etc. We have a dataset that contains 250 variants that individually hold 16 unique attributes. We have used Logistic Regression, Support Vector Machine, and Random Trees for this prediction.10-fold cross-validation had applied for training the data and the accuracy for Logistic Regression is 94.5 %, Support Vector Machine is 96.5% and Random Tree is 97.5%.

Keywords:  
Data Mining
Diabetes Prediction
Data Analysis
Predictive Analysis
Diagnostics
Healthcare
Machine Learning

Author(s) Name:  Md. Mehedi Hassan; Md. Al Mamun Billah; Md. Mushfiqur Rahman; Sadika Zaman; Md. Mehadi Hasan Shakil; Jarif Huda Angon

Publisher name:  IEEE

DOI:  10.1109/ICCCNT51525.2021.9579799

Volume Information:  

Paper Link:   https://ieeexplore.ieee.org/abstract/document/9579799