Applied Machine Learning for Early Diabetes Detection Based on Symptoms
Abstract
Purpose: Diabetes is a chronic disease that occurs either when the pancreas does not produce enough insulin or when the body cannot effectively use the insulin it produce. Diabetes is often referred to as a silent killer because this disease can affect all organs of the body and cause various symptoms. About 422 million people worldwide have diabetes, the majority living in low-and middle-income countries, and 1.5 million deaths are directly attributed to diabetes each year. Early diabetes detection is essential to prevent serious complications in patients based on symptoms.
Method: This study present a prediction using various Machine Learning (ML) algorithm based on age, gender and symptoms as predictor such as polyuria, feeling thirsty, easy itching, losing weight unintentionally, blurred vision, irritability and feeling tired. We have used such a dataset of 520 patients, which has been collected using direct questionnaires from the patients of Sylhet Diabetes Hospital, Bangladesh.
Results: This study compared several machine learning algorithms such as Logistic Regression, Naive Bayes, Classification and Regression Trees (CART), K-Nearest Neighbour, and Random Forest to develop diabetes prediction model. Several parameter, including classification accuracy (CA), F1 score, precision, and recall were used to evaluate the models. CART algorithm showed better parameter values, with CA 97,1%, recall 0.953, precision 0.932, and F1 score 0.901.
Conclusion: The use of machine learning models for early detection of diabetes with an accuracy rate of 97,1%. ML offers the ability to develop a quick prediction model for diabetes screening based on symptoms. We hope that with this study can contribute to the wider community by decrease the incidence of diabetes through recognizing suspicious symptoms. To prevent diabetes the future this machine learning model can be developed into a mobile application that the public can widely access.
