Web based Diabetes Prediction System with ML and Probabilistic Risk Stratification: Evaluation and Analysis

Abstract

Type 1 diabetes, a metabolic condition marked by elevated blood sugar, has become much more prevalent among young individuals. Early identification is essential since it is a chronic illness with a protracted incubation period. The lack of clear beginning symptoms might cause therapy to be delayed. Chronic damage and malfunction of many tissues, including the eyes (Diabetic retinopathy), kidneys (Diabetic Nephropathy), heart(cardiovascular), blood vessels (peripheral arterial), and nerves (Diabetic Neuropathy), may result from long-term high blood sugar levels. It is crucial to diagnose diabetes early. To do this, a number of factors are examined, including age, pregnancy, glucose, blood pressure, body mass index (BMI), insulin, and skin thickness. A comparative analysis of different algorithms is conducted to determine the most accurate one for predicting diabetes. Important keywords include type 1 diabetes, chronic disease, early detection, high blood sugar, chronic damage, predictive algorithms, and attributes analysis. The field of machine learning is becoming increasingly important in data science, as it focuses on how machines can learn from experience. The objective of this research is to combine multiple machine learning approaches to develop a system that reliably predicts the early development of diabetes in individuals. To accomplish this, the model's accuracy is determined using techniques like distance-based algorithm, binary regression, Classification and Regression Tree(CART). The most accurate algorithm is then chosen for estimating the probability of diabetes. The main objective of the study is to increase the precision of diabetes prediction by using the capabilities of machine learning.