Innovative Machine Learning Approaches for Prediction of Hypoglycemia in Patients with Type 2 Diabetes

Aug 3, 2026

Upload by Miles

Medical data science advances using machine learning to predict glucose levels and mitigate hypoglycemia risk in type 2 diabetes. The work applies supervised methods using regression and classification to evaluate prediction performance, and uses unsupervised learning to form clusters from variable similarity. A transfer learning strategy improves conventional model prediction accuracy. Data are split by a median glucose value of 67 mg/dL into two groups, with regression using 5-fold cross-validation and classification using 10-fold cross-validation. Unsupervised clustering feeds transfer learning for combined modeling.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages19

Reading Time48 min

AccessFree to download

Abstract

Introduction

Diabetes and hypoglycemia background

Clinical classification and risks

Limitations of existing tools

Motivation for machine learning prediction

Innovative machine learning approaches for prediction of hypoglycemia in patients with type 2 diabetes Sachin Ramnath Gaikwad1, Seeta Devi2, Sameer Shekhar3, Dipali Dumbre4 1Department of Artificial Intelligence and Machine Learning, Symbiosis Institute of Technology, Symbiosis International Deemed University, Pune, India 2Department of Obstetrics and Gynaecological Nursing, Symbiosis College of Nursing, Symbiosis International Deemed University, Pune, India 3Department of Computer Science and Engineering, Symbiosis Institute of Technology, Symbiosis International Deemed University, Pune, India 4Department of Medical and Surgical Nursing, Symbiosis College of Nursing , Symbiosis International Deemed University, Pune, India | Article history:
Received Nov 18, 2023 Revised Mar 23, 2024 Accepted Jun 1, 2024 | Medical data science advances using machine learning, which predicts glucose levels. A supervised machine learning technique is employed in which regression and classification methods are used to check the prediction performance. The unsupervised machine learning technique makes clusters based on variables' similarities. Furthermore, the prediction accuracy of conventional machine learning techniques is improved by proposing a transfer learning technique. Based on a median value of 67 mg/dL, the data set is divided into two groups: group 1 (BSL 57 mg/dL to 67 mg/dL) has 50.67% of the samples, and group 2 (with BSL 68 mg/dL to 79 mg/dL) has 49.33% of the samples. In regression analysis, 5-fold cross-validation is performed. The decision tree (DT) and gradient boosting (GB) individually provide a prediction accuracy of 18.2% . Regarding classification analysis, a 10-fold cross-validation configuration is used for training and testing the model. AdaBoost, GB, random forest, and neural network achieve an accuracy rate of 66.3% and an area under curve (AUC) score of 0.731. In unsupervised learning, the datasets are divided into three clusters. The clustering result is used in regression and classification models using transfer learning. The accuracy and precision of the AdaBoost and GB are as follows: 69.6%, 0.696 with f1 0.661 and 69.6%, 0.708 with f1 0.708, respectively.
This is an open access article under the CC BY-SA license.
| | --- | --- | | Keywords:
High risk factors Hypoglycaemia Innovative machine learning
approaches Predicition Type 2 diabetes | | Corresponding Author: Seeta Devi Symbiosis College of Nursing, Symbiosis International Deemed University Pune 412115, India Email: [drseetadevi1981@gmail.com](drseetadevi1981@gmail.com) Article Info ABSTRACT 1. INTRODUCTION Diabetes mellitus (DM) is a long-term medical disorder that involves high blood glucose (BG) levels caused by an inability to make or use insulin properly. Keeping normal blood sugar levels can help prevent diabetic complications affecting both minor and large blood vessels [1]–[4] . Tight management of blood sugar levels elevates the chances of experiencing hypoglycemic episodes but concurrently diminutions the likelihood of enduring chronic complications [5] . Hypoglycaemia is commonly detected as a blood sugar level of less than 70 mg/dL, which causes two physiological responses: a counterregulatory reaction and decreased cognitive ability [6]–[8] . Hypoglycemia can cause significant cardiovascular events and results in fear of low blood sugar, melancholy, and abnormal muscular movements, in addition to depression and deviation in physical well-being [9]–[13] . Clinically, hypoglycemia is categorized as moderate (MH) or severe (SH) based on whether the patient requires medical assistance during episodes of low blood sugar or if there is a loss of consciousness [14] . It has been estimated that patients with type 1 diabetes (T1D) had around 1.6 episodes of mild hypoglycemia and 0.029 episodes of severe hypoglycemia every week [15] . Furthermore, among people with T2 diabetes receiving treatment, the fre

FAQs

How is hypoglycemia defined in the study?

Hypoglycemia is detected as a blood sugar level of less than 70 mg/dL, which triggers counterregulatory responses and reduced cognitive ability.

Which machine learning approaches are used for prediction?

The study uses supervised learning with regression and classification, unsupervised learning for clustering, and a transfer learning technique to enhance accuracy.

How are the datasets and validation strategies organized?

The dataset is divided into two groups using a median value of 67 mg/dL. Regression uses 5-fold cross-validation, while classification uses a 10-fold cross-validation configuration.

Aug 3, 2026

Upload by Miles