Machine Learning Models Can Identify Diabetes from Limited Patient Data

Researchers used machine learning models on physical exam data to identify diabetic individuals among those with normal fasting glucose.
The study collected data from 3 hospitals and developed a computational model, with a deep neural network showing the best performance.
Key distinguishing features between diabetic and non-diabetic patients were identified, like BMI and age.
An optimal set of 13 features was determined to build robust models, which were validated on independent test sets.
A web tool called DRING was created to allow practical application of the models for screening undiagnosed diabetic cases.

news-medical.net

Relevant topic timeline:

8/31/2023

India's AI biotech takes on diabetes epidemic

India has the highest number of people with diabetes, with an estimated 101 million individuals living with the condition, and artificial intelligence technology offers potential solutions for early detection and personalized treatment.

9/4/2023

Machine learning helps identify metabolic biomarkers that could predict cancer risk

A study from the University of South Australia has utilized machine learning to identify metabolic biomarkers that can predict the risk of cancer, offering the potential for early detection and prevention.