The Study
Opportunistic screening of type 2 diabetes with deep metric learning using electronic health records
This study found that a computer program can look at people's medical records and guess who might get diabetes in the future, based on things like weight and medications. But it didn't change anyone's treatment or prove that catching it early helps—it just noticed patterns.
Analysis score
Maximum 0 for a computational/algorithm study.
Where the score came from
Scientists taught a computer to read doctors' notes and test results to find people who might get type 2 diabetes years in advance—and to group them into different types based on their health patterns.
Where does this study sit?
Reviews of RCTs (Meta-analyses)
Max 100Randomized Trials
Max 90Reviews of Cohort Studies
Max 85Cohort Studies
Max 72Reviews of Case-Control Studies
Max 63Case-Control Studies
Max 58Cross-Sectional & Case Series
Max 50Expert Opinion
Max 50 / 100
Quality score
Based on clinical experience or non-systematic literature reviews. The lowest level of evidence as they are most susceptible to bias and personal perspective.
Key takeaways
Summary
Based on the study abstract and findings.
- 1Yes—this means doctors could spot high-risk patients earlier and give the right treatment to the right group, like giving metformin sooner to those who respond best.
- 2The computer predicted diabetes 7 years ahead with 75.4% accuracy (AUC 0.754).
- 3It found 3 types: one (Green) had fewer health problems and lowered blood sugar by 0.64% after metformin; another (Red) had more problems and only lowered it by 0.27%.
Score breakdown, methodology, conflicts of interest, evidence analysis & raw study data
Publication
Journal
Scientific Reports
Year
2025
Authors
Qixuan Jin, Haoran Zhang, L. Szczerbiński, Jiacheng Zhu, Walter Gerych, Xuhai Xu, Kai Wang, Sarah Hsu, Ravi Mandla, Aaron J Deutsch, Alisa K. Manning, Josep M. Mercader, Thomas Hartvigsen, M. Udler, Marzyeh Ghassemi
Related Content
Claims (6)
Type 2 diabetes subtypes identified using deep metric learning show no meaningful connection to inherited genetic risk scores, suggesting these subtypes are shaped by lifestyle and clinical factors instead of genetic inheritance.
A machine learning model using routine medical record data can identify people who will develop type 2 diabetes seven years before diagnosis, and it does so more accurately than current methods based on standard risk factors or blood sugar measurements.
Using electronic health records and machine learning, researchers have identified three distinct forms of type 2 diabetes. One form, labeled Red, is consistently associated with higher rates of obesity-related conditions, cardiovascular disease, and mental health disorders than another form, labeled Green. These groupings are confirmed in two separate patient datasets.
Among people with type 2 diabetes, those classified as having fewer other health conditions show a larger drop in blood sugar levels after starting metformin than those classified as having more health conditions.
A machine learning model using routine clinical data from electronic health records can classify type 2 diabetes into distinct subtypes, and these subtypes are consistently identifiable across two large U.S. healthcare systems without genetic tests or specialized biomarkers.
Deep learning models can identify young-onset type 2 diabetes using routine medical records from primary and secondary care.
Not medical advice. For informational purposes only. Always consult a qualified healthcare professional before making health decisions.