Cardiovascular complications in a diabetes prediction model using machine learning: a systematic review
Problem
Predicting cardiovascular disease risk in patients with type 2 diabetes using machine learning-based prediction models.
Why it matters
Type 2 diabetes patients face high cardiovascular risk, and early prediction could enable preventive care.
What the AI did
Systematic review of ML prediction models for CVD risk in T2DM, with PROBAST risk-of-bias assessment; neural networks identified as the most reliable algorithm.
Results
Neural network: 76.6% precision, 88.06% sensitivity, AUC 0.91; overall applicability concern low
Impact
Review suggests neural networks are reliable for CVD prediction in diabetics, but no new deployed model is shown.
Record
| Venue | Cardiovascular Diabetology |
|---|---|
| Published | 2023 |
| Work type | review |
| Authors | Ooi Ting Kee, Harmiza Harun, Norlaila Mustafa, Nor Azian Abdul Murad, Siok‐Fong Chin, Rosmina Jaafar, Noraidatulakma Abdullah |
| Citations | 80 (OpenAlex, at import time) |
| References | 62 |
| Model or method | Neural network (identified as most reliable) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1186/s12933-023-01741-7 |
| Open access | gold |
| Keywords | machine learning, type 2 diabetes, cardiovascular disease, prediction model, systematic review, neural network |
Importance rating
real but narrower advance
Systematic review of ML models for CVD risk in T2DM; neural network most reliable (AUC 0.91), but no new deployed model and abstract states no scale of affected patients.