Researchers Develop PNN Predicting Prediabetes Risk
In a 2025 Scientific Reports study, researchers developed a pattern neural network (PNN) that combined total antioxidant status with clinical markers to predict prediabetes in 199 Indian adults. The PNN achieved overall accuracy of 98.3% and outperformed SVM, KNN, and logistic regression, with waist circumference and antioxidant activity as top predictors. Authors note single-center, cross-sectional design and recommend external validation.
Key Points
- 1Demonstrates PNN combining antioxidant status and clinical markers achieves 98.3% accuracy predicting prediabetes in 199 adults
- 2Highlights oxidative stress (lower antioxidant capacity) as a strong, mechanistically relevant predictor alongside waist circumference
- 3Enables potential low-cost, rapid screening tool subject to external validation and multi-site prospective testing
Scoring Rationale
Peer-reviewed high-accuracy demonstration with strong predictive signals, but limited by single-center design and modest sample size requiring validation.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Health & Insurance problems
