Artificial Intelligence Algorithm for Predicting Cardio-Cerebrovascular Risk in Type 2 Diabetes Shows Strong Performance but Needs Refinement

Strategic Insights -

Researchers from the University of Padova have made significant progress in developing an artificial intelligence (AI)-based algorithm for predicting cardio-cerebrovascular complications in patients with type 2 diabetes mellitus (T2D). The study analyzed medical records of 532 T2D patients from the Diabetology Unit in Padova, Italy, using the Metaclinic AI Prediction Module. The results showed that the AI algorithm showed strong predictive ability for cerebrovascular complications, but failed to reliably predict heart disease risk.

Key Takeaways:

  • The AI-based algorithm demonstrated excellent agreement with clinical diagnostics for cerebrovascular disease in both the 'Very high' and 'Low' risk groups.
  • However, the agreement between AI predictions and clinical diagnostics for heart disease was poor, with a Cohen's k coefficient of 0.00 in both risk groups.
  • The study highlights the potential clinical value of the algorithm for cerebrovascular risk assessment, but notes that it needs refinement for cardiac prediction.
  • The research involved a dataset of 532 T2D patients from the Diabetology Unit in Padova, Italy, and was supported by Universita Degli Studi Di Padova.
  • Francesco Piarulli, Eugenio Ragazzi, Chiara Celeste Celsan, Annunziata Lapolla, and Giovanni Sartore were among the researchers who conducted this study.
  • The study's findings suggest that AI-based algorithms may be useful for predicting cardiovascular complications in patients with T2D, but further refinement is needed to improve their accuracy.

Statistics:

  • 532: The number of T2D patients whose medical records were analyzed in the study.
  • 63: The number of patients identified as 'Very high' risk for heart disease.
  • 122: The number of patients identified as 'Low' risk for heart disease.
  • k = 0.00: The Cohen's k coefficient for agreement between AI predictions and clinical diagnostics for heart disease in both risk groups.
  • k = 0.89: The Cohen's k coefficient for agreement between AI predictions and clinical diagnostics for cerebrovascular disease in the 'Very high' risk group.
  • k = 0.83: The Cohen's k coefficient for agreement between AI predictions and clinical diagnostics for cerebrovascular disease in the 'Low' risk group.

Sources:

  • Piarulli, F., et al. (2025). Artificial intelligence algorithm for predicting cardio-cerebrovascular risk in type 2 diabetes: concordance with clinical and instrumental assessments. Diabetology & Metabolic Syndrome, 17(1), 1-11.
  • Diabetology & Metabolic Syndrome. (2025). Retrieved from http://www.dmsjournal.com/
  • BMC. (Publisher).