health
Modelo de IA predice riesgos de diabetes y cardiovasculares con datos de la glucosa
Un equipo internacional de investigadores ha desarrollado un sistema de inteligencia artificial, que han bautizado como ‘GluFormer’, capaz de predecir el riesgo de desarrollar diabetes […]
TL;DR
- An international research team developed an AI system called 'GluFormer' to predict diabetes and cardiovascular event risk using continuous glucose monitoring data.
- GluFormer was trained on over 10 million glucose measurements from 10,000 adults and successfully applied to 19 external cohorts.
- The AI model improved predictions of glucose levels and other parameters, outperforming traditional basal measurements.
- In prediabetes patients, GluFormer identified those with a higher likelihood of clinically relevant glucose level increases within two years.
- The system was validated in a long-term study, accurately predicting diabetes risk and cardiovascular mortality.
- A multimodal version incorporating dietary data could predict individual glucose responses to different foods.
- Researchers believe GluFormer can make continuous glucose monitoring a predictive tool for risk stratification, enabling earlier interventions and personalized medicine in metabolic health.