Links between physical activity, time-in-range and glucose predictability in people with type 1 diabetes.

BACKGROUND

Type 1 diabetes (T1D) management remains particularly challenging in the presence of external disturbances such as stress or physical activity (PA). The glycemic impact of PA is still not fully understood and lacks standardized modeling approaches. This work seeks to describe the relationship between PA, time-in-range and glucose predictability in people with T1D (PwT1D).

METHODS

This study uses data from the Type 1 Diabetes and Exercise Initiative (T1DEXI) clinical trial, the largest trial to date in PwT1D undergoing both free-living and structured exercise. A characterization pipeline extracts summary statistics including glucose, carbohydrate intake and PA signals. These features are used to perform unsupervised clustering of subjects using various techniques. The relevance of the cluster-defining variables is then assessed, and their relationship to the performance of a long short-term memory (LSTM) neural network trained to forecast glucose 1 h into the future is analyzed.

RESULTS

The spectral clustering algorithm successfully separates individuals into three groups based on glycemic control metrics. Results indicate that subjects with higher levels of weekly PA exhibit lower prediction errors. This suggests that regular PA enhances the predictability of glucose trends, enabling more accurate forecasting models.

CONCLUSION

Since fear of hypoglycemia remains one of the main barriers to PA in PwT1D, these findings are particularly relevant: regular exercise not only promotes better glycemic regulation but also improves the performance of predictive models, which could strengthen automated insulin delivery systems, support more reliable decision-support tools and contribute to safer and more confident engagement in PA.