Large language models predict blood sugar lows better with clear information display
It's All in the Way You Say It: The Role of Information Representation in LLM-Based Glycemic-Event Prediction
Artificial Intelligence
Summary
Predicting high or low blood sugar is important for people with type 1 diabetes. This study looks at how language-based AI models can predict these events using different ways of showing their health data. The researchers found that these models work better for predicting low blood sugar when the data is presented clearly, but traditional methods still perform best for predicting high blood sugar. Adding extra details like meals or exercise did not always make the models perform better. How the information is shared with the AI plays a big role in prediction accuracy.
Large Language ModelsType 1 DiabetesGlycemic Event PredictionPostprandial HyperglycemiaHypoglycemiaPhysiological Time-SeriesPrompt-based InferenceOhioT1DM DatasetSupervised LearningContextual Information
Authors
Andrea Apicella, Pasquale Arpaia, Matteo Orefice, Andrea Pollastro, Roberto Prevete
Abstract
Large Language Models (LLMs) are increasingly being investigated for physiological time-series prediction, yet their effectiveness may depend not only on the model itself, but also on how physiological information is represented and presented at inference time. This study investigates prompt-based general-purpose LLMs for postprandial hyperglycemia and hypoglycemia prediction in individuals with type 1 diabetes. Using the OhioT1DM dataset, we evaluate multiple open-weight LLMs under zero-shot and few-shot inference across prediction horizons of 30, 60, and 90 minutes. The analysis varies both the textual representation of the available physiological information and the amount of information exposed to the model, ranging from glucose observations alone to derived descriptors and additional contextual variables related to insulin, meals, carbohydrates, and physical activity. Performance is compared with conventional patient-specific supervised models and with Gluco-LLM, a language-model-based architecture explicitly adapted to glucose time-series forecasting. Results show a marked task-dependent behavior. Conventional supervised models achieve the strongest performance for hyperglycemia prediction, whereas the best observed prompt-based LLM configurations improve performance for hypoglycemia across all investigated horizons. The effectiveness of prompt-based inference is also strongly influenced by how physiological information is represented, while providing additional contextual information does not lead to a systematic improvement. Overall, these findings highlight physiological information representation as a central design factor in prompt-based LLM approaches to glycemic-event prediction.