AI-Enhanced Wearables Could Transform Type 2 Diabetes and Prediabetes Care

Artificial intelligence (AI)–powered wearable devices, particularly continuous glucose monitors, are rapidly changing how people with prediabetes and Type 2 diabetes understand and manage their blood sugar. By collecting data every few minutes rather than a few times a day, these tools offer a far more detailed picture of glucose patterns. However, research in this area has been fragmented, often focusing on specific devices or AI models, making it difficult to assess the field as a whole.

To address this gap, researchers from the University at Buffalo (Buffalo, NY, USA) conducted the first comprehensive meta-review of AI-enhanced wearable technologies for prediabetes and Type 2 diabetes. Published in NPJ Digital Medicine, the study reviewed evidence from the past several years and concluded that while these devices hold enormous potential, important challenges must be resolved before they become routine tools in clinical care.

The research team analyzed 60 high-quality studies selected from nearly 5,000 peer-reviewed papers that examined the integration of artificial intelligence and wearable technology in diabetes management. The findings show that AI-enhanced wearables can predict glucose changes up to one to two hours in advance, allowing individuals to anticipate and avoid dangerous swings. These systems can also personalize guidance based on daily routines, physical activity, sleep patterns, and stress levels, while helping clinicians manage large volumes of patient data more efficiently.

Despite these advantages, the review highlights several limitations. Many AI systems operate as “black boxes,” providing predictions without explaining the underlying reasoning. This lack of transparency can make it difficult for patients and clinicians to trust or act on the recommendations. Other challenges include small and homogeneous study populations, limited external validation, the absence of standardized datasets, inconsistent data quality, and barriers related to cost and integration into existing clinical workflows.

The researchers also found that the performance of AI-enhanced wearables depends heavily on the type of AI model used. Time-series models such as long short-term memory networks are well-suited for continuous glucose data, while newer transformer-based models can integrate multiple data streams, including heart rate, sleep, and activity. However, simpler models may be easier to interpret and adopt in clinical settings, highlighting the need to balance performance with explainability.

Overall, the study suggests that AI-enhanced wearables could support earlier risk detection, more personalized care, and potentially delay progression from prediabetes to diabetes. However, broader validation, improved transparency, and better clinical integration will be essential before these technologies can reach their full potential.

“For people living with diabetes, AI-enabled wearables have the potential to provide more timely and personalized guidance, helping them avoid glucose swings and manage daily decisions with greater confidence,” said Raphael Fraser, PhD, corresponding author and associate professor of medicine. “For clinicians, the key takeaway is that these tools may help identify risks earlier and support more efficient care.”

Related Links:
University at Buffalo