AI Wearables Target Diabetes Precision
A new review in Health and Metabolism discusses wearable CGMs and smartwatches integrated with AI analytics for precision diabetes care, combining real-time data with multi-omics analysis. Meanwhile, Excel users are sharing patient medical analysis dashboards tracking total patients, costs, insurance status, and payment data. A bioRxiv preprint also explored federated learning for diabetes prediction under non-IID data conditions.
AI is shifting continuous glucose monitors (CGM) from reactive to predictive tools. Instead of just showing past glucose levels, machine learning algorithms can now anticipate fluctuations, with some models forecasting changes up to two hours in advance, giving users a crucial window to act. This predictive power comes from analyzing real-time data streams alongside personal health metrics. Leading systems from companies like Medtronic and Dexcom are increasingly incorporating AI to improve the Mean Absolute Relative Difference (MARD), a key metric for glucose monitor accuracy where a lower percentage indicates higher accuracy. The integration of multi-omics—genomics, proteomics, metabolomics—provides a more holistic view of an individual's diabetes. By analyzing how genes, proteins, and metabolites interact, researchers can move beyond broad categorizations like "Type 2" to identify specific disease subtypes. This allows for more precise, personalized treatment strategies that target the underlying molecular drivers of the disease in each person. Federated learning, a method of training AI models across multiple decentralized devices or servers holding local data samples, faces a key hurdle with non-IID (non-independent and identically distributed) data. In healthcare, this means patient data from one hospital can be vastly different from another due to demographics or local practices, which can skew the AI model's accuracy and fairness. To combat this, researchers are developing new algorithms that can account for these variations without compromising patient privacy. Startups like Flower Labs and Sherpa.ai are creating platforms that allow for collaborative model training on this heterogeneous data, a crucial step for building robust and unbiased predictive tools. The investment landscape reflects this technological push, with digital health startups raising $14.2 billion in the U.S. in 2025, a significant increase from previous years. Companies focusing on AI-driven healthcare solutions captured the majority of this funding, signaling strong investor confidence in the technology's potential to transform chronic disease management.
