Hypoglycemia — low blood glucose — is a dangerous and potentially fatal condition most commonly experienced by people with Type 1 diabetes. Diabetics aim to keep their blood sugar levels in the healthy 70-180 mg/dL range as much as possible by managing diet, exercise, and insulin injections. Too much glucose in the blood — hyperglycemia — is also dangerous, but the effects tend to build gradually. Hypoglycemia can occur quickly, however, and must be treated quickly.

Medtronic and IBM Watson Health worked together to develop AI-driven hypoglycemia prediction for Medtronic’s IQcast prediction feature in the Sugar.IQ personal diabetes assistant app. The target users for IQcast are diabetics on multiple daily injections (MDI) regimens. The AI app gains insight into a patient’s diabetes management and their time spent under, within, and over the target range. According to the two companies, it can be difficult for patients to predict accurately when they will experience low blood sugar levels. The goal with IQcast is to predict the likelihood that a hypoglycemic event will occur within the next one to four hours. In data presented at the American Diabetes Association Scientific Sessions in 2017 and 2018, patients who used the Medtronic Guardian Connect standalone glucose monitoring system in conjunction with IQcast were able to remain in the healthy glucose range an extra 36 minutes a day: 30 minutes less in hyperglycemia and six fewer minutes in hypoglycemia.

The focus on developing tools that help patients with chronic illness and conditions not only react to health problems but predict when they are likely to occur is a big step forward. Accurate event prediction empowers patients to take preventative measure. This is just one more demonstration that “digital health” in the form of apps on smartphones can have a major impact on improving healthcare.