Patients who suffer from diabetes are at a much higher risk of experiencing a hypoglycemic event. In other words, their blood sugar may drop to dangerously low levels. Typically, patients need to prick their fingers multiple times per day to obtain a small blood sample to test their blood sugar level.
However, a team from the University of Warwick believes that there is a better way. They created an artificial intelligence (AI) system that is capable of detecting low blood sugar readings from electrocardiograph (ECG) data. This is obtained by a wearable device that is far less invasive than a finger prick. The discovery may revolutionize blood sugar monitoring for diabetic patients.
The innovative new monitoring method from the University of Warwick was researched by Dr. Leandro Pecchia. Using two pilot studies with healthy volunteers his team discovered that AI can detect tiny variations in ECG readings that occur when a person’s blood sugar drops to low levels.
Rather than needing to prick their finger, a patient simply wears a heartrate monitor that can be as simple as an off-the-shelf device. Considering the prevalence of wearables in today’s world, this technique could be extremely useful.
Pecchia says, “Fingerpricks are never pleasant and in some circumstances are particularly cumbersome. Taking fingerprick during the night certainly is unpleasant, especially for patients in [the] pediatric age.”
The study and its corresponding research were published in the journal Scientific Reports on Monday.
Pecchia goes on to say, “Our innovation consisted in using artificial intelligence for automatic detecting hypoglycemia via few ECG beats. This is relevant because ECG can be detected in any circumstance, including sleeping.”
Although the technique itself is certainly impressive, the manner in which it was discovered is noteworthy in its own right. Pecchia and his team analyzed data from individual patients rather than looking at that of a cohort. They used each patient’s unique data to train the AI system that would eventually detect their blood sugar levels.
According to the team, differences in the readouts between subjects caused problems when trying to train the AI with cohort ECG data. However, by taking a personalized approach and teaching the AI with each patient’s data, the team was able to create a system that works with remarkable accuracy.
Pecchia claims that the AI can detect low blood sugars with an accuracy rate of 82 percent. Though that might not sound too impressive it is on par with traditional finger-prick monitors. However, with that same level of accuracy comes the increased benefit of not needing to be poked several times per day.
Pecchia says, “Our approach enable[s] personalized tuning of detection algorithms and emphasize[s] how hypoglycemic events affect ECG in individuals. Based on this information, clinicians can adapt the therapy to each individual.”
The team is now looking for partners to help test the system with a wider range of patients. Thanks to the incredible processing of AI, breakthroughs like this one will help people with diabetes all around the world. It also serves as a prime example of the great things that can happen at the intersection of technology and medicine.