AI is now better than human doctors reading ECGs

A team of researchers from the University of Warwick have built an artificial intelligence they say can detect hyoglycaemic events — drops in blood sugar that can pose a real threat to those with or even without diabetes — looking only at data from a wearable electrocardiograph (ECG). Patients usually have to prick their fingers, sometimes multiple times a day, to analyze small amounts of their blood for blood sugar levels. “Fingerpicks are never pleasant and in some circumstances are particularly cumbersome,” lead researcher Leandro Pecchia said in a statement. “Taking fingerpick during the night certainly is unpleasant, especially for patients in paediatric age.” The results are impressive. Two pilot studies with healthy volunteers found that the AI was 82 percent reliable in detecting hypoglaecemia. That may not sound like much, but it’s about on par with current continuous glucose monitors (CGM) used in hospitals, which require needles to be inserted in the patient, according to the researchers.