Listening to Hypoglycemia – Voice as a Biomarker for Detection of a Medical Emergency using Machine Learning

<p dir="ltr">Objective</p><p dir="ltr">Hypoglycemia is a hazardous diabetes-related emergency. We aimed to develop a machine learning (ML) approach for noninvasive hypoglycemia detection using voice data.</p><p dir="ltr">Research Design and Methods</p><p dir="ltr">We collected voice data (540 recordings) with a smartphone in standardized euglycemia and hypoglycemia in two sequential clinical studies in people with type 1 diabetes. Using this data, we trained and evaluated an ML approach to detect hypoglycemia solely based on voice features.</p><p dir="ltr">Results</p><p dir="ltr">Twenty-two individuals were included (11 female, age 37.3±12.4y, HbA1c 7.1±0.5%). The ML approach detected hypoglycemia non-invasively with high accuracy (area under the receiver operating characteristic curve [AUROC] of 0.90±0.12 for reading a text aloud and 0.87±0.15 for rapid repetition of syllables [diadochokinetic task]). </p><p dir="ltr">Conclusion</p><p dir="ltr">An ML approach exclusively based on voice data allows for noninvasive hypoglycemia detection, corroborating the potential of ML-based approaches to infer acute health states through voice.</p>