Recent research, conducted by scientists at Klick Labs, explores the use of AI to identify type 2 diabetes based on voice analysis. This innovative approach leverages subtle vocal changes affected by emotions, intoxication, or tiredness to detect health shifts. The study involved 267 participants who recorded a fixed phrase multiple times daily over two weeks. AI extracted 14 vocal characteristics, such as pitch and intensity, achieving an 89 percent accuracy for women and 86 percent for men, considering factors like age and sex. Larger, diverse testing groups are needed, but this technology offers a promising, non-invasive alternative to traditional blood tests for diabetes screening.
Source: https://www.mcpdigitalhealth.org/article/S2949-7612(23)00073-1/fulltext