Advancements in artificial intelligence (AI) continue to revolutionise the medical field, and now researchers have discovered a new application: identifying type 2 diabetes through voice analysis. Scientists from Klick Labs have developed an AI model that can detect subtle changes in a person’s voice, potentially providing a faster and more accessible method for diagnosing this prevalent condition.
Traditional methods of detecting type 2 diabetes often require time-consuming processes, including blood tests and laboratory analysis. However, this innovative AI technology has the potential to remove these barriers entirely. By simply engaging in a conversation, individuals can now contribute to their own diagnosis, making the process more efficient and convenient.
The study conducted by Klick Labs involved 267 participants, both with and without type 2 diabetes. Over a two-week period, these individuals recorded a fixed phrase six times a day using a smartphone app. A staggering total of 18,465 recordings were collected and processed to extract 14 different vocal characteristics, such as pitch and intensity.
Using a subset of these recordings, the researchers trained the AI model to recognise the unique vocal patterns associated with type 2 diabetes. The model was trained based on factors such as sex, age, body mass index (BMI), and the presence or absence of diabetes. The remaining samples were then used to test the accuracy of the AI’s predictions.
The results were impressive, with the AI model achieving an accuracy level of 89 percent for women and 86 percent for men in detecting type 2 diabetes. Interestingly, the key vocal signals that identified the condition differed between genders. In men, intensity and amplitude variation played a crucial role, while variations in pitch were more indicative in women.
Although the initial findings are promising, the researchers acknowledge the need for larger and more diverse groups to validate these results. Nevertheless, this AI technology has the potential to transform how medical professionals screen for type 2 diabetes. The current method of blood analysis and subsequent waiting period for results could be replaced by a simple smartphone app, reducing both time and cost.
Type 2 diabetes affects approximately 1 in 11 adults worldwide, but it is estimated that many individuals remain undiagnosed. This AI technology could help identify those who are unaware of their condition, enabling early intervention and treatment. By reducing the number of undiagnosed cases, the burden of managing diabetes within the population could also be alleviated.
The findings of this study have been published in Mayo Clinic Proceedings: Digital Health, a reputable peer-reviewed journal. While further research is necessary to validate these results, the potential impact of this AI technology on healthcare is undeniable.
AI technology is proving to be a valuable tool in the field of medical diagnosis. The ability to detect type 2 diabetes through voice analysis offers a convenient and accessible solution, potentially benefiting millions of people worldwide. As researchers continue to refine and expand this technology, we can expect significant advancements in the early detection and management of various health conditions.
