by A recent study conducted by scientists from biotech firm Klick Labs has found that a simple 10-second smartphone voice recording may be able to detect type 2 diabetes. Currently, individuals who want to test for diabetes have to visit a clinic for blood tests and wait for results. However, this new voice test could potentially deliver immediate results on-the-spot.
The study involved 267 participants who had already been diagnosed as non-diabetic or type 2 diabetic. Each participant was asked to record a specific spoken phrase six times a day for two weeks using a smartphone app. The resulting recordings were between six to 10 seconds long, depending on the individual’s speaking speed.
The researchers analyzed 18,465 recordings, focusing on 14 acoustic features such as pitch and intensity. They found that these features varied consistently between the diabetic and non-diabetic participants, although they were undetectable to the human ear. However, signal processing software was able to pick up on these differences.
This suggests that developing type 2 diabetes causes subtle changes in a person’s voice. Taking this theory into account, the scientists developed an artificial intelligence-based program that analyzes voice recordings along with patient information such as age, sex, height, and weight. When tested on volunteers, the program accurately identified type 2 diabetic women with an 89% success rate and diabetic men with an 86% success rate.
While these numbers still have room for improvement, they already outperform traditional fasting blood glucose tests, which were 85% accurate for both sexes. Other tests like glycated hemoglobin and oral glucose tolerance tests had higher accuracy rates of 91% and 92%, respectively.
The next step for the researchers is to conduct voice tests on a larger, more diverse population to further refine the technology. According to Jaycee Kaufman, the first author of the study, this new voice technology has the potential to remove the barriers of time, travel, and cost associated with current methods of diabetes detection.
*Note:
1. Source: Coherent Market Insights, Public sources, Desk research
2. We have leveraged AI tools to mine information and compile it
