From traditional face-to-face medical care, online medical care is also beginning to spread. It appears that medical technology is being developed that will make online medical care even more popular. The technology for detecting diseases based on voices and sounds is called vocal biomarkers. By analyzing sounds, it has become possible to identify various diseases. Research into this vocal biomarker became active around the world in the 2010s. Particularly in the United States, this research is being promoted as a cheap and early detection method for diseases. This is due to the proliferation of smartphones and smart speakers that can record the sounds of conversations and coughs. The medical insurance system in the United States tends to have a heavy burden on medical expenses. In the United States, there is a need for easy and inexpensive medical technology.
The University of Texas Southwestern Medical Center has developed an AI that detects Alzheimer's disease based on voice. We asked 114 people with mild cognitive impairment and 92 people without disabilities to explain what they saw in the paintings. You show them a painting and ask them to explain the content for 1 to 2 minutes, and the AI makes a judgment. First, analyze the smoothness of their speaking style and the complexity of their grammar. Conventional tests take several minutes of time and cost hundreds to thousands of yen to examine vocabulary and other abilities. Unlike conventional tests, AI is an excellent tool that can detect signs of disease in about 10 minutes. AI detects mild cognitive impairment with an accuracy of approximately 80%, which is slightly higher than the 70% accuracy of conventional simple tests.
Colds are said to be the cause of all diseases, but now diabetes may be the cause of all diseases. According to the International Diabetes Federation, as of 2020, there will be approximately 540 million adult patients with diabetes worldwide. Approximately half of these 540 million people do not receive medical treatment. When blood sugar levels remain high, nerves and muscles are damaged, which can affect your voice. Diabetes is prone to depression and cognitive decline, and these symptoms may have been reflected in the voice. Focusing on this, a technology was developed to detect diabetes using voice. It has become possible to identify male and female patients with type 2 diabetes with an accuracy of 86-89% based on the characteristics of speech captured on a smartphone. In Japan, blood tests and medical examinations for diabetes cost around 10,000 yen. This audio tool is an affordable inspection tool that can be used anytime, anywhere, and is reasonably priced. Advances in vocal biomarkers may become tools that contribute to health promotion.