Dementia is a major health issue in America. Everyone likely knows of a family member, a neighbor or a friend who has or had dementia. It is likely to become even more of a concern as people live longer and the population ages. Diagnosis at the early stages can be challenging, and we need new approaches for early intervention. Especially with a disease as devastating as this, quality of life for patients is important. Biomedical technology to the rescue! Researchers have developed a machine learning model that can identify mild cognitive impairment in voice recordings.
For this study, the scientists evaluated voice recordings from neuropsychological tests from more than 1,000 people. The results were remarkable, with 88% accuracy in distinguishing normal cognition from mild dementia and 92% accuracy in identifying normal cognition from more advanced dementia. That is impressive and could represent a real breakthrough.
The approach involved using natural language processing to transcribe the audio recordings to written text. This technology is similar to the voice-to-text capabilities your smartphone has, but the system the scientists used was more sophisticated. The scientists then trained an AI process to identify signs of cognitive impairment, and processed the text through it. The results showed that one particular portion of the neuropsychological test was useful in accurately diagnosing dementia. That section of the test was an exercise where patients were asked to describe a picture with one word. Scientists want to further improve the tests to bring the numbers up even higher.
A previous study in Japan used speaking rate and pause intervals (pauses to find the right word) as a marker for Alzheimer's. That study resulted in a 90% accuracy in identification of Alzheimer's patients based on telephone audio files from more than 100 patients. This new study focused not on the way patients were speaking, but what they said.eople without fear of infection. Those who continue to shed virus beyond this time are at risk for more severe issues.
Scientists also thought that following voice patterns of an individual over time would help in identifying these subtle changes in cognitive function. This development of a diagnostic biomarker based on analysis of a person's voice could improve early diagnosis and help people who may develop this devastating disease.
So how far away are we from assessing cognitive decline by telehealth or a virtual visit? One day, you could have a conversation with your doctor or a nurse, read some text or respond to a few questions. Your voice is analyzed and a score is generated for your medical file documenting your cognitive decline or not! Of course, if there are concerns, there would likely be a follow-up with your physician.
This is becoming a major health issue in the U.S. as 6.5 million people have been diagnosed with Alzheimer's. Another estimated 5 million are living with mild cognitive impairment. These numbers will continue to rise as our population ages. Early diagnosis and intervention can slow the disease to improve people's quality of life.
More Information
Detection of dementia on voice recordings using deep learning: a Framingham Heart Study
Identification of reliable, affordable, and easy-to-use strategies for detection of dementia is sorely needed. Digital technologies, such as individual voice recordings, offer an attractive modality to assess cognition but methods that could automatically analyze such data are not readily available...
AI Detects Dementia in Voice Recordings
There are over 10 million new cases of dementia each year worldwide, according to the World Health Organization. Researchers conduct neuropsychological tests to assess patients' cognitive functioning, but transcribing the audio recordings takes a lot of time. Artificial intelligence (AI) algorithms can be used to analyze voice recordings of individuals to detect their dementia status efficiently and accurately...