Thanks for your feedback, this will help us improve our content for you!
AAIC 2023 | Voice-recognition based screening for MCI and dementia in older Chinese adults
Xin Xu, PhD, School of Public Health and The Second Affiliated Hospital of School of Medicine, Zhejiang University, Hangzhou, China, shares the findings of a large-scale implementation study of a voice recognition-based digital cognitive screener for mild cognitive impairment (MCI) and dementia in older Chinese adults. The tool was developed in response to the challenges posed by China’s very large elderly population, achieving a wide reach for community-based cognitive screening. By using voice recognition technology, individuals can easily operate the screener despite education or language barriers. Administration efficiency and implementational feasibility was assessed in over 10,000 participants, which demonstrated a completion rate of 97.5% in 5.6-6.1 minutes, regardless of gender, age, or education status. The predictive validity for MCI and dementia was close to 90%. This digital tool offers a more efficient alternative, producing time savings and requiring no specialized staff for administration, while having comparable effectiveness to the pen and paper equivalent. This interview took place at the Alzheimer’s Association International Conference® (AAIC) 2023 in Amsterdam, Netherlands.
These works are owned by Magdalen Medical Publishing (MMP) and are protected by copyright laws and treaties around the world. All rights are reserved.
Transcript (edited for clarity)
So for this digital cognitive screening tool, actually when we were doing the dementia screening in China, you know, the elderly population in China is massive. So in order to do the community based cognitive screening, it’s almost impossible when you need to screen thousands and tens of thousands of elderly in different areas of China. So what we did is that we want to tackle this barrier by introducing a digital based cognitive screening because they can reach a wide audience...
So for this digital cognitive screening tool, actually when we were doing the dementia screening in China, you know, the elderly population in China is massive. So in order to do the community based cognitive screening, it’s almost impossible when you need to screen thousands and tens of thousands of elderly in different areas of China. So what we did is that we want to tackle this barrier by introducing a digital based cognitive screening because they can reach a wide audience. But the most challenging part is that most of the elderly in China, due to the educational barrier and the language barrier, they might not be able to operate the verbal test or phone based cognitive screening tools. So what we did is that we partnered with the DAMO Academy back in China to develop this voice recognition when the elderly actually they don’t need to operate any of the phones on their hands. They just need to interact with the computer or with the app on their hand and just listen to the prompts and provide the answers and the voice recognition tool will just automatically recognize their words and their answers to identify if they are correct or not. By doing that, we will be able to really efficiently apply this screening tool simultaneously in more than ten communities covering more than 10,000 elderlies with acceptance rates of over 90% and we also tested its effectiveness and it is as perfect as the pen and paper version of the cognitive screening tools. So now it has been adopted by the government. It has been used in more than 100,000 of the elderly in China. For the pilot study, which we published last year, it was the development phase where we just conduct the pilot study in only 100+ elderly participants and for this year’s really large scale implementation phase, we not only test the validity and also the efficacy, but most importantly, the feasibility, whether it’s viable as a sustainable solution for the Chinese community to use and we also update our voice bank.
So one of the main language barriers for Chinese elderly is that we use different dialects for different parts of China. So it’s very difficult for just the traditional the voice bank or the analytical platform to identify those dialects. So for the large scale study, we really updated the multimodal understanding model. So we are able to understand different dialects for different elderly from different areas in China.
So for the predictive validity for dementia and MCI detection, it reached out about 90%, 85 to 90% accuracy there. So that is beyond our expectation actually, we have to say we didn’t see that coming. For the time cost and also the labor cost wise, we have said that it’s so much more efficient. Basically we just need six minutes with the elderly without the face to face, one on one administration. So you can use multiple phones administering multiple tests at the same time and also for the labor costs because we no longer need to employ clinical psychologists or those trained research assistants or the personnel to conduct tests. You can actually just have any facilitators or any of the social workers, anyone who can just tap on the screen to help with the screening process. So it’s really efficient. That’s what our large scale implementation study has found. It is now cross validated against different languages and dialects in four provinces in China, including Beijing, Shanghai, one of the Zhejiang province and one of the northern part of China.