Diagnosing Alzheimer’s disease can be a relatively slow process involving lengthy in-person neuropsychological exams, transcriptions by clinicians, reviews and detailed analysis.
Researchers at Boston University, however, have developed a new tool that could now automate the process, with the eye to eventually move it to online.
Their machine learning-powered computational model can detect cognitive impairment from audio recordings of neuropsychological tests, with no in-person appointment needed.
Ioannis Paschalidis, a co-author on the paper and a professor of engineering, said: “This approach brings us one step closer to early intervention.”
He also believes faster and earlier detection of Alzheimer’s could drive larger clinical trials focusing on patients that are in the early stages of the disease and potentially enable clinical interventions that slow cognitive decline.
“It can form the basis of an online tool that could reach everyone and could increase the number of people who get screened early,” he said.
The research team educated their model using audio recordings of neuropsychological interviews from over 1,000 individuals in the Framingham Heart Study, a long running Boston University project which studies cardiovascular disease and other physiological conditions.
By using automated online speech recognition tools and a machine learning technique known as natural language processing, helps computers understand text.
They had their program transcribe the interviews and then encode them into numbers.
A final model was trained on how to assess the likelihood and severity of a patient’s cognitive impairment using demographic data, the text encodings and real diagnoses from neuropsychologists and neurologists.
Paschalidis declared the model was not only able accurately distinguish between healthy individuals and those with dementia, but it was also able to detect differences between those with mild cognitive impairment and dementia.
It also turned out that the quality of the recordings and how participants spoke was less important than the content of what they were saying.
Ioannis said: “It surprised us that speech flow or other audio features are not that critical; you can automatically transcribe interviews reasonably well, and rely on text analysis through AI to assess cognitive impairment.”
Although the team needs to validate its results still, against other sources of data, their findings suggest that their tool could support clinicians in diagnosing cognitive impairment using audio recordings, including those from telehealth or virtual appointments.
Additionally, the model also gives insight into what parts of the neuropsychological exam could be potentially more important than others in determining whether a patient has impaired cognition.
The researching team’s model splits the exam transcripts into different sections based on the clinical tests performed.
They discovered that the Boston Naming Test, in which clinicians ask patients to label a picture using one word, is most informative for an accurate dementia diagnosis.
“This might enable clinicians to allocate resources in a way that allows them to do more screening, even before symptom onset,” said Paschalidis.
Early diagnosis of dementia is crucial as it gives patients and caregivers more time to be able to create an effective plan for support and treatment.
Furthermore, it is crucial for researchers working on therapies to slow and prevent Alzheimer’s disease progression.
Paschalidis said: “Our models can help clinicians assess patients in terms of their chances of cognitive decline, and then best tailor resources to them by doing further testing on those that have a higher likelihood of dementia.”
