Your patient’s voice holds early Alzheimer’s clues—here’s how to bring the findings into real-world practice
Industry Buzz
I've had patients complain of changes in speech and language-related dynamics long before they develop dementia.
—James Jackson, Psy.D
I would expect the AI to have a high number of false positives, as speech issues more often in elderly patients may be due to strokes which cause problems with speech.
—Clifford Segil, DO
The number of people living with Alzheimer's disease (AD) is expected to reach 14 million by 2060.[] Before the disease sets in, many patients experience mild cognitive impairment (MCI)—and a reported 3-15% of those patients will be diagnosed with AD each year.[]
One way of potentially predicting this progression? Analyzing speech patterns.[]
To test this theory, a Boston University-based research team developed an AI algorithm pre-programmed with voice recordings from over 1,000 individuals, enabling it to identify potential AD risks.
Next, 166 participants (which included 90 progressive MCI and 76 stable MCI cases, ages 63–97 with 107 females) took a neuropsychological test, which included voice recordings, to assess their speech patterns. The researchers also looked at the participants’ age, sex, and education level.
The researchers already knew who among the 166 had developed AD. In 78.5% of cases, they said they could accurately predict the MCI-to-AD progression within six years.
“Voice recordings from neuropsychological exams coupled with basic demographics can lead to strong predictive models of progression to dementia from mild cognitive impairment,” the researchers say. The algorithm now includes the voice recordings from the 166 participants.[]
Related: The results are in: What 2 major new studies show about semaglutide and Alzheimer’sWhat the experts think about this research
Sanam Hafeez, Psy.D, a neuropsychologist and director of Comprehend the Mind says the research helps to confirm the suspicion many experts have had all along: “Speech patterns, as well as behavioral changes, are seen alongside the more commonly known memory loss symptoms,” Dr. Hafeez says. “Further, cognitive testing…taxes the brain more than an informal conversation, which is why the speech patterns are more evident.”
Deepak S. Nair, MD, vice president of Neuroscience Service Line, shares another perspective: “In the setting of [MCI], every year, about 10-15% of patients will progress to a formal diagnosis of Alzheimer's dementia,” he says. “Over a lifetime, that conversion/progression rate is as high as 80%. Predicting this six-year conversion likelihood with improved accuracy will be an important aspect of patient management and counseling. Early recognition allows patients to discuss their goals of care and begin advanced care planning, which are both vital to effective care.”
This kind of work will revolutionize our approach to triage and early diagnosis/prediction of Alzheimer's… I can also see potential applications across various neurodegenerative disorders.
—Deepak S. Nair, MD
In his own clinical practice, Dr. Nair has seen language impairments in patients with cognitive impairment, including reduced vocabulary, word-finding difficulty, and decreased fluency (overall speech output), like more pauses and filler sounds. “These can be very subtle, and easily missed by clinicians unfamiliar with the patient's normal baseline,” he says. Dr. Hafeez also says she notices that patients with cognitive impairment or AD, “start to revert back to earlier language or dialects in those with multiple or bilingual backgrounds."
Translating Research into Practice
Consider these tips for your next patient exam
Listen for subtle speech changes: Experts note that early cognitive decline often appears not through memory lapses alone, but through shifts in speech—reduced vocabulary, word-finding difficulty, increased pauses, or decreased fluency. As Dr. Hafeez notes, some patients may even revert to earlier languages or dialects.
Watch speech during cognitive testing: Dr. Hafeez explains that structured tests—more than casual conversation—tend to reveal early language changes. Use memory, attention, and executive-function tasks as opportunities to observe speech more carefully.
Use speech clues to guide early planning: Dr. Nair emphasizes that spotting higher-risk MCI patients earlier allows for timely discussions about goals of care and future planning—critical steps in effective management.
Remember speech is a meaningful, low-burden marker: Even without AI tools, speech patterns offer valuable insight into early cognitive changes that may otherwise be missed. Your in-person assessment can catch subtle clues that automated tools may eventually support.
Not every healthcare practitioner is a fan of AI, though
Clifford Segil, DO, a neurologist at Providence Saint John’s Health Center in Santa Monica, CA, shares a totally different perspective: “I would never want a family member of mine diagnosed with a neurodegenerative disease from a machine—and this may be the way of the future,” Dr. Segil says.
“I would expect the AI to have a high number of false positives, as speech issues more often in elderly patients may be due to strokes which cause problems with speech. There are so many neurological entities that can affect speech that I find it challenging to believe AI in the year 2024 can tell the difference between aphasia being from a mini-stroke or worsening dementia without the aid of an MRI or in-person neurological exam,” he adds.
Nothing can replace a living, breathing expert, he says. “As a clinical neurologist I find it challenging to believe AI is going to take the place of neurologists and an in-person neurological exam from a neurologist to diagnose someone with something like AD,” Dr. Segil continues.
Others think AI hasn’t even begun to showcase the full scope of its AD-predicting power
James Jackson, Psy.D, director of Behavioral Health and professor of Medicine and Psychiatry at Vanderbilt University, says the research isn’t perfect. Still, its ideas—that speech is the “canary in the coal mine” are intriguing.
Dr. Jackson says he’s seen speech issues in his clinical experience, as well: “I've had patients complain of changes in speech and language-related dynamics long before they develop dementia,” he says. “I think that is the experience of many other psychologists and neurologists, too. The good news here is that if we can, indeed, prove that an individual is on a likely path toward Alzheimer's disease well before it happens, this provides a critical window in which we can intervene.”
As for the use of AI, Dr. Jackson says there are reasons to be cautious, and that diagnosing dementia and AD is complex and clinically intensive. “I think we are barely on the cusp of realizing what AI can do to help inform our work,” he says.
Read Next: Primary care may finally have an Alzheimer’s test of its own