New AI system successfully identifies Alzheimer’s disease using speech evaluation

By analyzing speech patterns, researchers at Boston University have developed an AI system that may predict with nearly 80% accuracy whether someone with mild cognitive impairment will develop Alzheimer’s disease inside six years.

The study, published within the journal Alzheimer’s & Dementia, uses AI to extract beneficial diagnostic information from cognitive assessments, accelerating Alzheimer’s diagnosis and, in turn, treatment. 

The team’s AI model achieved an accuracy of 78.5% and a sensitivity of 81.1% in predicting progression from mild cognitive impairment (MCI) to Alzheimer’s disease inside a six-year timeframe. This beats other traditional and non-invasive tests.

Crucially, though, the system relies solely on easily obtainable data: speech transcribed from cognitive assessments and basic demographic information like age, sex, and education level.

Cognitive assessments just like the Boston Naming Test involve a clinician talking to the patient. The audio from these tests is usually recorded for further evaluation. 

“We desired to predict what would occur in the following six years—and we found we are able to reasonably make that prediction with relatively good confidence and accuracy,” said Ioannis (Yannis) Paschalidis, director of the BU Rafik B. Hariri Institute for Computing and Computational Science & Engineering and one in all the study’s lead researchers.

“If you’ll be able to predict what’s going to occur, you’ve got more of a chance and time window to intervene with drugs, and no less than try to take care of the soundness of the condition and stop the transition to more severe types of dementia.”

More in regards to the study

Here’s a breakdown of how the study worked:

1. The research team began by collecting audio recordings of cognitive assessments from 166 participants diagnosed with mild cognitive impairment (MCI). They then tracked these individuals over a six-year period to find out who progressed to Alzheimer’s disease and who remained stable.
2. The team used advanced speech recognition technology to transcribe the audio recordings and prepare the information for evaluation. 
3. Next, the researchers applied sophisticated natural language processing techniques to extract a wide selection of linguistic features and patterns that they believed could potentially function indicators of Alzheimer’s risk.
4. They then used the speech features and demographic information to develop multiple machine learning models.
5. These AI models were designed to predict the likelihood that a given individual would progress from mild cognitive impairment to Alzheimer’s disease based on their unique speech patterns and private characteristics.
6. The models achieved an accuracy of 78.5% and a sensitivity of 81.1% in predicting which participants would develop Alzheimer’s throughout the six-year study period.
7. In a final evaluation, the research team identified cognitive tests with probably the most predictive power for Alzheimer’s risk, comparable to the Boston Naming Test, similarity tests, and the Wechsler Adult Intelligence Scale.

“Digital is the brand new blood,” said Rhoda Au, a professor at BU’s Chobanian & Avedisian School of Medicine and co-author of the study. 

“You can collect it, analyze it for what is understood today, store it, and reanalyze it for whatever latest emerges tomorrow.”

One of probably the most interesting facets of the study found that certain parts of the cognitive assessments were especially predictive of future Alzheimer’s risk. 

“Our evaluation revealed that subtests related to demographic questions, the Boston Naming Test, similarity tests, and the Wechsler Adult Intelligence Scale emerged as the highest features driving the performance of our model,” the researchers note. 

This could inform the event of more targeted cognitive assessments, further streamlining the screening process.

While the outcomes are promising, the researchers admit the necessity for further validation in larger, more diverse populations. 

Speech recognition can open the door to early diagnosis

Speech evaluation has proven a beneficial technique for predicting Alzheimer’s and other diseases.

In a 2020 study much like the Boston University study, University of Sheffield researchers demonstrated their AI’s ability to tell apart between participants with Alzheimer’s disease or mild cognitive impairment and people with functional cognitive disorder or healthy controls with an accuracy of 86.7%. 

Researchers at Klick Labs also developed an AI model that may detect type 2 diabetes using temporary voice recordings of just 6 to 10 seconds. Advanced diabetes can impact the voice through nerve damage, impaired blood flow, and dry mouth, leading to detectable changes. 

The study analyzed 18,000 recordings to discover subtle acoustic differences between diabetic and non-diabetic individuals.

When combined with aspects like age and BMI, the model achieved a maximum test accuracy of 89% for ladies and 86% for men.

Together, these studies prove that AI-supported noninvasive tests and diagnostic methods could lead on to quicker, more practical treatment, even when specialist doctors and equipment are absent.