Natural language processing techniques for studying language in pathological ageing: A scoping review.
Gloria Gagliardi
PMID 36960885WHAT IT FOUND
Most studies use computer language analysis to tell dementia speech from healthy speech, but methods and reporting are so inconsistent that no tool is ready for clinical use.
Key findings
01The review was based on 179 papers.
02Most studies compare dementia speech with healthy controls rather than tracking progression or subtypes.
03Reported accuracy exceeds 90% for Alzheimer's disease detection and is around 75–80% for mild cognitive impairment, but methods and reporting are too inconsistent for direct comparison.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The review did not test any language analysis tool for patient care. The reviewed studies mostly compare existing groups, not prospective clinical use. Studies used different tasks, recordings, features, and algorithms, so results cannot be compared directly. Only English-language peer-reviewed articles were included, and preprints were excluded. Many datasets are small, and lack of data limits how far findings apply. Some validation methods may inflate reported performance. Age-related voice changes can look like disease-related changes. Only a few studies have been implemented in clinical practice.
Declared interests
The author reported no known competing financial interests.
The easy way to misread this
Do not read the high accuracy figures as evidence that computer language analysis can be used clinically. The review says methods and reporting are not standardized, few studies are implemented in clinical practice, and results are not robust enough to aggregate.