Learning Language and Acoustic Models for Identifying Alzheimer’s Dementia From Speech
Bibliographic record
Abstract
Alzheimer’s dementia (AD) is a chronic neurodegenerative illness that manifests in a gradual decline of cognitive function. Early identification of AD is essential for managing the ensuing cognitive deficits, which may lead to a better prognostic outcome. Speech data can serve as a window into cognitive functioning and can be used to screen for early signs of AD. This paper describes methods for learning models using speech samples from the DementiaBank database, for identifying which subjects have Alzheimer’s dementia. We consider two machine learning tasks: 1) binary classification to distinguish patients from healthy controls, and 2) regression to estimate each subject’s Mini-Mental State Examination (MMSE) score. To develop models that can use acoustic and/or language features, we explore a variety of dimension reduction techniques, training algorithms, and fusion strategies. Our best performing classification model, using language features with dimension reduction and regularized logistic regression, achieves an accuracy of 85.4% on a held-out test set. On the regression task, a linear regression model trained on a reduced set of language features achieves a root mean square error (RMSE) of 5.62 on the test set. These results demonstrate the promise of using machine learning for detecting cognitive decline from speech in AD patients.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.005 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".