CognoStroke: Automated Cognitive and Mood Assessment on the Hyper-Acute Stroke Unit
Bibliographic record
Abstract
Background: Cognitive and mood impairments are common in Stroke Survivors (SSs), leading to worse outcomes and poorer quality of life measures. Current methods of assessment of mood and cognitive performance are time consuming and rely on health care professionals. This makes assessment in hyper-acute stroke units (HASU) difficult. Here we describe the use of CognoStroke, an automated assessment of mood and cognitive impairment in the HASU. Methods: Using conversational interaction delivered through a virtual, web-based agent (CognosStroke), speech analysis was performed using three large language models (GPT2, Facebook.BART-based, and RobERTa-base) to classify thresholds levels of MoCA (threshold: 22,23,24,25,26), GAD-7 (above 5 and 10), and PHQ-9 (above 5 and 10). Results are presented as Macro F1-scores (MFSs). Patients were asked about barriers to using CogonStroke. Results: A total of 151 SSs agreed to perform CognoStroke, with 75 completing the full assessment. The best MFS of 0.723 was achieved using CognoStroke for thresholding a MoCA of 26. The MFS improved further to 0.783 when single prompts or a smaller combination of prompts from the CognoStroke bank were used. For the PHQ-9 a MFS of 0.686 was achieved thresholding above 10 and on the GAD-7 a MFS of 0.617 was achieved for thresholding above 5. Single prompts or smaller prompt combinations again achieved higher MFSs. Discussion: CognoStroke has potential to classify SSs into groups with high or low cognitive and mood thresholds, highlighting benefits for improving post-stroke cognitive assessment. Challenges of automated assessment on the HASU include patient computer access, anxiety in using technology, post-stroke fatigue, and computer literacy.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".