How artificial intelligence (AI) is used in aphasia rehabilitation: A scoping review
Bibliographic record
Abstract
Background In recent years, artificial intelligence (AI) has become commonplace in our daily lives, making its way into many different settings, including health and rehabilitation. While there is an increase in research on AI use in different sectors, information is sparse regarding whether and how AI is used in aphasia rehabilitation.Aims The objective of this scoping review was to describe and understand how AI is currently being used in the rehabilitation of people with aphasia (PWA). Our secondary goal was to determine if and how AI is being integrated into Augmentative and alternative communication (AAC) devices or applications for aphasia rehabilitation.Methods Using the Arksey and O’Malley (2005) Levac and colleagues (2010) frameworks, we identified the research question: In what way is artificial intelligence (AI) used in language rehabilitation for people with aphasia (PWA)? We then selected search terms and searched six databases which resulted in the identification of 663 studies. Based on the inclusion criteria, 28 suitable studies were retained. We then charted, collated and summarised the data in order to generate four main themes: (1) AI used for the classification or diagnosis of aphasia/aphasic syndromes or for the classification or diagnosis of primary progressive aphasia (PPA)/PPA variants; (2) AI used for aphasia therapy; (3) AI used to create models of lexicalization; and (4) AI used to classify paraphasic errors.Results None of the articles retained incorporated AI in AAC devices or applications in the context of aphasia rehabilitation. The majority of articles (n=17) used AI to classify aphasic syndromes or to differentiate PWA from healthy controls or persons with dementia. Another subset of articles (n=7) used AI in the attempt to augment an aphasia therapy intervention. Finally, two articles used AI to create a model of lexicalization and another two used AI to classify different types of paraphasias in the utterances of PWA.Conclusion Regarding performance accuracy of the diagnosis tools, results show that, regardless the type of AI approach used, models were able to differentiate between aphasic syndromes with a relatively high level of accuracy. Although significant advancements in AI and more interaction between the fields of aphasia rehabilitation and AI are required before AI can be integrated in aphasia rehabilitation, it nevertheless has the potential to be a central component of novel AAC devices or applications and be incorporated into innovative methods for aphasia assessment and therapy. However, for a transition to the clinic, new technologies or interventions using AI will need to be assessed to determine their efficacy and acceptance by both speech-language pathologists and PWA.
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 | 0.011 | 0.067 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.023 | 0.023 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.006 | 0.005 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.004 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 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".