The reviewed record of science sign in
Pith

arxiv: 2606.18054 · v1 · pith:KE6HLS2L · submitted 2026-06-16 · eess.AS

AI-based Cognitive-linguistic Features for Dementia Assessment in Picture Description

Lingfeng Xu , Prad Kadambi , Samuel Goldinger , Visar Berisha , Kimberly D. Mueller , Julie Liss This is my paper

Reviewed by Pith2026-06-26 22:46 UTCgrok-4.3pith:KE6HLS2Lopen to challenge →

classification eess.AS
keywords dementia assessmentpicture descriptionlarge language modelscognitive-linguistic featuresADReSS datasetcognitive screeningClaude 3.5 Sonnet
0
0 comments X

The pith

LLMs can rate seven cognitive-linguistic constructs in picture descriptions to distinguish impaired individuals from healthy controls at 85% accuracy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper defines seven constructs specific to the Cookie Theft picture description task and uses large language models to rate them for severity while also producing explanations. Claude 3.5 Sonnet yields the strongest results, generating scores that separate cognitively impaired people from healthy controls. On the ADReSS dataset this reaches 85% accuracy, and expert reviewers rate the model's outputs at an average agreement of 3.99 out of 5. The work shows that LLMs can turn qualitative clinical constructs into quantitative, interpretable measures for cognitive screening.

Core claim

Large language models can be prompted to evaluate seven tailored constructs in Cookie Theft picture descriptions, producing severity scores and example-based explanations that significantly distinguish cognitively impaired individuals from healthy controls, with Claude 3.5 Sonnet reaching 85% accuracy on the ADReSS dataset and 3.99/5 expert agreement.

What carries the argument

Seven author-defined constructs for the Cookie Theft picture description task, rated by LLMs to produce severity scores and explanations.

If this is right

  • LLMs can turn hard-to-quantify clinical constructs into numerical severity scores with supporting explanations.
  • The resulting scores achieve 85% accuracy at separating impaired from healthy participants on the ADReSS dataset.
  • Expert reviewers agree with the model's scores and explanations at an average of 3.99 out of 5.
  • This method supplies a concrete route toward automated yet interpretable cognitive screening tools.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same prompting approach could be tested on other picture-description tasks or languages to check whether the constructs generalize.
  • Combining these text-based scores with acoustic features from speech recordings might improve overall detection performance.
  • Deployment as a lightweight app could allow repeated at-home assessments that track change over time.

Load-bearing premise

The seven constructs validly capture the intended cognitive-linguistic impairments and LLM ratings align with real impairment rather than surface language patterns.

What would settle it

Running the same prompting on an independent picture-description dataset where the severity scores no longer show statistically significant separation between impaired and healthy groups.

Figures

Figures reproduced from arXiv: 2606.18054 by Julie Liss, Kimberly D. Mueller, Lingfeng Xu, Prad Kadambi, Samuel Goldinger, Visar Berisha.

Figure 1
Figure 1. Figure 1: In addition to assigning severity scores, the LLMs [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 1
Figure 1. Figure 1: The prompt used for LLM-based cognitive evaluation using QLoRA [40] by training a 4-bit adapter. The adapta￾tion was performed on the DementiaBank dataset, with out￾put severity scores provided by Claude, our best-performing LLM. Teacher forcing was applied during training. Since the LLM outputs contained substantially more non-numerical to￾kens (corresponding to explanatory text) than numerical tokens (co… view at source ↗
Figure 2
Figure 2. Figure 2: Normalized relative importance of each construct in the XGBoost model’s decision-making process. The model was trained using severity scores derived from (A) manual transcripts and (B) ASR-generated transcripts [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
read the original abstract

Picture descriptions provide valuable insights into several clinical constructs related to cognitive-linguistic abilities. However, operationalizing these constructs into quantitative measures remains challenging, limiting interpretability and clinical utility. We introduced seven constructs tailored to the Cookie Theft picture description task and prompted large language models (LLMs) to evaluate them, generating severity scores and example-based explanations. Among the examined LLMs, Claude 3.5 Sonnet performed the best, producing severity scores that significantly distinguish cognitively impaired individuals from healthy controls. The model achieves a high accuracy of 85% on the ADReSS dataset. Expert evaluation of Claude's scores and explanations yields a 3.99/5 average agreement. The findings demonstrate the potential of LLMs to operationalize clinical constructs and generate interpretable evaluations, offering a promising approach for accessible cognitive screening tools.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript introduces seven cognitive-linguistic constructs tailored to the Cookie Theft picture description task and prompts LLMs to generate severity scores and example-based explanations for each. On the ADReSS dataset, Claude 3.5 Sonnet yields the strongest results, producing scores that distinguish cognitively impaired from healthy controls at 85% accuracy while achieving an average expert agreement of 3.99/5 on the scores and explanations. The work positions this as a route to interpretable, accessible cognitive screening tools.

Significance. If the reported performance and expert alignment hold under fuller methodological scrutiny, the approach could advance automated, explainable assessment of cognitive-linguistic impairment by directly operationalizing clinical constructs rather than relying solely on surface linguistic features. The independent expert rating step supplies an external validity check, and the emphasis on severity scores plus explanations addresses a common limitation in black-box dementia-detection models.

major comments (2)
  1. Abstract: The central performance claim (85% accuracy and 'significantly distinguish') is presented without any reference to the classification procedure, dataset splits, statistical tests for group separation, or inter-rater reliability of the expert ratings. These omissions are load-bearing because the abstract supplies the only quantitative results visible to readers and the soundness assessment cannot be completed from the given text.
  2. Abstract: The seven constructs are asserted to be 'tailored' to the task and to operationalize clinical constructs, yet no information is supplied on their derivation, redundancy checks, or alignment with established instruments (e.g., Boston Diagnostic Aphasia Examination subscales). This directly affects the weakest assumption that LLM ratings track actual impairment rather than surface patterns.
minor comments (2)
  1. Abstract: The phrase 'high accuracy of 85%' would be clearer if accompanied by a parenthetical note on the exact task (binary impairment detection) and the baseline comparison, if any.
  2. The manuscript would benefit from an explicit statement of how the expert agreement score (3.99/5) was computed (e.g., mean across raters and items) and whether the experts were blinded to participant status.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and the recommendation of minor revision. The comments highlight opportunities to strengthen the abstract's self-contained nature, which we will address directly in the revised manuscript.

read point-by-point responses

  1. Referee: Abstract: The central performance claim (85% accuracy and 'significantly distinguish') is presented without any reference to the classification procedure, dataset splits, statistical tests for group separation, or inter-rater reliability of the expert ratings. These omissions are load-bearing because the abstract supplies the only quantitative results visible to readers and the soundness assessment cannot be completed from the given text.

    Authors: We agree that the abstract would benefit from additional methodological context to allow readers to assess the claims independently. In the revision, we will expand the abstract to briefly specify the classification approach (threshold-based binary classification on aggregated severity scores), reference the ADReSS dataset and its standard train/test splits, note the use of non-parametric statistical tests for group differences, and clarify that expert evaluation used a 5-point Likert scale with the reported mean agreement (full inter-rater metrics, if computed, appear in the Methods). This keeps the abstract concise while addressing the concern. revision: yes

  2. Referee: Abstract: The seven constructs are asserted to be 'tailored' to the task and to operationalize clinical constructs, yet no information is supplied on their derivation, redundancy checks, or alignment with established instruments (e.g., Boston Diagnostic Aphasia Examination subscales). This directly affects the weakest assumption that LLM ratings track actual impairment rather than surface patterns.

    Authors: The constructs were derived from clinical literature on cognitive-linguistic deficits in dementia and mapped specifically to the Cookie Theft task's demands (e.g., semantic content, syntactic complexity, pragmatic inference). Redundancy was minimized by ensuring each targets a distinct clinical dimension, with alignment to instruments such as the Boston Diagnostic Aphasia Examination and related scales. While the current abstract is brief, the full manuscript details this grounding in the Introduction. To improve transparency, we will insert a short clause in the revised abstract referencing their clinical basis and will ensure the Methods section explicitly lists the mappings and overlap checks. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper defines seven clinical constructs, prompts LLMs to rate picture descriptions on them, and evaluates the resulting severity scores against external ADReSS dataset labels plus separate expert ratings. No equations, fitted parameters, or self-citations appear in the derivation chain; the reported 85% accuracy and 3.99/5 expert agreement are computed directly against independent ground truth. The pipeline is self-contained with no reduction of outputs to inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the untested premise that the seven constructs are faithful translations of clinical concepts and that LLM outputs track those concepts rather than artifacts of the prompting or model training data.

axioms (1)
  • domain assumption The seven constructs are appropriate and sufficient operationalizations of cognitive-linguistic abilities relevant to dementia in the Cookie Theft task.
    Introduced by the authors without reference to prior validation studies or inter-rater reliability metrics in the abstract.

pith-pipeline@v0.9.1-grok · 5680 in / 1085 out tokens · 26847 ms · 2026-06-26T22:46:23.752626+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

59 extracted references · 8 canonical work pages · 4 internal anchors

  1. [1]

    AI-based Cognitive-linguistic Features for Dementia Assessment in Picture Description

    Introduction Cognitive impairment is a critical early indicator of several neu- rodegenerative diseases [1, 2, 3]. Effective identification en- ables timely diagnosis, accurate risk assessment, and early in- terventions that can slow disease progression and improve pa- tient outcomes [4, 5]. Neuropsychological assessments, many of which include speech-lan...

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  2. [2]

    Data preparation This study used data from the Pitt corpus of DementiaBank

    Methods 2.1. Data preparation This study used data from the Pitt corpus of DementiaBank

  3. [3]

    Control” were assigned to the Control group, with all others assigned to the Clinical group. In the W- ADRC dataset, samples labeled “Normal

    and the Wisconsin Alzheimer’s Disease Research Center (W-ADRC) clinical core [35]. All participants completed the standard Cookie Theft picture description task from the Boston Diagnostic Aphasia Examination [33]. Manual transcripts were preprocessed into clean paragraphs for LLM input. During this step, we removed most annotations originally added by cli...

    2024

  4. [4]

    Saliency of information: …

  5. [5]

    Semantic categories: … …

  6. [6]

    General cognition and perception: … Upon completing the evaluation of all aspects, provide a final summary of the identified issues and the assessed severities. Use this template to prepare your output: ------------------------------------------------ **Reasoning**: *<name of an aspect>*: <reasoning behind your evaluation of this aspect, including quotati...

    2048

  7. [7]

    Saliency of information

    Results 3.1. LLM performance on DementiaBank We begin by comparing the discriminative ability of the sever- ity scores generated by each LLM. As W-ADRC contains pro- tected health information (PHI), this analysis is conducted on DementiaBank, which does not carry PHI-related restrictions. Table 3 summarizes the performance of each LLM when using manually ...

  8. [8]

    Saliency of information

    Discussion This study introduced seven clinical constructs tailored to the Cookie Theft picture description task and prompted several LLMs to assess each construct by assigning severity scores and generating example-based explanations. As presented in Ta- ble 3, Claude 3.5 Sonnet outperforms other LLMs in all met- rics, indicating that advanced LLMs hold ...

  9. [9]

    Mental state language

    Consequently, the group differences in W-ADRC are less pronounced, which may partially account for LLaMA’s reduced performance on the W-ADRC task. Regarding the adaptation strategies, as shown in Table 5, adapting LLaMA using Demen- tiaBank did not improve its performance, suggesting that sim- ply adding data without optimizing the training approach may b...

  10. [10]

    Conclusion and future work In this study, we prompted several LLMs to operationalize seven task-specific clinical constructs and generate example- based explanations. Results highlight the potential of ad- vanced models like Claude 3.5 Sonnet to assess cognitive status from Cookie Theft picture descriptions, with preliminary SLP feedback indicating modera...

  11. [11]

    Acknowledgments This work was supported in part by NIH-NIA Grant 5R01AG082052, 1T32AG082658-01A1, and funding from the John and Tami Marick Family Foundation. We gratefully ac- knowledge Joshua Breger, Phoebe Crumpton, Elena Groves, Madeline Hale, Kristin Murphey, Iris Nowenstein, Ileana Ratiu, and Elizabeth Trueba for their support in evaluating LLM out-...

  12. [12]

    Generative AI Use Disclosure Generative AI tools were used solely for language editing and polishing of the manuscript

  13. [13]

    Mild cognitive impairment represents early-stage Alzheimer disease,

    J. C. Morris, M. Storandt, J. P. Miller, D. W. McKeel, J. L. Price, E. H. Rubin, and L. Berg, “Mild cognitive impairment represents early-stage Alzheimer disease,”Archives of neurology, vol. 58, no. 3, pp. 397–405, 2001

    2001

  14. [14]

    Validity of the MoCA and MMSE in the detection of MCI and dementia in Parkinson disease,

    S. Hoops, S. Nazem, A. Siderowf, J. Duda, S. Xie, M. Stern, and D. Weintraub, “Validity of the MoCA and MMSE in the detection of MCI and dementia in Parkinson disease,”Neurology, vol. 73, no. 21, pp. 1738–1745, 2009

    2009

  15. [15]

    Mild cog- nitive impairment in prediagnosed Huntington disease,

    K. Duff, J. Paulsen, J. Mills, L. Beglinger, D. Moser, M. Smith, D. Langbehn, J. Stout, S. Queller, and D. Harrington, “Mild cog- nitive impairment in prediagnosed Huntington disease,”Neurol- ogy, vol. 75, no. 6, pp. 500–507, 2010

    2010

  16. [16]

    Timely diagnosis for Alzheimer’s disease: A literature review on benefits and challenges,

    B. Dubois, A. Padovani, P. Scheltens, A. Rossi, and G. Dell’Agnello, “Timely diagnosis for Alzheimer’s disease: A literature review on benefits and challenges,”Journal of Alzheimer’s disease, vol. 49, no. 3, pp. 617–631, 2015

    2015

  17. [17]

    Lecanemab in early Alzheimer’s disease,

    C. H. Van Dyck, C. J. Swanson, P. Aisen, R. J. Bateman, C. Chen, M. Gee, M. Kanekiyo, D. Li, L. Reyderman, S. Cohenet al., “Lecanemab in early Alzheimer’s disease,”New England Journal of Medicine, vol. 388, no. 1, pp. 9–21, 2023

    2023

  18. [18]

    Describing the cookie theft picture: Sources of breakdown in Alzheimer’s dementia,

    L. Cummings, “Describing the cookie theft picture: Sources of breakdown in Alzheimer’s dementia,”Pragmatics and Society, vol. 10, no. 2, pp. 153–176, 2019

    2019

  19. [19]

    Connected speech and language in mild cognitive impairment and Alzheimer’s disease: A review of picture description tasks,

    K. D. Mueller, B. Hermann, J. Mecollari, and L. S. Turkstra, “Connected speech and language in mild cognitive impairment and Alzheimer’s disease: A review of picture description tasks,” Journal of clinical and experimental neuropsychology, vol. 40, no. 9, pp. 917–939, 2018

    2018

  20. [20]

    Alzheimer’s dis- ease and automatic speech analysis: A review,

    M. L. B. Pulido, J. B. A. Hern ´andez, M. ´Angel Ferrer Ballester, C. M. T. Gonz´alez, J. Mekyska, and Z. Sm´ekal, “Alzheimer’s dis- ease and automatic speech analysis: A review,”Expert Systems with Applications, vol. 150, p. 113213, 2020

    2020

  21. [21]

    V oice quality and speech fluency distinguish individuals with mild cog- nitive impairment from healthy controls,

    C. Themistocleous, M. Eckerstr ¨om, and D. Kokkinakis, “V oice quality and speech fluency distinguish individuals with mild cog- nitive impairment from healthy controls,”Plos one, vol. 15, no. 7, p. e0236009, 2020

    2020

  22. [22]

    Speech silence character as a diagnostic biomarker of early cog- nitive decline and its functional mechanism: A multicenter cross- sectional cohort study,

    H.-L. Wang, R. Tang, R.-J. Ren, E. B. Dammer, Q.-H. Guo, G.- P. Peng, H.-L. Cui, Y .-M. Zhang, J.-T. Wang, X.-Y . Xieet al., “Speech silence character as a diagnostic biomarker of early cog- nitive decline and its functional mechanism: A multicenter cross- sectional cohort study,”BMC medicine, vol. 20, no. 1, p. 380, 2022

    2022

  23. [23]

    Articu- latory precision from connected speech as a marker of cognitive decline in Alzheimer’s disease risk-enriched cohorts,

    L. Xu, K. Chen, K. D. Mueller, J. Liss, and V . Berisha, “Articu- latory precision from connected speech as a marker of cognitive decline in Alzheimer’s disease risk-enriched cohorts,”Journal of Alzheimer’s Disease, vol. 103, no. 2, pp. 476–486, 2025

    2025

  24. [24]

    High amyloid burden is associated with fewer spe- cific words during spontaneous speech in individuals with subjec- tive cognitive decline,

    S. C. Verfaillie, J. Witteman, R. E. Slot, I. J. Pruis, L. E. Vermaat, N. D. Prins, N. O. Schiller, M. van de Wiel, P. Scheltens, B. N. van Berckelet al., “High amyloid burden is associated with fewer spe- cific words during spontaneous speech in individuals with subjec- tive cognitive decline,”Neuropsychologia, vol. 131, pp. 184–192, 2019

    2019

  25. [25]

    Declines in connected language are associated with very early mild cognitive impairment: Results from the Wis- consin Registry for Alzheimer’s Prevention,

    K. D. Mueller, R. L. Koscik, B. P. Hermann, S. C. Johnson, and L. S. Turkstra, “Declines in connected language are associated with very early mild cognitive impairment: Results from the Wis- consin Registry for Alzheimer’s Prevention,”Frontiers in Aging Neuroscience, vol. 9, p. 437, 2018

    2018

  26. [26]

    WavBERT: Exploiting semantic and non-semantic speech using Wav2vec and BERT for dementia detection,

    Y . Zhu, A. Obyat, X. Liang, J. A. Batsis, and R. M. Roth, “WavBERT: Exploiting semantic and non-semantic speech using Wav2vec and BERT for dementia detection,” inInterspeech 2021, 2021, pp. 3790–3794

    2021

  27. [27]

    Transformer-based deep neural network language models for Alzheimer’s disease risk assessment from targeted speech,

    A. Roshanzamir, H. Aghajan, and M. Soleymani Baghshah, “Transformer-based deep neural network language models for Alzheimer’s disease risk assessment from targeted speech,”BMC Medical Informatics and Decision Making, vol. 21, no. 1, p. 92, 2021

    2021

  28. [28]

    Disflu- encies and fine-tuning pre-trained language models for detection of Alzheimer’s disease,

    J. Yuan, Y . Bian, X. Cai, J. Huang, Z. Ye, and K. Church, “Disflu- encies and fine-tuning pre-trained language models for detection of Alzheimer’s disease,” inInterspeech 2020, 2020, pp. 2162– 2166

    2020

  29. [29]

    Operationalizing clinical speech analytics: Moving from features to measures for real-world clinical impact,

    J. Liss and V . Berisha, “Operationalizing clinical speech analytics: Moving from features to measures for real-world clinical impact,” Journal of Speech, Language, and Hearing Research, vol. 67, no. 11, pp. 4226–4232, 2024

    2024

  30. [30]

    Using linguistic fea- tures longitudinally to predict clinical scores for Alzheimer’s dis- ease and related dementias,

    M. Yancheva, K. C. Fraser, and F. Rudzicz, “Using linguistic fea- tures longitudinally to predict clinical scores for Alzheimer’s dis- ease and related dementias,” inProceedings of SLPAT 2015: 6th workshop on speech and language processing for assistive tech- nologies, 2015, pp. 134–139

    2015

  31. [31]

    A comparative study of acoustic and linguistic fea- tures classification for Alzheimer’s disease detection,

    J. Li, J. Yu, Z. Ye, S. Wong, M. Mak, B. Mak, X. Liu, and H. Meng, “A comparative study of acoustic and linguistic fea- tures classification for Alzheimer’s disease detection,” inICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2021, pp. 6423–6427

    2021

  32. [32]

    Repeatability of commonly used speech and lan- guage features for clinical applications,

    G. M. Stegmann, S. Hahn, J. Liss, J. Shefner, S. B. Rutkove, K. Kawabata, S. Bhandari, K. Shelton, C. J. Duncan, and V . Berisha, “Repeatability of commonly used speech and lan- guage features for clinical applications,”Digital biomarkers, vol. 4, no. 3, pp. 109–122, 2020

    2020

  33. [33]

    GPT-4 Technical Report

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Ale- man, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkatet al., “Gpt-4 technical report,”arXiv preprint arXiv:2303.08774, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  34. [34]

    Claude 3.5 sonnet: Model card,

    Anthropic, “Claude 3.5 sonnet: Model card,” https://www-cdn. anthropic.com/fed9cc193a14b84131812372d8d5857f8f304c52/ Model Card Claude 3 Addendum.pdf, 2024, technical documen- tation provided by Anthropic

    2024

  35. [35]

    LLaMA: Open and Efficient Foundation Language Models

    H. Touvron, T. Lavril, G. Izacard, X. Martinet, M.-A. Lachaux, T. Lacroix, B. Rozi `ere, N. Goyal, E. Hambro, F. Azharet al., “Llama: Open and efficient foundation language models,”arXiv preprint arXiv:2302.13971, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  36. [36]

    Capabilities of GPT-4 on Medical Challenge Problems

    H. Nori, N. King, S. M. McKinney, D. Carignan, and E. Horvitz, “Capabilities of gpt-4 on medical challenge problems,”arXiv preprint arXiv:2303.13375, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  37. [37]

    Can LLMs like GPT-4 out- perform traditional ai tools in dementia diagnosis? Maybe, but not today,

    Z. Wang, R. Li, B. Dong, J. Wang, X. Li, N. Liu, C. Mao, W. Zhang, L. Dong, J. Gaoet al., “Can LLMs like GPT-4 out- perform traditional ai tools in dementia diagnosis? Maybe, but not today,”arXiv preprint arXiv:2306.01499, 2023

    work page arXiv 2023

  38. [38]

    Text dialogue analysis for pri- mary screening of mild cognitive impairment: Development and validation study,

    C. Wang, S. Liu, A. Li, and J. Liu, “Text dialogue analysis for pri- mary screening of mild cognitive impairment: Development and validation study,”Journal of Medical Internet Research, vol. 25, p. e51501, 2023

    2023

  39. [39]

    Alzheimer’s disease recognition from spontaneous speech using large language mod- els,

    J.-U. Bang, S.-H. Han, and B.-O. Kang, “Alzheimer’s disease recognition from spontaneous speech using large language mod- els,”ETRI Journal, vol. 46, no. 1, pp. 96–105, 2024

    2024

  40. [40]

    Macro-descriptors for Alzheimer’s disease detec- tion using large language models,

    C. Botelho, J. Mendonc ¸a, A. Pompili, T. Schultz, A. Abad, and I. Trancoso, “Macro-descriptors for Alzheimer’s disease detec- tion using large language models,” inInterspeech 2024, 2024, pp. 1975–1979

    2024

  41. [41]

    Profiling patient transcript using large language model reasoning augmentation for Alzheimer’s disease detection,

    C.-P. Chen and J.-L. Li, “Profiling patient transcript using large language model reasoning augmentation for Alzheimer’s disease detection,” in2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2024, pp. 1–4

    2024

  42. [42]

    C. A. Nyongesa, M. Hogarth, and J. Pa, “Artificial intelligence- driven natural language processing for identifying linguistic pat- terns in Alzheimer’s disease and mild cognitive impairment: A study of lexical, syntactic, and cohesive features of speech through picture description tasks,”Journal of Alzheimer’s Dis- ease, vol. 106, no. 1, pp. 120–138, 2025

    2025

  43. [43]

    Counting ability of large language models and impact of tokenization,

    X. Zhang, J. Cao, and C. You, “Counting ability of large language models and impact of tokenization,”arXiv preprint arXiv:2410.19730, 2024

    work page arXiv 2024

  44. [44]

    Transformers can do arithmetic with the right embed- dings,

    S. McLeish, A. Bansal, A. Stein, N. Jain, J. Kirchenbauer, B. Bar- toldson, B. Kailkhura, A. Bhatele, J. Geiping, A. Schwarzschild et al., “Transformers can do arithmetic with the right embed- dings,”Advances in Neural Information Processing Systems, vol. 37, pp. 108 012–108 041, 2024

    2024

  45. [45]

    Goodglass, E

    H. Goodglass, E. Kaplan, and S. Weintraub,BDAE: The Boston diagnostic aphasia examination. Lippincott Williams & Wilkins Philadelphia, PA, 2001

    2001

  46. [46]

    The natural history of Alzheimer’s disease: Description of study cohort and accuracy of diagnosis,

    J. T. Becker, F. Boiler, O. L. Lopez, J. Saxton, and K. L. McGo- nigle, “The natural history of Alzheimer’s disease: Description of study cohort and accuracy of diagnosis,”Archives of neurology, vol. 51, no. 6, pp. 585–594, 1994

    1994

  47. [47]

    Wisconsin alzheimer’s disease re- search center-notable discoveries and accomplishments in demen- tia research,

    S. Asthana and R. N. Golden, “Wisconsin alzheimer’s disease re- search center-notable discoveries and accomplishments in demen- tia research,”WMJ: Official Publication of the State Medical So- ciety of Wisconsin, vol. 123, no. 1, pp. 62–63, 2024

    2024

  48. [48]

    The mini-mental state examination: a brief cognitive assessment,

    M. Folstein, S. Folstein, and J. Folstein, “The mini-mental state examination: a brief cognitive assessment,”Principles and prac- tice of geriatric psychiatry, pp. 145–146, 2010

    2010

  49. [49]

    WhisperX: Time- accurate speech transcription of long-form audio,

    M. Bain, J. Huh, T. Han, and A. Zisserman, “WhisperX: Time- accurate speech transcription of long-form audio,” inInterspeech 2023, 2023, pp. 4489–4493

    2023

  50. [50]

    L’examen psychologique dans les cas d’encephalopathie traumatique,

    A. Rey, “L’examen psychologique dans les cas d’encephalopathie traumatique,”Archives de psychologie, vol. 28, pp. 286–340, 1941

    1941

  51. [51]

    Leakage and the Reproducibility Crisis in ML-based Science

    S. Kapoor and A. Narayanan, “Leakage and the reproducibility crisis in ML-based science,”arXiv preprint arXiv:2207.07048, 2022

    work page Pith review arXiv 2022

  52. [52]

    Qlora: Efficient finetuning of quantized llms,

    T. Dettmers, A. Pagnoni, A. Holtzman, and L. Zettlemoyer, “Qlora: Efficient finetuning of quantized llms,”Advances in neu- ral information processing systems, vol. 36, pp. 10 088–10 115, 2023

    2023

  53. [53]

    Bert: Pre- training of deep bidirectional transformers for language under- standing,

    J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre- training of deep bidirectional transformers for language under- standing,” inProceedings of the 2019 conference of the North American chapter of the association for computational linguis- tics: human language technologies, volume 1 (long and short pa- pers), 2019, pp. 4171–4186

    2019

  54. [54]

    Xgboost: A scalable tree boosting sys- tem,

    T. Chen and C. Guestrin, “Xgboost: A scalable tree boosting sys- tem,” inProceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ser. KDD ’16. New York, NY , USA: Association for Computing Machin- ery, 2016, p. 785–794

    2016

  55. [55]

    Alzheimer’s dementia recognition through spontaneous speech: The ADReSS challenge,

    S. Luz, F. Haider, S. de la Fuente, D. Fromm, and B. MacWhin- ney, “Alzheimer’s dementia recognition through spontaneous speech: The ADReSS challenge,” inInterspeech 2020, 2020, pp. 2172–2176

    2020

  56. [56]

    A unified approach to inter- preting model predictions,

    S. M. Lundberg and S.-I. Lee, “A unified approach to inter- preting model predictions,” inProceedings of the 31st Interna- tional Conference on Neural Information Processing Systems, ser. NIPS’17. Red Hook, NY , USA: Curran Associates Inc., 2017, p. 4768–4777

    2017

  57. [57]

    Alzheimer’s disease detection based on large language model prompt engineer- ing,

    T. Zheng, X. Xie, X. Peng, H. Chen, and F. Tian, “Alzheimer’s disease detection based on large language model prompt engineer- ing,” inInternational Conference on Social Robotics. Springer, 2024, pp. 207–216

    2024

  58. [58]

    Exploring linguistic feature and model combination for speech recognition based automatic AD detection,

    Y . Wang, T. Wang, Z. Ye, L. Meng, S. Hu, X. Wu, X. Liu, and H. Meng, “Exploring linguistic feature and model combination for speech recognition based automatic AD detection,” inInterspeech 2022, 2022, pp. 3328–3332

    2022

  59. [59]

    Is chatgpt a good causal reasoner? a comprehensive evaluation,

    J. Gao, X. Ding, B. Qin, and T. Liu, “Is chatgpt a good causal reasoner? a comprehensive evaluation,”arXiv preprint arXiv:2305.07375, 2023

    work page arXiv 2023