CoChat

Diagnosing People with Dementia Using Automatic Conversation Analysis

The DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index; none of the 10 sources its claims cite is checked for support yet.

Abstract

A recent study using Conversation Analysis (CA) has demonstrated that communication problems may be picked up during conversations between patients and neurologists, and that this can be used to differentiate between patients with (progressive neurodegenerative dementia) ND and those with (nonprogressive) functional memory disorders (FMD). This paper presents a novel automatic method for transcribing such conversations and extracting CA-style features. A range of acoustic, syntactic, semantic and visual features were automatically extracted and used to train a set of classifiers. In a proof-of-principle style study, using data recording during real neurologist-patient consultations, we demonstrate that automatically extracting CA-style features gives a classification accuracy of 95%when using verbatim transcripts. Replacing those transcripts with automatic speech recognition transcripts, we obtain a classification accuracy of 79% which improves to 90% when feature selection is applied. This is a first and encouraging step towards replacing inaccurate, potentially stressful cognitive tests with a test based on monitoring conversation capabilities that could be conducted in e.g. the privacy of the patient’s own home. \n \n

Overview

CoChat Summary

TL;DR

The paper presents an automatic conversation-analysis pipeline to distinguish neurodegenerative dementia (ND) from functional memory disorder (FMD) in neurologist-patient consultations. On 39 conversations, it reports 95% accuracy with verbatim transcripts, 79% with ASR transcripts, and 90% with ASR plus feature selection.

Key findings

  • Using verbatim transcripts and automatic CA-style features, the best classifier (linear SVM) achieved 95% cross-validation accuracy; across the five classifiers, accuracies ranged from 82% to 95% with an average of 87%.
  • With a reduced 20 audio-only feature set, the automatic system (“AutoCA”) reached 92% accuracy; adding ASR transcripts reduced accuracy to 79%.
  • Applying feature selection (top 10 features by RFE) improved the ASR-based system from 79% to 90%, while AutoCA remained at 95%.
  • The ASR models were trained/tested on the difficult conversational data with word error rates of 95.1%, 81.1%, and 40.6% for the three ASR configurations reported.

Methods

  • Design: Proof-of-principle study replicating CA-inspired diagnostic profiling from Elsey et al., comparing manual transcripts, automatic feature extraction, ASR-based transcripts, and feature selection.
  • Data/sample: 39 conversations total: 21 FMD and 18 ND; the dataset included around 10 hours of spontaneous speech from 105 speakers, with 7820 utterances after removing very short and overlapping segments.
  • Analysis: Extracted 22 CA-style features initially, then 20 audio-only features for the fully automatic setting; evaluated five classifiers (linear SVM, Random Forest, AdaBoost, Perceptron, SGD) using leave-one-out cross-validation over 39 test datapoints, with results reported mainly for the linear SVM.

Limitations

  • The paper states that the data were challenging for speech recognition, with high overlap, noisy recording conditions, and sometimes poor microphone placement; it therefore excluded overlapping speech and did not attempt a fully general recognition setup.
  • The authors note this is an initial/proof-of-principle study and that feature understanding is shallow, with no higher-level language understanding attempted.
  • Evident from the text: the sample is small and drawn from a specific clinical interaction setting, so the reported accuracies may not generalize beyond these 39 consultations.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.21437/interspeech.2016-857

Cites

18 of 18 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 Valid · 0 Not valid · 6 Not checked · 25 Own results and interpretations

9 more claims

Create a free account to see all 31 claims and to verify the 10 references still open.

See all claims — free account
  • It may state two separate claims as one. 11
  • Probably not a substantial claim. 7

Create a free account to see these claims.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 18 references

Research works printed but not linked: 2

  1. [1][1] S. Bel, K. Harkness, J. M. Dickson, and D. Blackburn, 'A diagnosis for 55: what is the cost of government initiatives in dementia case finding.' Age and Ageing. , vol. 44, pp. 344-345, 2015. -
  2. [17][17] H. E. Hamilton, Conversations with an Alzheimers patient: An interactional sociolinguistic study . Cambridge: Cambridge University Press, 1994. -

Cited by

Metadata

Authors
Mirheidari, B., Blackburn, D., Reuber, M., Walker, T., & Christensen, H.
Affiliations
  • University of Sheffield (GB)
  • Royal Hallamshire Hospital (GB)
Venue
Interspeech 2016
Publisher
ISCA
Published
2016-09-08
Pages
1220-1224
ISSN
2308-457X (ISSN-L)
OpenAlex
W2518348742
Cited by
25
References
18
Citations by year
2026: 1 · 2025: 1 · 2024: 1 · 2023: 1 · 2021: 4 · 2020: 4 · 2019: 8 · 2018: 4 · 2017: 1
FWCI
14.80
Type
conference paper
Language
en
Open access
closed
Locations
  • Interspeech (USB) (publishedVersion)
  • White Rose Research Online (University of Leeds, The University of Sheffield, University of York) (acceptedVersion)
Topics
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
  • Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
  • Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
Keywords
dementia diagnosis, conversation analysis, automatic transcription, neurodegenerative dementia, feature selection

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-10-11, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-11 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.