Uncovering the Potential for a Weakly Supervised End-to-End Model in Recognising Speech from Patient with Post-Stroke Aphasia DOI Creative Commons

Giulia Sanguedolce,

Patrick A. Naylor, Fatemeh Geranmayeh

et al.

Published: Jan. 1, 2023

Post-stroke speech and language deficits (aphasia) significantly impact patients' quality of life. Many with mild symptoms remain undiagnosed, the majority do not receive intensive doses therapy recommended, due to healthcare costs and/or inadequate services. Automatic Speech Recognition (ASR) may help overcome these difficulties by improving diagnostic rates providing feedback during tailored therapy. However, its performance is often unsatisfactory high variability in errors scarcity training datasets. This study assessed Whisper, a recently released end-to-end model, patients post-stroke aphasia (PWA). We tuned hyperparameters achieve lowest word error rate (WER) on aphasic speech. WER was higher PWA compared age-matched controls (10.3% vs 38.5%, p<0.001). demonstrated that worse related more severe as measured expressive (overt naming, spontaneous production) receptive (written spoken comprehension) assessments. Stroke lesion size did affect Whisper. Linear mixed models accounting for demographic factors, duration, time since stroke, confirmed Whisper left hemispheric frontal lesions.We discuss implications findings how future ASR can be improved PWA.

Language: Английский

Remapping and Reconnecting the Language Network after Stroke DOI Creative Commons
Victoria Tilton-Bolowsky, Melissa D. Stockbridge, Argye E. Hillis

et al.

Brain Sciences, Journal Year: 2024, Volume and Issue: 14(5), P. 419 - 419

Published: April 24, 2024

Here, we review the literature on neurotypical individuals and with post-stroke aphasia showing that right-hemisphere regions homologous to language network other regions, like right cerebellum, are activated in tasks support even healthy people. We propose recovery occurs largely by potentiating hemisphere networks previously supported a lesser degree modulating connection strength between nodes of undamaged left-hemisphere network. Based this premise (supported evidence review), interventions should be aimed at through Hebbian learning or augmenting connections neuroplasticity, such as non-invasive brain stimulation perhaps modulation neurotransmitters involved neuroplasticity. treatment studies have taken approach. conclude further rehabilitation aim is justified.

Language: Английский

Citations

6
Time-Lagged Multidimensional Pattern Connectivity (TL-MDPC): An EEG/MEG pattern transformation based functional connectivity metric DOI Creative Commons
Setareh Rahimi, Rebecca L. Jackson, Seyedeh-Rezvan Farahibozorg

et al.

NeuroImage, Journal Year: 2023, Volume and Issue: 270, P. 119958 - 119958

Published: Feb. 21, 2023

Functional and effective connectivity methods are essential to study the complex information flow in brain networks underlying human cognition. Only recently have begun emerge that make use of full multidimensional contained patterns activation, rather than unidimensional summary measures these patterns. To date, mostly been applied fMRI data, no method allows vertex-to-vertex transformations with temporal specificity EEG/MEG data. Here, we introduce time-lagged pattern (TL-MDPC) as a novel bivariate functional metric for research. TL-MDPC estimates among multiple regions across different latency ranges. It determines how well ROI X at time point tx can linearly predict Y ty. In present study, simulations demonstrate TL-MDPC's increased sensitivity effects compared approach realistic choices number trials signal-to-noise ratios. We TL-MDPC, its counterpart, an existing dataset varying depth semantic processing visually presented words by contrasting decision lexical task. detected significant beginning very early on, showed stronger task modulations approach, suggesting it is capable capturing more information. With only, observed rich between core representation (left right anterior lobes) control (inferior frontal gyrus posterior cortex) areas greater demands. promising identify patterns, typically missed approaches.

Language: Английский

Citations

11
The volume and the distribution of premorbid white matter hyperintensities: Impact on post‐stroke aphasia DOI Creative Commons
Veronika Vadinova, Aleksi J. Sihvonen, Fiona Wee

et al.

Human Brain Mapping, Journal Year: 2024, Volume and Issue: 45(1)

Published: Jan. 1, 2024

Abstract White matter hyperintensities (WMH) are a radiological manifestation of progressive white integrity loss. The total volume and distribution WMH within the corpus callosum have been associated with pathological cognitive ageing processes but not considered in relation to post‐stroke aphasia outcomes. We investigated contribution both WMH, extent lesion load recovery language after first‐ever stroke. Behavioural neuroimaging data from individuals ( N = 37) left‐hemisphere stroke were included at early subacute stage recovery. Spoken comprehension production abilities assessed using word sentence‐level tasks. Neuroimaging was used derive variables (volume critical regions) (WMH three callosal segments). did predict variance measures, when together demographic variables. However, forceps minor segment explained t −2.59, p .01) corrected socio‐demographic Premorbid lesions negatively aphasic This negative impact on is consistent converging evidence suggesting that disrupt neural networks supporting range functions.

Language: Английский

Citations

4
The relationship between activities of daily living and speech impediments based on evidence from statistical and machine learning analyses DOI Creative Commons
Jun Liu, Hong-Guo Li, Y. J. Mao

et al.

Frontiers in Public Health, Journal Year: 2025, Volume and Issue: 13

Published: Feb. 6, 2025

Speech impediments (SIs) are increasingly prevalent among middle-aged and older adults, raising concerns within public health. Early detection of potential SI in this demographic is critical. This study investigates the Activities Daily Living (ADL) as a predictive marker for SI, utilizing data from 2018 China Health Retirement Longitudinal Study (CHARLS), which includes 10,136 individuals aged 45 above. The Barthel Index (BI) was used to assess ADL, correlation between ADL examined through statistical analyses. Machine learning algorithms (Support Vector Machine, Decision Tree, Logistic Regression) were employed validate findings elucidate underlying relationship SI. poses significant challenges health quality life increasing demands on community-based home care services. In context global aging, it crucial investigate factors contributing While role biomarker remains unclear, aims provide new evidence supporting an early predictor analysis machine validation. Data derived CHARLS national baseline survey, comprising participants evaluated using BI, assessed based records "Speech impediments." Statistical analyses, including independent sample t-tests, chi-square tests, Pearson Spearman hierarchical multiple linear regression, conducted SPSS 25.0. algorithms, specifically Support (SVM), Tree (DT), Regression (LR), implemented Python 3.10.2. Analysis characteristics revealed that average BI score "With impediments" group 49.46, significantly lower than 85.11 "Without group. indicated negative (r = -0.205, p < 0.001). Hierarchical regression confirmed robustness across three models (B -0.001, β -0.168, t -16.16, 95% CI -0.001 0.000). validated findings, confirming accuracy with area under curve (AUC) scores SVM-AUC 0.648, DT-AUC 0.931, LR-AUC 0.666. inclusion improved overall performance, highlighting its positive impact prediction. various methodologies demonstrate finding further corroborated by algorithms. Impairment increases likelihood occurrence, underscoring importance maintaining populations mitigate risk

Language: Английский

Citations

0
An innovation scalp acupuncture prescription for post-stroke aphasia：A Neuroimaging-Based validation study DOI Creative Commons

Minjie Xu,

Binlong Zhang,

Yuhang Chen

et al.

Brain Research Bulletin, Journal Year: 2025, Volume and Issue: unknown, P. 111334 - 111334

Published: April 1, 2025

The coexistence of speech disorders in stroke patients can negatively impact their quality life and rehabilitation outcomes. Scalp acupuncture (SA) has shown potential as a non-pharmacological treatment for post-stroke aphasia (PSA). As the location SA PSA is controversial, this study aims to utilize neuroimaging techniques identifying validation promising target. was divided into two phases. In phase Ⅰ, three pipelines, including lesion mapping, meta-analysis, resting-state functional connectivity, were integrated targets. Ⅱ, Centro-square needling manipulations then applied evaluate prescription with PSA. left middle temporal gyrus (MTG) chosen one targets it had highest occurrence among outcomes pipelines. It been discovered that technique MTG immediately enhance reduced connectivity (FC) between frontal caused by diseases. Moreover, enhances FC superior gyrus, which may constitute therapeutic mechanism underlying its efficacy improving verb understanding scores on Chinese Rehabilitation Research Center Standard Aphasia Examination scale. summary, protocol integrating traditional medicine help refine locations

Language: Английский

Citations

0
Latent class cluster analysis to capture language profiles in the acute phase post-stroke DOI
Mara Barberis, Klara Schevenels, Robin Lemmens

et al.

Aphasiology, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 20

Published: April 23, 2025

Language: Английский

Citations

0
The role of language-related functional brain regions and white matter tracts in network plasticity of post-stroke aphasia DOI

Yue Han,

Yuanyuan Jing,

Yanmin Shi

et al.

Journal of Neurology, Journal Year: 2024, Volume and Issue: 271(6), P. 3095 - 3115

Published: April 12, 2024

Language: Английский

Citations

3
The development of the People with Aphasia and Other Layperson Involvement (PAOLI) framework for guiding patient and public involvement (PPI) in aphasia research DOI Creative Commons
Marina Charalambous,

Alexia Kountouri,

Jürg R. Schwyter

et al.

Research Involvement and Engagement, Journal Year: 2023, Volume and Issue: 9(1)

Published: Sept. 1, 2023

Patient and Public Involvement (PPI) in aphasia research requires researchers to include people with as partners from the beginning of study. Yet quality reporting on level type involvement is poorly documented absence a framework guide PPI research. This study aimed extract items statements relevant for development People Aphasia Other Layperson (PAOLI) designing implementing research, collaboration aphasia.The method recommended by EQUATOR network was followed. involved: (1) evidence scoping review, (2) thematic analysis in-depth interviews, stroke aphasia, topics be included pilot draft, (3) two round Delphi survey item/statement selection (4) an experts' consensus meeting. The team involved chronic stroke-induced aphasia. process co-design informed Dialogue model.Twenty-three panellists, 13 countries, voted one 87% (20/23) responding two. final PAOLI includes following 17 (with 66 descriptive statements): establish collaborations, recruit patients, gain consent, organize induction meetings, train patient partners, create communication links, engage conceptualize topics, priorities, reach consensus, work methods, develop proposals, assist dissemination results, promote implementation outcomes, support self-evaluation, monitor progress assess impact involvement. These were considered panellists most partners.The first international guiding Researchers are encouraged adopt improve their promoting meaningful within start.Aphasia disorder which results challenges everyday interactions impacts life. Qualitative involving often investigates Until very recently either excluded such teams or occasionally consultants but without contribution reported team. current builds that has identified standardized approach active teams. participation principles model involves engaging patients/clients about issues. prompted creation framework, close after stroke.To decide content two-round voting (Delphi survey), 23 different meeting finalize completed. statements) how to: important teams.The use supporting

Language: Английский

Citations

9
Individualized functional localization of the language and multiple demand network in chronic post-stroke aphasia DOI Creative Commons
Pieter De Clercq,

Alicia Ronnie Gonsalves,

Robin Gerrits

et al.

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: Jan. 15, 2024

Abstract Recent research found a distinct dissociation between brain regions supporting domain-general cognitive processes and core language functions. The question of whether individuals with post-stroke aphasia (IWA) exhibit comparable remains debated, particularly as previous studies overlooked individual variability in functional network organization heterogeneity. To address this gap, we employed an individualized localization approach to test the involvement multiple demand (MD) during processing chronic aphasia. We collected MRI data 15 IWA 13 age-matched controls. Participants performed spatial working memory task, triggering MD activation, well listening reading activation. compared both groups activation patterns investigated link severity. Involvement was examined by investigating task activity within subject-specific that are active task. each generalized across different modalities, but exhibited robust from other groups. Moreover, there no evidence either group. Additionally, showed weaker controls left-hemispheric regions, higher values left correlating improved performance IWA. In conclusion, our findings suggest does not contribute passive, receptive functions or healthy older adults. Instead, results align proposing normalized supports

Language: Английский

Citations

3
Using in vivo functional and structural connectivity to predict chronic stroke aphasia deficits DOI Creative Commons
Ying Zhao, Christopher R. Cox, Matthew A. Lambon Ralph

et al.

Brain, Journal Year: 2022, Volume and Issue: 146(5), P. 1950 - 1962

Published: Nov. 8, 2022

Abstract Focal brain damage caused by stroke can result in aphasia and advances cognitive neuroscience suggest that impairment may be associated with network-level disorder rather than just circumscribed cortical damage. Several studies have shown meaningful relationships between brain–behaviour using lesions; however, only a handful of incorporated vivo structural functional connectivity. Patients chronic post-stroke were assessed (n = 68) 39) MRI to assess whether predicting performance improved multiple modalities if additional variance explained compared lesion models alone. These neural measurements used construct predict four key language-cognitive factors: (i) phonology; (ii) semantics; (iii) executive function; (iv) fluency. Our results showed each factor (except ability) could significantly related measurement alone; connectivity did not explain above the models. We find evidence predictors linked core sites. First, predictive features found located within resting-state networks identified healthy controls, suggesting might reflect functionally specific reorganization (damage node network disruption entire network). Second, sites, multimodal information redundant prediction modelling. In addition, we observed optimum sparsity regularized regression differed for behavioural component across different imaging features, future should consider optimizing hyperparameters per target. Together, indicate was predicted alone does improve model profile language impairment.

Language: Английский

Citations

14