A5003904755- Dementia and Cognitive Impairment Research
- Chronic Disease Management Strategies
- Music and Audio Processing
- Natural Language Processing Techniques
- Speech Recognition and Synthesis
- Topic Modeling
- Machine Learning in Healthcare
- Neurobiology of Language and Bilingualism
- Text Readability and Simplification
- Speech and Audio Processing
- Ethics in Clinical Research
- Cell Image Analysis Techniques
- Neural dynamics and brain function
- Phonocardiography and Auscultation Techniques
- Visual perception and processing mechanisms
- Voice and Speech Disorders
- Health, Environment, Cognitive Aging
- Diverse Musicological Studies
- Interpreting and Communication in Healthcare
- Advanced Text Analysis Techniques
- Neuroethics, Human Enhancement, Biomedical Innovations
- Language Development and Disorders
- Health Systems, Economic Evaluations, Quality of Life
- Patient-Provider Communication in Healthcare
- Pharmaceutical Economics and Policy
University of Oxford
2024-2025
Abstract INTRODUCTION We evaluated preliminary feasibility of a digital, culturally‐informed approach to recruit and screen participants for the Alzheimer's Disease Neuroimaging Initiative (ADNI4). METHODS Participants were recruited using digital advertising completed surveys (e.g., demographics, medical exclusion criteria, 12‐item Everyday Cognition Scale [ECog‐12]), Novoic Storyteller speech‐based cognitive test). Completion rates assessment performance compared between underrepresented...
Story recall is a simple and sensitive cognitive test that commonly used to measure changes in episodic memory function early Alzheimer disease (AD). Recent advances digital technology natural language processing methods make this candidate for automated administration scoring. Multiple parallel stimuli are required higher-frequency monitoring.This study aims develop validate remote fully story task, suitable longitudinal assessment, population of older adults with without mild impairment...
We introduce Surfboard, an open-source Python library for extracting audio features with application to the medical domain. Surfboard is written aim of addressing pain points existing libraries and facilitating joint use modern machine learning frameworks. The package can be accessed both programmatically in via its command line interface, allowing it easily integrated within workflows. It builds on state-of-the-art analysis packages offers multiprocessing support processing large workloads....
Abstract Early detection of Alzheimer’s disease is required to identify patients suitable for disease-modifying medications and improve access non-pharmacological preventative interventions. Prior research shows detectable changes in speech dementia its clinical precursors. The current study assesses whether a fully automated speech-based artificial intelligence system can detect cognitive impairment amyloid beta positivity, which characterize early stages disease. Two hundred participants...
We introduce ParaBLEU, a paraphrase representation learning model and evaluation metric for text generation. Unlike previous approaches, ParaBLEU learns to understand paraphrasis using generative conditioning as pretraining objective. correlates more strongly with human judgements than existing metrics, obtaining new state-of-the-art results on the 2017 WMT Metrics Shared Task. show that our is robust data scarcity, exceeding performance only 50% of available training surpassing BLEU, ROUGE...
Introduction Neurodegenerative and psychiatric disorders (NPDs) confer a huge health burden, which is set to increase as populations age. New, remotely delivered diagnostic assessments that can detect early stage NPDs by profiling speech could enable earlier intervention fewer missed diagnoses. The feasibility of collecting data in those with should be established. Methods analysis present study will assess the obtaining data, collected using smartphone app, from individuals across three NPD...
Abstract Background The Alzheimer’s Disease Neuroimaging Initiative (ADNI) is a longitudinal observational study of individuals across the dementia spectrum that validates biomarkers for clinical trials. newest phase study, ADNI4, aims to substantially increase inclusion underrepresented populations improve generalizability ADNI results. Novel biomarkers, including speech‐based cognitive assessment and blood‐based plasma are integrated into design as key participant screening measures....
Abstract Background New disease modifying treatments for Alzheimer’s Disease (AD) require confirmation of amyloid biomarkers and evidence clinical and/or cognitive AD symptoms. Blood tests are poised to enter routine care. However, at‐risk individuals with positive vary in their level impairment, raising the need sensitive scalable assessments. Fully automated speech‐based testing screen Mild Cognitive Impairment (MCI) has shown promise bridge this gap, but its efficacy specifically within...
Abstract Background Scalable, efficient methods are needed to enroll diverse populations of older adults into AD observational studies and clinical trials. We evaluated preliminary feasibility a novel, digital, culturally informed approach recruit screen participants for the Alzheimer’s Disease Neuroimaging Initiative (ADNI4). Methods Digital advertising tailored towards Black/African American Latinx residing near six ADNI sites directed potential recruitment website ( Figure 1 ). They...
Neurons in primary visual cortex (V1) show a remarkable functional specificity their pre- and postsynaptic partners. Recent work has revealed variety of wiring biases describing how the short- long-range connections V1 neurons relate to tuning properties. However, it is less clear whether these connectivity rules are based on some underlying principle cortical organization. Here, we that emerges naturally recurrent neural network optimized predict upcoming sensory inputs for natural stimuli....
Abstract INTRODUCTION Speech‐based testing shows promise for sensitive and scalable objective screening Alzheimer's disease (AD), but research to date offers limited evidence of generalizability. METHODS Data were taken from the AMYPRED (Amyloid Prediction in Early Stage Disease Acoustic Linguistic Patterns Speech) studies ( N = 101, 46 mild cognitive impairment [MCI]) Neuroimaging Initiative 4 (ADNI4) remote digital 426, 58 self‐reported MCI, AD or dementia) in‐clinic 57, 13 MCI) cohorts,...
Abstract The Alzheimer's Disease Neuroimaging Initiative (ADNI) Private Partners Scientific Board (PPSB) Diversity, Equity, and Inclusion Working Group (DE&I WG) was established to work with the ADNI3 Diversity Task Force provide an industry perspective on increasing representation of diverse participants in build precompetitive cross‐industry knowledge engagement recruitment under‐represented (URPs). In this article, we review highlight role ongoing activities within ADNI PPSB DE&I...
Abstract Background Storyteller, a brief self‐administered test that uses speech analysis to measure cognitive functioning, has demonstrated ability predict Mild Cognitive Impairment (MCI) and mild Alzheimer’s disease (AD). The is being implemented globally in Sponsor clinical trials, ADNI, Sites for pre‐screening will be used across heterogeneous populations. Normative data Storyteller exists important contextualising performance, but not been previously published. Method US residents aged...
Abstract Background Changes in speech occur early‐stage Alzheimer’s disease. A range of approaches have been used for eliciting and automatically analysing speech, but limited research has directly compared these methods. Method Participants from the AMYPRED‐UK (NCT04828122) AMYPRED‐US (NCT04928976) studies completed four speech‐elicitation tasks: Automatic Story Recall Task (ASRT), Logical Memory Test (LMT), Semantic (Animals, Vegetables, Fruit) Phonemic Verbal Fluency (F, A, S) Tasks....
We introduce BlaBla, an open-source Python library for extracting linguistic features with proven clinical relevance to neurological and psychiatric diseases across many languages.BlaBla is a unifying framework accelerating simplifying research.The built on state-of-theart NLP frameworks supports multithreaded/GPU-enabled feature extraction via both native calls command line interface.We describe BlaBlas architecture validation of its 12 diseases.We further demonstrate the application BlaBla...
Abstract Background Vocal and linguistic changes in Alzheimer’s dementia have been documented. The current study assesses whether a fully automated speech‐based artificial intelligence (AI) system can detect early clinical impairment amyloid positivity, which characterise the earliest stages of disease (AD). Method Two studies were completed UK USA: AMYPRED‐UK (NCT04828122) AMYPRED‐US (NCT04928976). 200 participants with established beta (Aβ) diagnostic status (97 Aβ+, 103 Aβ‐ from prior PET...
We introduce Surfboard, an open-source Python library for extracting audio features with application to the medical domain. Surfboard is written aim of addressing pain points existing libraries and facilitating joint use modern machine learning frameworks. The package can be accessed both programmatically in via its command line interface, allowing it easily integrated within workflows. It builds on state-of-the-art analysis packages offers multiprocessing support processing large workloads....
Abstract BACKGROUND Story recall is a simple and sensitive cognitive test commonly used to measure changes in episodic memory function early Alzheimer’s disease (AD). Recent advances digital technology natural language processing methods make this candidate for automated administration scoring. Convenient low-burden daily assessments may provide more reliable data than one-off lengthy be suitable longer-term monitoring. OBJECTIVES Develop validate remote fully story task, longitudinal...
Abstract Background Changes in speech, language, and episodic semantic memory are documented Alzheimer’s disease (AD) years before routine diagnosis. Aims Develop an Artificial Intelligence (AI) system detecting amyloid-confirmed prodromal preclinical AD from speech collected remotely via participants’ smartphones. Method A convenience sample of 133 participants with established amyloid beta clinical diagnostic status (66 β +, 67 -; 71 cognitively unimpaired (CU), 62 mild cognitive...
We propose a method for learning de-identified prosody representations from raw audio using contrastive self-supervised signal. Whereas prior work has relied on conditioning models bottlenecks, we introduce set of inductive biases that exploit the natural structure to minimize timbral information and decouple speaker representations. Despite aggressive downsampling input having no access linguistic information, our model performs comparably state-of-the-art speech DAMMP, new benchmark spoken...
Abstract Background Changes in speech occur early‐stage Alzheimer’s disease. A range of approaches have been used for eliciting and automatically analysing speech, but limited research has directly compared these methods. Method Participants from the AMYPRED‐UK (NCT04828122) AMYPRED‐US (NCT04928976) studies completed four speech‐elicitation tasks: Automatic Story Recall Task (ASRT), Logical Memory Test (LMT), Semantic (Animals, Vegetables, Fruit) Phonemic Verbal Fluency (F, A, S) Tasks....