Automated text‐level semantic markers of Alzheimer's disease

Abstract

Introduction Automated speech analysis has emerged as a scalable, cost‐effective tool to identify persons with Alzheimer's disease dementia (ADD). Yet, most research is undermined by low interpretability and specificity. Methods Combining statistical and machine learning analyses of natural speech data, we aimed to discriminate ADD patients from healthy controls (HCs) based on automated measures of domains typically affected in ADD: semantic granularity (coarseness of concepts) and ongoing semantic variability (conceptual closeness of successive words). To test for specificity, we replicated the analyses on Parkinson's disease (PD) patients. Results Relative to controls, ADD (but not PD) patients exhibited significant differences in both measures. Also, these features robustly discriminated between ADD patients and HC, while yielding near‐chance classification between PD patients and HCs. Discussion Automated discourse‐level semantic analyses can reveal objective, interpretable, and specific markers of ADD, bridging well‐established neuropsychological targets with digital assessment tools.

Detecting Parkinson’s disease and its cognitive phenotypes via automated semantic analyses of action stories

“…This reinforces the view that semantic abnormalities in PD are mainly driven by action concepts. Indeed, while PD patients are consistently affected in this category 1 , they evince no major alterations in more general semantic measures, including processing of abstract 12 and social concepts 9 , semantic granularity 29 , and ongoing semantic variability 29 , among others. Note, also, that the P-RSF metric allows identifying specific semantic memory domains that are compromised and spared, favoring interpretability.…”

“…First, our sample size was moderate, especially in the subgroup analyses. Although previous natural discourse studies on PD 8 , 20 , 35 and other neurodegenerative disorders 29 have yielded robust results with similar and smaller groups, replications with more participants would be needed. Relatedly, results stemmed from the distance between the original texts’ verbs and the ones produced by participants in each training fold, meaning that they might change if new participants were tested and produced verbs that were not present in such folds.…”

“…Following standard preprocessing approaches, all words in each transcript were converted to lowercase letters. Accents, numbers, punctuation signs, and stop words were then removed, and the remaining words were lemmatized 29 . Feature extraction was performed for AT retellings and nAT retellings separately.…”

Natural language signatures of psilocybin microdosing

“…As a first step, this approach consisted in extracting linguistic features from unconstrained speech produced by the participants during the acute effects, using them as input to machine learning algorithms. NLP is characterized for being an objective, non-invasive, cost-effective, and scalable tool to investigate ecologically valid data (Sanz, 2022; Tagliazucchi, 2022). Unlike standard questionaries, which constrain reports to a possibly sub-optimal pre-selected set of questions, this approach is capable of automatically identifying and capturing informative semantic and grammatical features of speech, allowing to distinguish between different experimental conditions (Agurto et al, 2020; Bedi et al, 2014; Corcoran et al, 2018; Norel et al, 2020; Sanz, 2022; Sanz et al, 2021).…”

“…NLP is characterized for being an objective, non-invasive, cost-effective, and scalable tool to investigate ecologically valid data (Sanz, 2022; Tagliazucchi, 2022). Unlike standard questionaries, which constrain reports to a possibly sub-optimal pre-selected set of questions, this approach is capable of automatically identifying and capturing informative semantic and grammatical features of speech, allowing to distinguish between different experimental conditions (Agurto et al, 2020; Bedi et al, 2014; Corcoran et al, 2018; Norel et al, 2020; Sanz, 2022; Sanz et al, 2021). Importantly, besides informing the contents of the drug-elicited experience, NLP allows to investigate the modulation of language production itself, which can be informative of drug action beyond what is reported during the subjective acute effects (Tagliazucchi, 2022).…”