INESC-ID RCAAP Portal
Not a member yet
15883 research outputs found
Sort by
A Reproducibility Study on Consistent LLM Reasoning for Natural Language Inference over Clinical Trials
Speech reference intervals: an assessment of feasibility in depression symptom severity prediction
This study investigates the feasibility of using reference intervals (RIs) derived from speech features to predict depression symptom severity. The authors generated and compared RIs from three reference datasets varying in size, elicitation prompt, and health information, and found that RI-based models performed similarly to conventional feature-based models. By establishing a standardized framework for speech reference intervals, this work aims to increase interpretability and improve clinical applications for depression diagnosis and treatment
Acoustic and Linguistic Biomarkers for Cognitive Impairment Detection from Speech
This research publication presents a comprehensive approach for automatically assessing cognitive decline from spontaneous speech, specifically targeting early detection of dementia and mild cognitive impairment. The authors explore various acoustic and linguistic feature sets, including knowledge-based features, text-based neural embeddings, and macro-descriptors derived from large language models, to identify distinct manifestations of cognitive impairment in speech production. By combining these feature sets with multiple classification strategies and fusing the top-performing systems using logistic regression, the researchers achieve improved final classification performance, demonstrating a promising method for detecting early cognitive decline through speech analysis