INESC-ID RCAAP Portal
Not a member yet
15883 research outputs found
Sort by
Advancing the prediction and understanding of placebo responses in chronic back pain using large language models
No full textThis study investigates the use of large language models to predict placebo responses in chronic low-back pain. The authors fine-tune these models on contextual features extracted from patient interviews and demonstrate that semantic features accurately predict placebo responders, achieving a classification accuracy of 74% in unseen data. By leveraging natural language processing techniques, this method provides interpretable insights into psychosocial factors underlying placebo responses, highlighting nuanced linguistic patterns linked to responder status
Aligning Web Query Generation with Ranking Objectives via Direct Preference Optimization
No full textAdvancing the Prediction and Understanding of Placebo Responses in Chronic Back Pain Using Large Language Models
No full textBackground Placebo analgesia is a widely studied clinical phenomenon, yet placebo responses vary widely across individuals. Prior research has identified biopsychosocial factors that determine the likelihood of an individual to respond to placebo, yet generalizability and ecological validity in those studies have been limited due to the inability to account for dynamic personal and treatment effects. Methods We assessed fine-tuned large language models (LLMs) for the prediction of placebo responses in chronic low-back pain using contextual features extracted from patient interviews, as they speak about their lifestyle, pain, and treatment history. Interviews were conducted as part of two RCTs designed to study the placebo effect. These interviews were collected after treatment in the first trial (discovery cohort) and prior to treatment in the second trial (validation cohort). Results Semantic features extracted with LLMs can predict which individuals respond to a placebo, with an accuracy of 74% in unseen data, and validating with 70% accuracy in an independent cohort. Furthermore, in contrast to previous work, LLMs eliminated the need for pre-selecting search terms, enabling a fully data-driven approach, and provided interpretable insights into psychosocial factors underlying placebo responses. Conclusions These findings expand on prior research by integrating state-of-art NLP techniques to address limitations in interpretability and context sensitivity of the traditional methods in related work. This method highlights the role of language models to link language and psychological states, paving the way for a deeper quantitative exploration of biopsychosocial phenomena, and to understand how they relate to treatment outcomes. Significance Statement This study paves the way for a deeper yet quantitative exploration of biopsychosocial phenomena through language, and to understand how they relate to treatment outcomes, namely placebo. In this case it highlights nuanced linguistic patterns linked to responder status, which tap into semantic dimensions such as “anxiety,” “resignation,” and “hope”.</p
Exploring Few-Shot Approaches to Automatic Text Complexity Assessment in European Portuguese
No full textThis study explores few-shot approaches to automatic text complexity assessment in European Portuguese, leveraging Large Language Models (LLMs) and prompt-based techniques. The authors investigate the effectiveness of prompts with varying degrees of information and example selection strategies, revealing that even a single example significantly improves model performance and that few-shot approaches outperform fine-tuned models. By utilizing the iRead4Skills dataset, which provides annotated complexity levels for adult native speakers, this work contributes to the development of more accurate and reliable automatic text complexity assessment tools