Rapid prototyping of a support system for patient selection in clinical trials

Abstract

[Abstract]: Clinical trials are crucial for validating new medical treatments, such as drugs and surgical procedures, by providing empirical evidence of their efficacy. However, these trials are costly, time-consuming, and often encounter challenges during patient recruitment. Automated methods offer advantages by reducing the manual review of medical records, expand- ing the pool of potential participants, and identifying eligible candidates. The TREC Clinical Trial track addresses this by proposing a patient-trial selection paradigm aimed at streamlining recruitment through the matching of patient and trial descriptions. This project evaluates various Information Retrieval and Natural Language Processing approaches to this process, focusing on the effectiveness of different models in improving clinical trial recruitment.

[Resumo]: Os ensaios clínicos son cruciais para validar novos tratamentos médicos, como fármacos e procedementos cirúrxicos, achegando probas empíricas da súa eficacia. Con todo, estes ensaios son custosos, requiren moito tempo e teñen dificultades á hora de recrutar pacientes. Os métodos automatizados ofrecen vantaxes ao reducir a revisión manual de historiais médicos, ampliar o grupo de posibles participantes e identificar os candidatos elixibles. A tarefa Clinical Trial proposta polo TREC aborda esta cuestión propoñendo un paradigma de selección de pacientes e ensaios destinado a axilizar o recrutamento mediante o emparellamento de descricións de pacientes e ensaios. Este proxecto avalía varios enfoques de Recuperación de Información e Procesamento da Linguaxe Natural para este proceso, centrándose na eficacia de diferentes modelos para mellorar o recrutamento de ensaios clínicos.