Automated matching of relevant research studies to patient records for EBM Project Summary There is an urgent need to align the delivery, administration, and evaluation of healthcare with the most up-to-date evidence available from published results of clinical trials. This could improve healthcare outcomes and cost-effectiveness on a large scale. But the processes for bringing evidence to bear are currently limited and are applied unevenly. The standard approach to evidence-based medicine depends upon sustained labor-intensive efforts by physicians to identify and study new evidence relevant to their patients. The alternative is to provide automated software tools, but currently available tools are incomplete and are limited to applying simplified pre-digested guidelines to the available pre-structured patient data, which misses many important aspects expressed in the original clinical record. The proposed project will create a software system that overcomes these limitations by automatically identifying published research studies that contain the evidence most relevant to specific patient clinical records. The proposed software will do this by creating and then matching automatically-constructed summaries of both research studies and clinical records (including processing their full original unstructured text). This project will develop a prototype software system and conduct a rigorous evaluation of its accuracy. If successful, this effort will lead to software products that can improve efficiency and precision of healthcare evaluations and significantly raise physician awareness of relevant evidence affecting patient care.
Automated matching of relevant research studies to patient records for EBM Project Narrative The wealth of evidence in published medical research studies is applied unevenly today. If successful, the proposed project will lead to software products that automatically and accurately identify the medical evidence most relevant for individual patients based on their complete medical records. Such software products would have the potential to improve healthcare outcomes and cost-effectiveness on a large scale.
