In the Colombian health system, clinical relevance denials are a challenge for healthcare providers, as they involve the withholding of payment for medical services presumed to be unjustified. These denials can lead to substantial financial losses and must be appealed with a thorough review of scientific literature. However, this process is hampered by information overload and the lack of technological tools. Therefore, a computational tool was developed to aid in the search of evidence-based medicine during the management of clinical relevance denials by using deep learning techniques. The development process began with requirements engineering and progressed through the design of user interface and functional logic, and the integration of a web scraping API with a large language model to retrieve articles from Google Scholar. The final product was assessed by a panel of experts, who evaluated its performance, search capability and usability. The tool obtained a score of 90 on the system usability scale. The retrieved articles had an average rating of 4,61 in quality and 4,1 in relevance based on a 5-point Likert scale. Additionally, 90,47% of the articles corresponded to evidence-based medicine. Therefore, the tool has the potential of being a support tool for the appeal process of clinical relevance denials by offering current, aligned and relevant results.