Understanding how Patient-Reported Outcomes (PROs) have been utilized in previous studies is important for Clinical Trial design, but also challenging as it can require reading thousands of articles to identify specific instruments and relevant symptoms. We present an artificial intelligence (AI) algorithm to search within PubMed to generate a set of disease-specific insights and symptoms, tying these to PRO measures where relevant. The goal is to uncover the instruments used in the literature, and the domains those instruments measure. To start, we use AI to surface nearly 125,000 clinical-outcome results from 18,000 papers related to Head and Neck Squamous Cell Carcinoma. Some results were PROs or target, and some were off target, such as survival rates or adverse events. Our algorithm then filtered these clinical outcomes to those that contained generic PRO terms, such as “index” or “scale.” This yielded 2,528 potential outcomes more aligned with PROs. However, these still contain non-PRO measures, such as “tumor index value.” Next, our algorithm identified the most commonly occurring nouns/noun-phrases within the filtered set of 2,528 outcomes, surfacing the more prominent PRO themes. We then picked out the key terms such as “voice” or “quality of life” as separate themes. Some of these themes are specific PRO instruments and some are symptom scales. For instance, the “quality of life” theme includes the “EORTC quality of life questionnaire,” while “voice” includes “voice handicap index.” This approach identified 47 themes, associated with 441 papers. The whole process, from initial data curation to PRO identification, took less than 30 minutes. The algorithm presented here allowed us to understand which patient-centric themes and symptoms are important, and gather how previous trials used PROs. Further, it has the potential to be more thorough and insightful than a standard literature review.