Mapping Health Research Involving Human Subjects in Saudi Arabia: <i>A text mining and topic modeling analysis</i>

Objectives: To identify dominant and underrepresented research topics and to classify published studies by design and data sources in health research involving human subjects from Saudi Arabia. Methods: This retrospective study analyzed PubMed-indexed health articles related to Saudi Arabia from 1959 to March 2025. Text mining was used to identify studies conducted in the country involving human subjects. Eligible articles were classified by study design and data sources. Topic modeling using non-negative matrix factorization was applied to identify topics based on the primary research focus. Results: A total of 36,452 records of studies conducted in Saudi Arabia were initially identified. Of these, approximately 89.8% were original research, 5.1% were case reports or case series, and 5.1% were reviews. Screening of original research articles yielded 18,239 studies involving human subjects, with cross-sectional designs being the most common (57.7%) and clinical trials the least represented (1.4%). Regarding data sources, around 63.4% of studies involving human subjects used surveys, while 9.7% relied on health records. Topic modeling revealed that inpatient care and clinical outcomes constituted the most prominent topic (19.5%), whereas asthma and allergic conditions formed the least represented topic (1.3%). Conclusion: Despite increasing research output in Saudi Arabia, clinical trials remain limited. Promoting interventional research through enhanced infrastructure, collaboration, and training is essential. Topic modeling offers valuable insights to guide research priorities and align national efforts with population health needs.