Joshua A. Waters

Surgical video interpretation is a promising medical artificial intelligence application. However, no existing video annotation method preserves the spatiotemporal complexity of surgeon reasoning. Here we show that verbal reasoning and visual attention can be converted into structured, machine-actionable records of intraoperative behaviours. Our method decomposes transcribed verbal commentary into video-anchored semantic feedback chunks, which are classified via a large language model, with spatial grounding to surgical scenes via eyegaze or cursor tracking. We demonstrate method validity and scalability on structured and unstructured annotation tasks. For quality feedback on full-length colorectal procedures, the method reached near-human fidelity for chunking (cosine similarity: 0.95±0.01) and semantic classification across observations (Cohen’s κ: 0.71±0.07) and evaluative triggers (Cohen’s κ: 0.67±0.14), with excellent usability ratings. For structured critical view of safety assessment in laparoscopic cholecystectomy, implicit annotation yielded excellent agreement with explicit reviewer ratings (Cohen’s κ: 0.83, 0.49 and 0.81 across three criteria). We anticipate this method will advance surgical data science by enabling scalable construction of meaningfully annotated surgical video datasets.

Background: Anal cancer rates are rising among women, immunocompromised individuals, and Men who have sex with Men (MSM), independent of HIV status. While high‐resolution anoscopy remains the diagnostic standard, limited access has increased interest in alternative screening. HPV testing now complements or replaces cytology in cervical cancer prevention, but evidence for anal screening, especially using extended genotyping platforms like BD Onclarity, remains limited. Methods: This single center cross‐sectional study enrolled high‐risk individuals undergoing anal cancer screening from May 2024 to May 2025. Anal cytology and high‐risk HPV (HR‐HPV) DNA testing were performed using the BD Onclarity assay on the BD COR system. Demographics, cytology results, HRHPV genotypes, and available histopathology were analyzed. Results: Among 117 high‐risk participants (median age 42; 106 males, 8 females, 3 transgender women), 74.3% were ‐HIV positive, predominantly MSM‐. Abnormal cytology occurred in 41% of cases including ASC‐US ( n = 29; 60.4%), LSIL ( n = 16; 33.3%), ASC‐H ( n = 2; 4.1%), and HSIL ( n = 1; 2.1%). Valid BD Onclarity results were available for 104 individuals, with 61.5% testing HR‐HPV positive. Non‐16/18 HR‐HPV types were most common overall, while HPV16 predominated in abnormal cytology cases. Twelve individuals underwent biopsy, revealing no dysplasia ( n = 2), AIN1 ( n = 5), AIN2 ( n = 1), and AIN3 ( n = 4). Conclusions: Extended HR‐HPV genotyping using the BD Onclarity assay demonstrated a high prevalence of HR‐HPV in this high‐risk population, with multiple genotype infections commonly observed. While non‐HPV16/18 genotypes predominated overall, HPV16 was more frequently associated with abnormal cytologic and high‐grade histologic findings. Given the limited number of biopsy‐confirmed high‐grade lesions, these associations should be interpreted as descriptive and hypothesis‐generating. Cytology remains central to anal cancer screening, with extended HPV genotyping serving as a complementary adjunct rather than a standalone stratification tool. Larger longitudinal studies with systematic histopathologic follow‐up are needed to clarify the clinical significance of extended HPV genotyping in this population.