A cross-sectional survey investigating surgeon perceptions of pre-operative risk prediction models incorporating radiomic features.

Purpose: Incisional hernias are a significant source of morbidity in the United States that impact quality of life and can cause life-threatening complications. Complex patient factors, collected as structured and unstructured data, contribute to the risk of developing an incisional hernia following abdominal surgery. It is unknown how risk prediction models derived from imaging data, or radiomic features, can enhance pre-operative surgical planning. This study investigates surgeons' perspectives regarding risk prediction models derived from radiomic features and assesses the model's impact on surgeon behavior.

Methods: An online cross-sectional survey assessing perceptions of a pre-operative risk prediction model was administered to surgeons across the US from April 23, 2024- May 30, 2024. Surgeons' beliefs of the risk model's impact on surgeon behavior and its applicability in the clinical setting were assessed.

Results: A total of 166 completed surveys were analyzed. Mean age was 52.3 (SD 10.1), 71.1% were male, 78.9% were White, and 90.4% were not Hispanic or Latino. The majority of the respondents were general surgeons (58%), colorectal surgeons (14%), thoracic surgeons (12%), and urologists (7%). The mean level of accuracy predicted from radiomic features needed to prompt a change in management was 74.5% (SD 15.1%). The mean at which FPR and FNR were unacceptable was 25.9% (SD 16.9%) and 26.1% (SD 21.7%), respectively. Most believed a risk prediction model tool would improve their peri-operative management.

Conclusion: A majority of surgeons were positively supportive of incorporating a hernia risk-prediction clinical decision tool incorporating radiomic features in their clinical practice.