Surgeon-Informed Clinical Decision Support Software for Surgical Risk Prediction and Outcomes Tracking

The effective implementation of clinical decision support software (CDSS) requires integration into clinical context and workflow. To prospectively define surgeon-centric design principles, this study employed a modified-Delphi method to guide the development of a novel CDSS app combining incisional hernia risk prediction with patient outcomes tracking.

A three-round, online modified-Delphi panel of 30 surgeons from eight specialties assessed a pilot CDSS app. Feasibility, importance, and acceptability of the application's risk model, user interface, and workflow integration were assessed using Likert scales and qualitative analysis.

Surgeons established clear parameters for model performance (median acceptable false negative rate <10%, false positive rate <15%, minimum area under the receiver operator characteristic curve ≥0.8). Visual risk depictions and transparent weighting of risk factors were highly important for surgeon risk interpretation and patient communication. Team-based data entry, user accounts, data entry and outcome trackers, customizable reminders, and mobile and desktop interfaces were highly important for mitigating user burden, particularly for longitudinal outcomes tracking. Thematic analysis revealed: risk models are primarily used for patients already perceived as high-risk to enhance patient communication, variable statistical literacy impacts surgeon understanding and application of risk model, acceptable model performance metrics vary with clinical context, and minimizing user burden are crucial for successful CDSS adoption.

These findings provide surgeon-informed foundational concepts for developing surgical CDSS tools. For CDSS to be successful, it must function as an interpretable and efficient tool that augments clinical judgment. Addressing end-user needs for model transparency, statistical literacy, and workflow integration are essential for adoption.