True Mitotic Count Prediction in Gastrointestinal Stromal Tumors: Bayesian Network Model and PROMETheus (Preoperative Mitosis Estimator Tool) Application Development

• DOI: 10.2196/50023
• Publication Date: 2024-07-21T11:47:13Z
• Authors: Salvatore Lorenzo...; Manuela Cammelli; Ilaria Santori; Marta Tassan-Mangina; Laura Samà; Laura Ruspi; Federico Sicoli; Piergiuseppe Colombo; Luigi Maria Terra...; Vittorio Quagliuolo; Ferdinando Carlo ...
• Keywords: Preprint; Stromal tumor

Background Gastrointestinal stromal tumors (GISTs) present a complex clinical landscape, where precise preoperative risk assessment plays pivotal role in guiding therapeutic decisions. Conventional methods for evaluating mitotic count, such as biopsy-based assessments, encounter challenges stemming from tumor heterogeneity and sampling biases, thereby underscoring the urgent need innovative approaches to enhance prognostic accuracy. Objective The primary objective of this study was develop robust reliable computational tool, PROMETheus (Preoperative Mitosis Estimator Tool), aimed at refining patient stratification through estimation count GISTs. Methods Using advanced Bayesian network methodologies, we constructed directed acyclic graph (DAG) integrating pertinent clinicopathological variables essential accurate prediction on surgical specimen. Key parameters identified incorporated into model encompassed size, location, biopsy specimens, surface area evaluated during biopsy, response therapy, when applicable. Rigorous testing procedures, including prior predictive simulations, validation utilizing synthetic data sets were employed. Finally, trained comprehensive cohort real-world GIST cases (n=80), drawn repository Istituto di Ricovero e Cura Carattere Scientifico (IRCCS) Humanitas Research Hospital, with total 160 analyzed. Results Our exhibited excellent diagnostic performance data. Different architecture selected based lower deviance out-of-sample capabilities. Posterior checks (retrodiction) further corroborated model’s Subsequently, developed. This is an intuitive tool that dynamically computes predicted specimens tumor-specific attributes, area, biopsy-derived using posterior probabilities derived model. Conclusions deployment represents potential advancement GISTs, offering clinicians means anticipate counts solid base stratify patients studies. By facilitating tailored strategies, poised revolutionize decision-making paradigms, ultimately translating improved outcomes enhanced precision management