Advancing Renal Mass Diagnosis and Management through AI-Enhanced Decision Support Systems (RMM-DSS)

Renal masses, both benign and malignant, pose significant diagnostic and management challenges, with kidney cancer ranking among the top 10 most common cancers in both men and women in the U.S. Treatment options vary based on patient and tumor characteristics and include active surveillance, biopsy, surgical resection, and thermal ablation. While surgery can be effective, many small renal masses—particularly those under 4 cm—are benign, and unnecessary surgical removal can expose patients to avoidable risks. Imaging techniques like CT are critical for diagnosis, but manual interpretation is time-consuming and subject to inter-reader variability, contributing to inconsistencies in diagnosis and treatment planning. Integrating imaging with electronic health records (EHRs), which capture key risk factors such as obesity and smoking, can support more accurate risk stratification and clinical decision-making. Despite the promise of artificial intelligence (AI) and real-world data (RWD) to improve renal mass diagnosis and management, adoption in clinical settings remains limited. Barriers include challenges in developing reliable segmentation algorithms, integrating multimodal data, ensuring usability within clinical workflows, and the lack of clear, evidence-based guidelines for managing renal masses— particularly small, incidentally detected lesions—which contributes to clinical uncertainty and variability. To address these gaps, our multidisciplinary team—drawing on expertise in data science and renal research and leveraging large-scale datasets from the UF Health Integrated Data Repository (IDR) and public imaging datasets—proposes the following aims: 1) Develop transformer-based vision-language models (VLMs) for medical image segmentation, clinical concept extraction, and automated radiology report generation, trained on UF and public datasets and validated against expert annotations. 2) Create a robust multimodal framework that integrates EHRs, imaging, and clinical notes—while addressing missing modalities—to support renal mass risk stratification and identification of key clinical factors. 3) Design, develop, and evaluate the RENAL MASS DIAGNOSIS AND MANAGEMENT DECISION SUPPORT SYSTEM (RMM-DSS) using a user-centered design (UCD) approach. This tool will deliver personalized diagnostic and treatment insights and integrate seamlessly into clinical workflows. Iterative co-design and usability testing—including deployment in the Epic sandbox—with radiologists, urologists, and other stakeholders across multiple health systems will ensure clinical relevance and usability. Expected outcomes include: (1) novel AI-driven tools for medical imaging and text processing that enable automated segmentation and report generation; (2) a robust multimodal framework to enhance decision- making; and (3) a high-fidelity, usable prototype of the RMM-DSS. This work has broader potential to improve small renal mass management and inform similar efforts in other clinical domains. Project Number: 1R01CA303940-01A1 | Fiscal Year: 2026 | NIH Institute/Center: National Cancer Institute (NCI) | Principal Investigator: Jie Xu (+1 co-PI) | Institution: UNIVERSITY OF FLORIDA, GAINESVILLE, FL | Award Amount: $632,670 | Activity Code: R01 | Study Section: Clinical Informatics and Digital Health Study Section[CIDH] View on NIH RePORTER: https://reporter.nih.gov/project-details/11367712