AI Clinical Decision Tools in Multidisciplinary Team Discussions for Colorectal Cancer

Abstract

Abstract Multidisciplinary team (MDT) discussions have become a cornerstone of colorectal cancer (CRC) management, integrating the expertise of surgeons, oncologists, radiologists, pathologists, and allied health professionals to facilitate personalized, evidence-based care. However, increasing complexity in treatment options, particularly with the rise of neoadjuvant strategies and immunotherapies, has rendered decision-making more challenging. Traditional tools—including clinical guidelines, risk calculators, and nomograms—offer structured decision support but lack flexibility and personalization. Artificial intelligence (AI), particularly through machine learning (ML), radiomics, and large language models (LLMs), is emerging as a transformative adjunct to clinical decision-making in CRC. Machine learning models have demonstrated strong predictive performance for treatment response, recurrence risk, and surgical complications, while radiomics and deep learning have improved diagnostic accuracy and treatment response prediction using imaging and endoscopy. LLMs such as ChatGPT have shown promising concordance with MDT recommendations in early studies, especially for standard clinical scenarios. However, limitations remain in handling complex, nuanced cases. Despite their growing capabilities, AI and LLMs are not yet integrated into routine MDT workflows due to concerns about interpretability, regulatory oversight, and ethical challenges. Future directions include developing real-time, multimodal AI-MDT platforms, improving explainability, ensuring equitable data representation, and integrating AI training into medical education. This review outlines current evidence on AI integration within CRC MDTs, highlighting both its clinical potential and the barriers that must be addressed to ensure safe, effective, and equitable implementation. Ultimately, AI is poised to augment—not replace—human expertise, enhancing the consistency, efficiency, and personalization of multidisciplinary CRC care.

Ähnliche Arbeiten