AI outperforms clinicians in post-surgical decisions for colorectal cancer

In a rigorous analysis of 202 T1 colorectal cancer cases, large language models (ChatGPT-4o and DeepSeek) outclassed human clinicians in adherence to treatment guidelines after endoscopic resection. The study revealed a worrying guideline adherence rate below 80% among clinicians, while AI tools significantly improved decision accuracy across varied experience levels and professional backgrounds, regardless of language input. This underlines the transformative potential of AI in enhancing surgical decision-making.