ľ¹Ï¸£ÀûÓ°ÊÓ

Reconstructive surgery after events like cancer treatment is life-changing, but it comes with the risk of serious complications. Currently, surgeons have limited tools to accurately predict which individual patient is most likely to face issues such as infection, tissue death, or the failure of a reconstruction. Traditional risk calculators often fall short because they cannot fully account for the complex interplay between a patient's unique health profile and the specifics of their surgery.

This doctoral research explores a powerful solution: artificial intelligence (AI). This thesis demonstrates how AI, and specifically a technique called machine learning, can be used to create smarter and more accurate predictive tools for plastic and reconstructive surgery. By training computer algorithms on real-world data from hundreds of patients, this study developed new models to forecast specific complications following breast, abdominal wall, and head and neck reconstruction.

The results show that these AI models consistently outperformed older statistical methods. They more accurately predicted risks such as skin flap necrosis after mastectomy, infection after a breast implant, and hernia recurrence after abdominal wall repair. Crucially, the AI was able to identify complex patterns among numerous patient and surgical risk factors that traditional tools often miss.

The clinical potential is significant. These tools can empower surgeons and patients with personalized risk information, leading to better-informed decisions and targeted strategies to prevent complications. This work provides not only the scientific foundation for these models but also a crucial framework for their safe and ethical implementation into surgical care, paving the way for a more data-driven and personalized future in surgery.