CLEVELAND -- Implementation of AI may allow for better personalization of care, identification of surgical risks and delivery of safe surgery, according to a presenter at Cleveland Clinic's A.I. Summit for Healthcare Professionals.

"We are at an exciting and critical junction where we need to know about [AI in medicine] and be active in the decisions," Christopher Weight, MD, , MS, professor and director of urologic oncology and vice chair of the department of urology - research at Cleveland Clinic Lerner College of Medicine at Case Western, said in his presentation.

While reliable strategies exist to identify patients who will do well after surgery and those who are at high risk for complications, Weight said collecting data on patients is costly and has a 5% to 10% error rate. However, a large language model has been shown to reliably extract and analyze data from a large group of patients to identify who may be at high risk.

"We can use large language models and prompt engineering to allow high-volume, detailed data to be pulled out in a structured way," Weight said. "This gives us the tools to start using other AI tools to be better at personalizing care."

AI models can also predict whether a patient will have a good outcome or not after surgery, Weight said. By training a model to predict age based on CT scan, Weight and colleagues found that not only did the model accurately predict a patient's true age, but any discrepancy in age determined whether the patient's hospital length of stay would be impacted.

"If the model thought you were younger than you were, you got out of the hospital faster," Weight said. "We also wanted to look at overall survival. We found a similar pattern that AI age discrepancy was more important than the true age of the patient in predicting overall survival."

Autonomous robotic surgery is another area that has received attention in recent years. A survey published in 2020 showed 38% of participants believed surgery performed by a fully autonomous robot was already happening, despite there being no FDA approvals for fully autonomous surgical instruments or robots, according to Weight.

"Level 1 is where most of our surgical robots are that you hear about today," Weight said. "These [current surgical robots] scale and translate surgical hand movements into instruments inside the body. They have no decision-making capability."

He said the highest level of autonomy that has been FDA approved is level 3, which means the robot can understand the surgical scenario, plan and execute specific tasks.

"So far we have been nowhere near a level 4 or level 5 autonomous robotic surgery," Weight said.