Great piece Ben. Do you think collection of patient reported outcomes and analysis of the results will shed more light on the question of unindicated surgery? In other words, an insufficient clinically important difference between pre-and postoperative scores could be revealing. Or low preoperative scores could indicate an overly aggressive approach to surgery.
I think there is so much subtlety and nuance when it comes to who's going to do well and who isn't that it's tough to design global appropriateness criteria. There are so many factors that go into a successful outcome from joint and spine surgery, some surgeon dependent and some patient dependent.
I'm actually fascinated by the potential of digital twins and AI predictive models that allow for more granular, individualized recommendations taking into account as many factors as possible. Experience, expertise, and intuition help a lot, but I was surprised so many times of the course of my career. Patients I thought would fly through surgery struggled and some that I was more concerned about did great. The human element beats a lot of predictive models.
I couldn’t agree with you more. Using machine learning and combining as many patient factors as possible to create a predictive model can really help us identify patients who are not likely to benefit from the surgery. We did a project like this at The Baptist and it was published in the NEJM Catalyst. As a spine surgeon, I knew I had to be really careful choosing appropriate patients for surgery because of the complicating issues of chronic pain. I’d like to think I did well, but certainly there were cases that didn’t do well despite my expectation for a good result.
Great piece Ben. Do you think collection of patient reported outcomes and analysis of the results will shed more light on the question of unindicated surgery? In other words, an insufficient clinically important difference between pre-and postoperative scores could be revealing. Or low preoperative scores could indicate an overly aggressive approach to surgery.
I think there is so much subtlety and nuance when it comes to who's going to do well and who isn't that it's tough to design global appropriateness criteria. There are so many factors that go into a successful outcome from joint and spine surgery, some surgeon dependent and some patient dependent.
I'm actually fascinated by the potential of digital twins and AI predictive models that allow for more granular, individualized recommendations taking into account as many factors as possible. Experience, expertise, and intuition help a lot, but I was surprised so many times of the course of my career. Patients I thought would fly through surgery struggled and some that I was more concerned about did great. The human element beats a lot of predictive models.
I couldn’t agree with you more. Using machine learning and combining as many patient factors as possible to create a predictive model can really help us identify patients who are not likely to benefit from the surgery. We did a project like this at The Baptist and it was published in the NEJM Catalyst. As a spine surgeon, I knew I had to be really careful choosing appropriate patients for surgery because of the complicating issues of chronic pain. I’d like to think I did well, but certainly there were cases that didn’t do well despite my expectation for a good result.