Orthopedic data science: fluoroscopic image analysis for the objective assessment of technical skill

Most adverse events occurring in clinical settings are sown long before they occur. Some systematic failures and their effects go undetected for years, perceived only after the occurrence of some triggering event, such as an adverse surgical outcome. Because of their dormancy, these latent errors are most easily prevented when treated at their source. A motivating example of this is a resident surgeon advancing out of a residency program without obtaining sufficient proficiency in a vital technical skill. This is a lapse in skill assessment methodology. Over the last two decades, residency programs have been switching to more competency-based graduation criteria and begun using data science to advance skills assessment methods toward greater objectivity. The orthopedic surgical discipline is a leader in these efforts.

Fluoroscopic wire navigation is a common orthopedic skill involving a surgeon using a moveable x-ray machine to track the progress of their tools as they navigate through bone to place an implant. The advantage of this skill is that it is less invasive and improves patient care. However, mastering this skill can be challenging, and resident surgeons require frequent assessments and personalized feedback from their mentor to better understand how to improve. Traditional assessment methods have limited efficacy due to attributes like bias. This work presents an alternative, data science focused approach to evaluating fluoroscopic wire navigation without those limitations. The method uses the x-rays taken during surgery to produce unbiased and informative assessments. From these assessments, residents and their mentors can inexpensively produce reliable, accurate, and precise measures of the residents’ competency. Ultimately, this these assessments may improve patient safety through better prepared residents.