Research

Explores just how pernicious shortcutting can be when using CNNs on medical images. Ask deep learning to find a way to predict something strange from a set of images, and far too often, it can, whether it makes sense or not.

A pilot study looking at how ChatGPT answers compare to those of clinicians.

While orthopaedic surgeons intuitively know they can’t tell how much pain a patient feels from X-ray or MRI, it hasn’t been confirmed. This work does that across almost 5,000 patients.

Explores just how pernicious shortcutting can be when using CNNs on medical images. Ask deep learning to find a way to predict something strange from a set of images, and far too often, it can, whether it makes sense or not.

How much of an osteoarthritis patient’s pain is structural? This work moves beyond traditional/human derived biomarkers of knee health and sees how much pain can be predicted by a CNN looking at a knee X-ray.

Patient-reported outcomes (PROs) are used in research, clinical practice, and by federal reimbursement models to assess outcomes for patients who have knee osteoarthritis (OA) and total knee arthroplasty (TKA). But do they reflect observed evaluation we see standardized functional tests?

A look at how tightly population opioid abuse correlates to orthopaedic prescription levels.

Research is now replete with CNN papers. This paper skips the equations and explains how these algorithms “see” in a way a surgeon can understand.