Classification of Independent Medical Examination Reports Using Supervised Learning Methods

Abstract

An independent medical examination (IME) is requested by an insurance provider or self-insured employer to determine the extent of an injured worker’s disability, including if the injury or ailment is permanent or non-permanent. An IME report is the summary document providing a physician’s medical opinion about a patient based on their experience. Our aim is to classify de-identified IME reports as fitting one of two categories: Permanent and Not Permanent injury. We apply Naïve Bayes (NB) and Support Vector Machine (SVM) classifiers to this task and consider various hyperparameter combinations for each. The machine learning models generated by our work are useful in helping medical professionals identify trends in their work, enabling more equitable and effective treatment and insurance coverage.