Predicting the outcome of a Chronic Subdural Hematoma (CSDH) referral using a deep learning algorithm, to determine the admission/acceptance or rejection of a CSDH patient for neurosurgical intervention. We have created a multi-level perceptron artificial neural network that takes in 7 input variables to predict whether a patient’s referral would be accepted by neurosurgery. The model, named Artificial Neural network for Chronic subdural HematOma Referral outcome prediction (ANCHOR), was trained on 1050 patients and tested on an unseen cohort of 450 patients, yielding accuracy of 96.2%, sensitivity of 92.2%, specificity of 98.1%, positive predictive value of 95.6% and an AUC of 0.951. The model is continuously being trained and improved behind the scenes and thus any and all feedback from physicians is appreciated. All the source code is available upon request.

Disclaimer: The model is not externally validated and thus not recommended for clinical use.