Novel Prehospital Prediction Model of Large Vessel Occlusion Using Artificial Neural Network

Background: Identifying large vessel occlusion (LVO) patients in the prehospital triage stage to avoid unnecessary and costly delays is important but still challenging. We aim to develop an artificial neural network (ANN) algorithm to predict LVO using prehospital accessible data including demographics, National Institutes of Health Stroke Scale (NIHSS) items and vascular risk factors. Methods: Consecutive acute ischemic stroke patients who underwent CT angiography (CTA) or time of flight MR angiography (TOF-MRA) and received reperfusion therapy within 8 hours from symptom onset were included. The diagnosis of LVO was defined as occlusion of the intracranial internal carotid artery, M1 and M2 segments of the middle cerebral artery (MCA) and basilar artery on CTA or TOF-MRA before treatment. Patients with and without LVO were randomly selected at a 1:1 ratio. The ANN model was developed using backpropagation algorithm, and 10-fold cross-validation was used to validate the model. The comparison of diagnostic parameters between the ANN model and previously established prehospital prediction scales were performed. Results: Finally 300 LVO and 300 non-LVO patients were randomly selected for the training and validation of the ANN model. The mean Youden index, sensitivity, specificity and accuracy of the ANN model based on the 10-fold cross-validation analysis were 0.640, 0.807, 0.833, and 0.820, respectively. The area under the curve (AUC), Youden index and accuracy of the ANN model were all higher than other prehospital prediction scales. Conclusions: The ANN can be an effective tool for the recognition of LVO in the prehospital triage stage.