Combining Motif information and K-Nearest-Neighbors for Forecasting Hospital Outpatient Visits Flow

Effective hospital outpatient visits flow forecasting is an important task for modern hospitals to implement intelligent management of medical resources. Since outpatient visits flow may be nonlinear and dynamic, we propose a hybrid model, which combines motif information and k-nearest-neighbors regressor. Time series motif is a previously unknown pattern appearing frequently in a time series. In the proposed approach, we first discover time series motif by using a segmentation-based method and then exploit motif information for forecasting in combination with a k-nearest-neighbors (kNN) model. The proposed approach is called kNN+Motif. To demonstrate that our kNN+Motif method is robust, we applied the new approach to forecast the outpatient visits flow in Ho Chi Minh City Hospital of Dermato-Venereology. The experiment was implemented to compare the proposed forecasting model against the single k-nearest-neighbors model and artificial neural network (ANN) model. The experimental results demonstrate that the proposed kNN+Motif model is more effective than the single k-nearest-neighbors method and ANN model. Besides, the kNN+Motif can run much faster than thesingle kNN model.