Hospital Services Quality Prediction Using A New Deep Fuzzy Learning Model

Abstract

Hospital services play a crucial role in people's lives, and continuous improvement in these services is essential. The prediction is paramount to determine the quality and suitability of the hospital's services. Yet the complexity of predicting increases, especially when dealing with intangible services such as hospital services. Many traditional prediction methods suffer from a gradual decline in accuracy and a simultaneous increase in the error rate especially in time series data. This paper introduces a new model, called Deep Fuzzy Learning Prediction (DFLP), for accurately predicting hospital service quality. The model combines fuzzy logic and deep learning neural network techniques and utilizes a set of quality parameters. DFLP's nodes represent different memberships among fuzzy sets, and the model adapts its internal fuzzy membership functions based on input data and time. Each node within the DFLP model box represents a different membership among the fuzzy sets. The membership scenarios vary when adding or removing a group of hospitals from the prediction process. This is due to changes in the fuzzy membership centers, resulting in membership changes. The DFLP model offers an effective way to predict and adapt to the dynamic nature of hospital services across time series. The model is tested on data from more than 4500 hospitals in the United States collected by the Centers for Medicare and Medicaid Services (CMS). The results show a significant convergence between the model's prediction and the actual services provided by hospitals. The results were compared with Long Short-Term Memory (LSTM) networks and Bidirectional LSTM (Bi-LSTM) networks and showed the superiority of the proposed model over traditional prediction networks. The proposed model achieved an accuracy of 96.68%, while LSTM and Bi-LSTM obtained 92.30% and 94.45%, respectively