Text Classification for Patient Experience Improvement: A Neural Network Approach

Patients' experience reflects the quality of care offered by the hospital. Recognizing its importance will help achieve better healthcare service delivery and increase the likelihood of receiving rewards rather than penalties as an enhancement incentive. Centers for Medicare & Medicaid Services (CMS) has elaborated a variety of patient experience surveys. This research utilizes the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, administered to inpatients, at Cooper University Hospital from 2018 to 2022. Alongside the multiple-choice questions, the survey includes a free-text section that reflects patients' feedback and from which the comments’ dataset is extracted. Text preprocessing and vectorization are conducted in parallel with keywords extraction based on which main topics are identified. This paper focuses on a prevalent and precise topic that revolves around Food Service-related issues. A sample of the comments is then manually labeled for classification purposes. After that, Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) are constructed to categorize these comments into corresponding classes, which allows issues identification and potential adjustments in the affiliated hospital's department. The results are compared with those given by the Linear Support Vector Machine (Linear SVM) using five performance metrics: accuracy, F1-score, recall, precision, and Area Under the Curve (AUC) score. The reported results constitute the first step of a broader project that revolves around categorizing a larger dataset and aims for patient experience amelioration.