Online review analysis-based multi-criteria decision-making for evaluating patient satisfaction: A case study of the Haodf website

The online reviews provided by patients contain many aspects of patient satisfaction (PS). An accurate understanding of PS can help hospitals and doctors quickly find the direction of medical service improvement and help patients select appropriate doctors. However, online reviews are texts in which patients show true feelings without constraints. Therefore, identifying, measuring, and representing PS are difficult. To solve these problems, we propose the online review analysis-based multi-criteria decision-making (MCDM) method. First, an aspect extraction method integrating dependency parsing and attention-based aspect extraction (ABAE) is proposed, and nine criteria for PS evaluation are extracted from the online reviews of the Haodf website. Second, a sentiment analysis method based on multiple dictionaries and dependency relations is developed to measure PS under each criterion in reviews. Then, an MCDM method based on a probabilistic linguistic term set representing PS is used to assess PS when considering patients’ loss aversion. Finally, the proposed method is verified in the evaluation of lung cancer patients’ satisfaction with doctors. The results show that our extracted criteria have higher coherence and accuracy compared to those extracted by other aspect extraction methods, and the proposed online review analysis-based MCDM method outperforms state-of-the-art methods in PS identification, measurement, and representation.