Data Mining of the Reviews from Online Private Doctors

Do Patients Assess Physicians Differently in Video Visits as Compared with In-Person Visits? Insights from Text-Mining Online Physician Reviews

Introduction: Use of both in-person and video visits have become a common norm in health care delivery, especially after the COVID-19 pandemic. It is imperative to understand how patients feel about their providers and their experiences during in-person and video visits. This study examines the important factors that patients use in their reviews and differences in the relative importance. Methods: We performed sentiment analysis and topic modeling on online physician reviews from April 2020 to April 2022. Our dataset comprised 34,824 reviews posted by patients after completing in-person or video visits. Results: Sentiment analysis yielded 27,507 (92.69%) positive and 2,168 (7.31%) negative reviews for in-person visits, and 4,610 (89.53%) positive and 539 (10.47%) negative reviews for video visits. Topic modeling identified seven factors patients used in their reviews: Bedside manners, Medical Expertise, Communication, Visit Environment, Scheduling and Follow-up, Wait times, and Costs and insurance. Patients who gave positive reviews after in-person consultations more frequently mentioned communication, office environment and staff, and bedside manners. Those who gave negative reviews after in-person visits mentioned longer wait times, providers' office and staff, medical expertise, and costs and insurance problems. Patients with positive reviews after video visits emphasized communication, bedside manners, and medical expertise. However, patients posting negative reviews after video visits frequently mentioned problems with appointment scheduling and follow-up, medical expertise, wait times, costs and insurance, and technical problems in video visits. Conclusions: This study identified key factors that influence patients' assessment of their providers in in-person and video visits. Paying attention to these factors can help improve the overall patient experience.

Topic evolution and sentiment comparison of user reviews on an online medical platform in response to COVID-19: taking review data of Haodf.com as an example

Introduction

Throughout the COVID-19 pandemic, many patients have sought medical advice on online medical platforms. Review data have become an essential reference point for supporting users in selecting doctors. As the research object, this study considered Haodf.com, a well-known e-consultation website in China.

Methods

This study examines the topics and sentimental change rules of user review texts from a temporal perspective. We also compared the topics and sentimental change characteristics of user review texts before and after the COVID-19 pandemic. First, 323,519 review data points about 2,122 doctors on Haodf.com were crawled using Python from 2017 to 2022. Subsequently, we employed the latent Dirichlet allocation method to cluster topics and the ROST content mining software to analyze user sentiments. Second, according to the results of the perplexity calculation, we divided text data into five topics: diagnosis and treatment attitude, medical skills and ethics, treatment effect, treatment scheme, and treatment process. Finally, we identified the most important topics and their trends over time.

Results

Users primarily focused on diagnosis and treatment attitude, with medical skills and ethics being the second-most important topic among users. As time progressed, the attention paid by users to diagnosis and treatment attitude increased-especially during the COVID-19 outbreak in 2020, when attention to diagnosis and treatment attitude increased significantly. User attention to the topic of medical skills and ethics began to decline during the COVID-19 outbreak, while attention to treatment effect and scheme generally showed a downward trend from 2017 to 2022. User attention to the treatment process exhibited a declining tendency before the COVID-19 outbreak, but increased after. Regarding sentiment analysis, most users exhibited a high degree of satisfaction for online medical services. However, positive user sentiments showed a downward trend over time, especially after the COVID-19 outbreak.

Discussion

This study has reference value for assisting user choice regarding medical treatment, decision-making by doctors, and online medical platform design.

Validity of patients' online reviews at direct-to-consumer teleconsultation platforms: a protocol for a cross-sectional study using unannounced standardised patients

INTRODUCTION

As direct-to-consumer teleconsultation (hereafter referred to as 'teleconsultation') has gained popularity, an increasing number of patients have been leaving online reviews of their teleconsultation experiences. These reviews can help guide patients in identifying doctors for teleconsultation. However, few studies have examined the validity of online reviews in assessing the quality of teleconsultation against a gold standard. Therefore, we aim to use unannounced standardised patients (USPs) to validate online reviews in assessing both the technical and patient-centred quality of teleconsultations. We hypothesise that online review results will be more consistent with the patient-centred quality, rather than the technical quality, as assessed by the USPs.

METHODS AND ANALYSIS

In this cross-sectional study, USPs representing 11 common primary care conditions will randomly visit 253 physicians via the three largest teleconsultation platforms in China. Each physician will receive a text-based and a voice/video-based USP visit, resulting in a total of 506 USP visits. The USP will complete a quality checklist to assess the proportion of clinical practice guideline-recommended items during teleconsultation. After each visit, the USP will also complete the Patient Perception of Patient-Centeredness Rating. The USP-assessed results will be compared with online review results using the intraclass correlation coefficient (ICC). If ICC >0.4 (p<0.05), we will assume reasonable concordance between the USP-assessed quality and online reviews. Furthermore, we will use correlation analysis, Lin's Coordinated Correlation Coefficient and Kappa as supplementary analyses.

ETHICS AND DISSEMINATION

This study has received approval from the Institutional Review Board of Southern Medical University (#Southern Medical Audit (2022) No. 013). Results will be actively disseminated through print and social media, and USP tools will be made available for other researchers.

TRIAL REGISTRATION

The study has been registered at the China Clinical Trials Registry (ChiCTR2200062975).

A sentiment analysis driven method based on public and personal preferences with correlated attributes to select online doctors

This paper proposes a method to assist patients in finding the most appropriate doctor for online medical consultation. To do that, it constructs an online doctor selection decision-making method that considers the correlation attributes, in which the measure of attribute correlation is derived from the history real decision data. To combine public and personal preference with correlated attributes, it proposes a Choquet integral based comprehensive online doctor ranking method. In detail, a two stage classification model based on BERT (Bidirectional Encoder Representations from Transformers) is used to extract service features from unstructured text reviews. Then, 2-additive fuzzy measure is adopted to represent the patient public group aggregated attribute preference. Next, a novel optimization model is proposed to combine the public preference and personal preference. Finally, a case study of dxy.com is carried out to illustrate the procedure of the method. The comparison result between proposed method and other traditional MADM (multi-attribute decision-making) methods prove its rationality.

A Study of Mobile Medical App User Satisfaction Incorporating Theme Analysis and Review Sentiment Tendencies

Mobile medicine plays a significant role in optimizing medical resource allocation, improving medical efficiency, etc. Identifying and analyzing user concern elements from active online reviews can help to improve service quality and enhance product competitiveness in a targeted manner. Based on the latent Dirichlet allocation (LDA) topic model, this study carries out a topic-clustering analysis of users’ online comments and builds an evaluation index system of mobile medical users’ satisfaction by using grounded theory. After that, the evaluation information of users is obtained through an emotional analysis of online comments. Then, in order to fully consider the uncertainty of decision makers’ evaluations, rough number theory and the fuzzy comprehensive evaluation method are used to confirm the conclusions of experts and indicators and to evaluate the satisfaction of mobile medical users. The empirical results show that users are satisfied with the service quality and content quality of mobile medical apps, and less satisfied with the management and technology qualities. Therefore, this paper puts forward corresponding countermeasures from the aspects of strengthening safety supervision, strengthening scientific research, strengthening information audit, attaching importance to service quality management and strengthening doctors’ sense of gain. This study uses text mining for index extraction and satisfaction analysis of online reviews to quantitatively evaluate user satisfaction with mobile medical apps, providing a reference for the improvement of mobile medical apps. However, there are still certain shortcomings in the current study, and subsequent studies can screen false reviews for a deeper study of online review information.

Investigating the effect of service feedback and physician popularity on physician demand in the virtual healthcare environment

Purpose

This study examines how service feedback and physician popularity affect physician demand in the context of virtual healthcare environment. Based on the signaling theory, the critical factor of environment uncertainty (i.e. disease risk) and its impact on physician demand is also investigated. Further, the research on the endogeneity of online reviews in healthcare is also examined in the current study.

Design/methodology/approach

A secondary data econometric analysis using 3-wave data sets of 823 physicians obtained from two PRWs (Healthgrades and Vitals) was conducted. The analysis was run using the difference-in-difference method to consider physician and website-specific effects.

Findings

The study's findings indicate that physician popularity has a stronger positive effect on physician demand compared with service feedback. Improving popularity leads to a relative increase in the number of appointments, which in turn enhance physician demand. Further, the impact of physician popularity on physician demand is positively mitigated by the disease risk.

Originality/value

The authors' research contributes to a better understanding of the signaling transmission mechanism in the online healthcare environment. Further, the findings provide practical implications for key stakeholders into how an efficient feedback and popularity mechanism can be built to enhance physician service outcomes in order to maximize the financial efficiency of physicians.

Exploring the User Interaction Network in an Anxiety Disorder Online Community: An Exponential Random Graph Model with Topical and Emotional Effects

The increasing number of people with anxiety disorders presents challenges when gathering health information. Users in anxiety disorder online communities (ADOCs) share and obtain a variety of health information, such as treatment experience, drug efficacy, and emotional support. This interaction alleviates the difficulties involved in obtaining health information. Users engage in community interaction via posts, comments, and replies, which promotes the development of an online community as well as the wellbeing of community users, and research concerning the formation mechanism of the user interaction network in ADOCs could be beneficial to users. Taking the Anxiety Disorder Post Bar as the research object, this study constructed an ADOC user interaction network based on users' posts, comments, and personal information data. With the help of exponential random graph models (ERGMs), we studied the effects of the network structure, user attributes, topics, and emotional intensity on user interaction networks. We found that there was significant reciprocity in the user interaction network in ADOCs. In terms of user attributes, gender homogeneity had no impact on the formation of the user interaction network. Experienced users in the community had obvious advantages, and experienced users could obtain replies more easily from other members. In terms of topics, pathology popularization showed obvious homogeneity, and symptoms of generalized anxiety disorder showed obvious heterogeneity. In terms of emotional intensity, users with polarized emotions were more likely to receive replies from users with positive emotions. The probability of interaction between two users with negative emotions was small, and users with opposite emotional polarity tended to interact, especially when the interaction was initiated by users with positive emotions.

An improved approach based on dynamic mixed sampling and transfer learning for topic recognition: a case study on online patient reviews

Purpose

This study aimed to adapt existing text-mining techniques and propose a novel topic recognition approach for textual patient reviews.

Design/methodology/approach

The authors first transformed multilabel samples for adapting model training forms. Then, an improved method was proposed based on dynamic mixed sampling and transfer learning to improve the learning problem caused by imbalanced samples. Specifically, the training of our model was based on the framework of a convolutional neural network and self-trained Word2Vector on large-scale corpora.

Findings

Compared with the SVM and other CNN-based models, the CNN+ DMS + TL model proposed in this study has made significant improvement in F1 score.

Originality/value

The improved methods based on dynamic mixed sampling and transfer learning can adequately manage the learning problem caused by the skewed distribution of samples and achieve the effective and automatic topic recognition of textual patient reviews.

Peer review

The peer-review history for this article is available at: https://publons.com/publon/10.1108/OIR-01-2021-0059.