Design of a smart medical service quality evaluation system based on a hybrid multi-criteria decision model Authors: Gang Wang, Qigan Shao Published: 2024 ABSTRACT As medical informatization continues to progress, the sophistication of smart medicine (SM) systems has led to a marked enhancement in the caliber of medical services provided. Nevertheless, the existing body of literature is sparse when it comes to frameworks for evaluating service quality, specifically within the domain of smart medicine (SM). Drawing on a hybrid multi-criteria decision-making model that fuses the best–worst method (BWM) with the VIKOR approach, this research has crafted an innovative framework for assessing the service quality of SM. BWM was used to obtain the weights of all dimensions and indicators under each dimension. Then, the service quality of hospitals H1, H2 and H3 in Xiamen was evaluated using the VIKOR method. The results showed that smart appointments, diagnosis and treatment are three important dimensions to evaluate SM service quality in medical institutions. The stability and robustness of the model were proved through sensitivity analysis. Findings suggest that hospitals can strengthen the construction of their appointment information platforms, the quality management of internal doctors, and the information connection between self-service terminals and information platforms to improve hospital service quality in the construction of smart medicine. Introduction Smart medicine (SM) is a concept originating from the “smart planet” strategy proposed by the International Business Machines Corporation (IBM) in 2009, which consists of six major fields, namely smart power, smart medicine, smart cities, smart traffic, smart supply chains, and smart banks1. SM has been successively developed from traditional medicine, digital medicine, and information-based medicine2. The big data of traditional medicine are formed by digitizing the existing massive medicine information and cases, and on this basis, the application and innovation of SM technology are carried out. Artificial intelligence (AI) technology offers the capability to delve into medical data, facilitating the identification and diagnosis of diseases, as well as bolstering medical health management and educational initiatives. As established based on informatization, SM is superior to traditional medical informatization in terms of intellectualization and the accessibility and quality of services3. By means of information sharing, SM has established an individual-based electronic health record, realizing face-to-face data exchange between medical professionals and patients. It has also achieved the sharing of information with regional health information institutions, showing its great value for application in the medical industry4. With the advancement of SM, the level of medical services provided by hospitals has gradually improved, and the variety has been enriched in a multilayer manner5. SM has the potential to markedly elevate service quality through increased efficiency, tailored services, and informed decision-making powered by data. However, it also faces challenges including reliance on technology, data security concerns, and regulatory ethical issues. It is essential to strike a balance between fostering innovation and managing associated risks. Although the SM system and the service level of hospitals are being gradually improved, there are still several prominent problems, such as insufficiency, unsatisfactory quality, excessive concentration, inappropriate allocation, and uncoordinated development of medical resources6. The great disparity in the number of doctors and patients has resulted in a severe deficiency of doctors and an extension of their working hours; furthermore, due to the limited number of hospital beds and the excessive number of patients, hospitals cannot take care of all patients at the same time7–9. However, previous literature pointed out that there was a lack of objective methods for the evaluation of the effectiveness of SM services. For the general public, smart medicine (SM) signifies the availability of more convenient and cost-effective medical services, along with a broader spectrum of health services. On the institutional side, SM translates to a more judicious distribution of medical resources, the delivery of more efficient and secure medical services, and a more nuanced approach to management10. The general public is the service recipient, while the medical institutions are service providers. The value of SM care is created by the service providers and service recipients11. In the process of disease intervention, a set of intelligent auxiliary diagnosis and treatment systems based on AI is built to provide doctors with real-time auxiliary diagnosis recommendations, standardized and personalized treatment plan recommendations, and real-time medical knowledge support. Medical evaluation should be performed based on medical diagnosis and treatment. In this study, it was assumed that the user evaluation process is an important source of the determination of the quality of SM services. Meanwhile, to evaluate the quality of SM services, the current development level of SM should be taken into account. To bridge this research gap, this study aimed to achieve the following objectives: (1) determine the indicators for evaluating SM service quality; (2) determine the significance of these indicators and their interrelationships; (3) develop a model to select alternative SM departments; and (4) identify key factors for sustainable improvement of SM service quality. The determination of the quality of SM services is a multidimensional decision management problem. The multiple criteria decision-making (MCDM) model is often used for evaluating complex systems. Different from classical regression statistical analysis methods, the MCDM does not require exploration of data distribution characteristics and independence assumptions12. Indicators are obtained for creating a framework for expert judgement, and then the weights and performances of alternatives can be calculated. However, the MCDM model has rarely been applied to SM service quality assessment. Qualitative research or traditional statistical methods are often used to study the quality of healthcare. At present, many problems in various fields have been resolved using the MCDM model, such as stacker selection, material supplier selection, evaluation of the quality of urban life, project management, food waste treatment, green supply chain risks, and supplier selection13–19. However, most of the studies on the quality of medical services were performed based on traditional medicine. A questionnaire survey was conducted on service quality based on 262 patients in a hospital in Turkey with the fuzzy “service quality” (SER-QUAL) model, and the results showed that the gap scores for all items were negative, implying that the patients were generally dissatisfied with the hospital20,21. A detailed case study on nine public hospitals in Sicily, Italy was performed through combining fuzzy SERQUAL with the analytic hierarchy process (AHP), in which four core service standards and 15 essential service items were assessed, to identify the aspects of dissatisfaction with public health services in the region22. In this study, SM is considered a new medical model that has developed from traditional medicine. The medical treatment process based on traditional medicine remains unchanged and is used as the basis for evaluating the quality of intelligent medical services. In addition, evaluating the service quality of SM also requires a “smart” perspective, that is, how effective the IoT technology is in medical service. In this paper, the service quality of SM in three hospitals in China is evaluated with MCDM to fill the gap in the literature and to construct an evaluation system to measure the service quality of SM. The best and worst method (BWM), as an MCDM method, was developed by Rezaei for calculating the weight of indicators of a system23. Compared to traditional methods such as AHP, the BMW method does not require respondents to evaluate each option in detail, but focuses on the best and worst features of each option, simplifying the questionnaire and reducing the burden on respondents. The BMW method separates the best and worst features, making it easier for respondents to make decisions. It can therefore flexibly adjust the design of the problem according to research needs, such as customizing the options for the best and worst features for different groups or situations. The typical multi-criteria compromise methods, VIKOR and TOPSIS, have their own characteristics. The scheme obtained by VIKOR is the most similar to the ideal solution, while the best solution obtained by TOPSIS is not always similar to the ideal solution; at the same time, the scheme obtained by VIKOR is a compromise scheme with priorities24. The proposed model offers a novel Index system to evaluate SM service quality. In addition, this study proposed a model which combines the BWM method and the VIKOR technique. The contributions are as follows:This study presents a controllable evaluation framework and hybrid MCDM model as a reference for evaluating the quality of SM services.The study identified key factors that affect the quality of SM services and provides direction for achieving sustainable development by focusing on SM.Three hospitals were used as case studies to display the applicability of the proposed model, and this study further demonstrates the practicality and effectiveness of the model through sensitivity analysis. This study presents a controllable evaluation framework and hybrid MCDM model as a reference for evaluating the quality of SM services. This study presents a controllable evaluation framework and hybrid MCDM model as a reference for evaluating the quality of SM services. The study identified key factors that affect the quality of SM services and provides direction for achieving sustainable development by focusing on SM. The study identified key factors that affect the quality of SM services and provides direction for achieving sustainable development by focusing on SM. Three hospitals were used as case studies to display the applicability of the proposed model, and this study further demonstrates the practicality and effectiveness of the model through sensitivity analysis. Three hospitals were used as case studies to display the applicability of the proposed model, and this study further demonstrates the practicality and effectiveness of the model through sensitivity analysis. As seen in Fig. 1, this study was divided into three phases: (1) the first phase determined the indicators of the service quality of smart medicine based on the literature review and expert opinions; (2) the second phase calculated and ranked the weights of these indicators with the BWM method; and (3) the third phase ranked the alternative hospitals with optimal indicators with the VIKOR method.Fig. 1Research framework. Research framework. Literature review Assessing hospital service quality is paramount, as it is intrinsically linked to patient satisfaction and the effectiveness of treatments. This evaluation can lead to a reduction in medical errors, an optimization of resource distribution, and an enhancement of hospitals' market competitiveness. In addition, continuous improvement in service quality helps meet regulatory requirements, enhances the hospital’s image, and improves employee job satisfaction and retention. In this way, hospitals can provide safer and more effective medical services for patients and promote overall health management and disease prevention. For the evaluation of medical quality, experts have proposed solutions from the perspective of MCDM. Li et al.25 proposed a novel hybrid model which integrated the fuzzy best–worst method with the entropy and MARCOS approach to evaluate the service quality of mobile medical apps. Tushar et al.26 utilized the Delphi method to confirm the factors and the grey-DEMATEL method to evaluate the service quality of hospitals. Tuzkaya et al.27 proposed an interval valued intuitionistic fuzzy-preference ranking organization method for enrichment evaluations (IVIF-PROMETHEE) method to evaluate the service quality of public hospitals based on patients’ views. Handayani et al.28 used the entropy approach to evaluate the service quality of the hospital from the perspective of stakeholders. Shafii et al.29 applied the fuzzy analytical hierarchy process (FAHP) and the technique for order preference by similarity to ideal solution (TOPSIS) to evaluate the service quality of hospitals. Wu et al.30 used the analytic hierarchy process (AHP) to get the key indicators for evaluating the service quality of acupuncture. Akdag et al.31 used AHP and TOPSIS to assess the service quality of hospitals. Gregório et al.32 proposed the ELECTRE method to consider interactions between criteria of public hospitals. Mete et al.33 proposed a hybrid MCDM model which integrated Entropy and the BWM method to evaluate countries’ COVID-19 risks. Krishankumar et al.34 proposed a criteria importance through inter-criteria correlation (CRITIC) approach to select healthcare service providers, and Chowdhury et al.35 integrated machine learning, TOPSIS and the entropy method to detect COVID-19. Through literature analysis, it is not difficult to find that most of the literature focuses on using a variety of quantity of traditional medical services, and the evaluation literature on the quality of smart medical services is rare, as seen in Table 1. Due to the application of intelligent technology, the evaluation indicators of service quality in smart medical care are very different from those in traditional medical care. As far as we know, there are very few solutions to measure SM performance. In order to address this research gap, this paper extracted indicators for evaluating SM through literature analysis, and designed a system to measure the quality of SM services.Table 1Review of related literature on hospital service quality assessment.TopicMethodFindingsSourceService quality of medical appBWM, entropy, MARCOSEstablished a new mobile medical app service assessment index system25Healthcare service managementDelphi, grey-DEMATELIdentified 11 causal factors and nine effect group factors26Healthcare service quality for divisionsIVIF-PROMETHEE,IVIF-TOPSISThe proposed methodology can be a useful tool for hospital management27Service quality of private hospitalsEntropyHospitals should pay more attention to human resources as the top dimension28Hospital service qualityFAHP, TOPSISResponsiveness, assurance, security, tangibles, health communication and patient orientation were identified as affecting hospital services quality29Acupuncture service qualityAHPOptimizing the operating procedures will help to improve acupuncture services30Hospital service qualityFuzzy set, TOPSISMultiple Criteria Decision Making (MCDM) was to evaluate the service quality of Turkish hospitals31Multi-criteria decision-aiding for public hospitalsELECTREPerceived service quality is directly related to achieving optimal health outcomes32COVID-19 risks evaluationEntropy, BWMCOVID-19 risk assessment of countries is the most significant novelty for the relevant literature33Healthcare service qualityCRITICDeveloped model reduces human intervention34Performance machine learning techniqueEntropy, TOPSISProposed method outperforms the state-of-the-art models35 Table 1Review of related literature on hospital service quality assessment.TopicMethodFindingsSourceService quality of medical appBWM, entropy, MARCOSEstablished a new mobile medical app service assessment index system25Healthcare service managementDelphi, grey-DEMATELIdentified 11 causal factors and nine effect group factors26Healthcare service quality for divisionsIVIF-PROMETHEE,IVIF-TOPSISThe proposed methodology can be a useful tool for hospital management27Service quality of private hospitalsEntropyHospitals should pay more attention to human resources as the top dimension28Hospital service qualityFAHP, TOPSISResponsiveness, assurance, security, tangibles, health communication and patient orientation were identified as affecting hospital services quality29Acupuncture service qualityAHPOptimizing the operating procedures will help to improve acupuncture services30Hospital service qualityFuzzy set, TOPSISMultiple Criteria Decision Making (MCDM) was to evaluate the service quality of Turkish hospitals31Multi-criteria decision-aiding for public hospitalsELECTREPerceived service quality is directly related to achieving optimal health outcomes32COVID-19 risks evaluationEntropy, BWMCOVID-19 risk assessment of countries is the most significant novelty for the relevant literature33Healthcare service qualityCRITICDeveloped model reduces human intervention34Performance machine learning techniqueEntropy, TOPSISProposed method outperforms the state-of-the-art models35 Review of related literature on hospital service quality assessment. In order to ensure the fairness, accuracy, and practicality of the evaluation process, and thus promote the continuous improvement and development of medical service quality, this study proposes four principles for selecting smart medical service quality indicators, as follows:Evaluation indicators should objectively reflect the actual situation of smart medical services, and the data sources should be reliable.The evaluation indicators should comprehensively cover key areas of smart medical services, such as technology application, service process, patient experience, safety, efficiency, and so on, to ensure the comprehensiveness and representativeness of the evaluation results.The indicator system should have a clear hierarchical structure, forming an orderly evaluation framework that facilitates systematic analysis and comprehensive evaluation.Evaluation indicators should be centered on the needs and experiences of patients, emphasizing patient satisfaction and service quality, ensuring the humanization of medical services and maximizing patient benefits. Evaluation indicators should objectively reflect the actual situation of smart medical services, and the data sources should be reliable. Evaluation indicators should objectively reflect the actual situation of smart medical services, and the data sources should be reliable. The evaluation indicators should comprehensively cover key areas of smart medical services, such as technology application, service process, patient experience, safety, efficiency, and so on, to ensure the comprehensiveness and representativeness of the evaluation results. The evaluation indicators should comprehensively cover key areas of smart medical services, such as technology application, service process, patient experience, safety, efficiency, and so on, to ensure the comprehensiveness and representativeness of the evaluation results. The indicator system should have a clear hierarchical structure, forming an orderly evaluation framework that facilitates systematic analysis and comprehensive evaluation. The indicator system should have a clear hierarchical structure, forming an orderly evaluation framework that facilitates systematic analysis and comprehensive evaluation. Evaluation indicators should be centered on the needs and experiences of patients, emphasizing patient satisfaction and service quality, ensuring the humanization of medical services and maximizing patient benefits. Evaluation indicators should be centered on the needs and experiences of patients, emphasizing patient satisfaction and service quality, ensuring the humanization of medical services and maximizing patient benefits. Based on the above four principles, six dimensions are initially formulated based on the review of the existing literature and the assessment of the service quality of smart medicine: smart appointment, smart consultation, smart diagnosis and treatment, smart nursing, smart settlement, and smart healthcare. Smart appointment Smart appointment refers to the process for a patient to make an appointment in advance including selecting the physician, scheduling a visit, and locating the hospital by setting up a personal file based on his/her ID card or medical insurance card with a smart phone or PC. Compared with the traditional process, patients do not need to wait in line at the hospital for registration; therefore, smart appointments can significantly save the patients’ time and reduce the hospital labor cost, thus improving the efficiency of hospital management and the patients’ visiting experience and satisfaction. The outpatient process based on a “mutual health data bank” could realize the intelligence of the outpatient process, and internet appointments should be taken as an important criterion for evaluating smart appointments36. When constructing the smart medicine service platform of hospitals in Turkey, the service platform information security guarantee system should be established based on the relevant national security guarantee standards to ensure the information security of residents when using the service platform37. A real-name filing system should be established, and the residents should complete real-name registration and archiving with their ID cards or health insurance cards. The patients making appointments with a smartphone app should also be provided with the GPS location of the hospital and map guidance, to save time by preventing them from making detours. Real-name archiving and smart navigation are two important dimensions of weighing smart appointments. Short message service (SMS) reminders prior to physical examinations could effectively reduce the rate of missing appointments, and improve the actual examination rate and appointment satisfaction38. Based on the above-mentioned findings, four indicators were set under smart appointment, namely real-name registration and archiving, Internet appointments, appointment reminders, and intelligent navigation. Smart consultation Smart consultation refers to the process whereby patients who have made an Internet appointment can check the waiting order, calling status and average waiting time prior to their arrival at the hospital, while those who have not made an Internet appointment can make an on-site appointment with the multi-functional self-service terminal. Short message service (SMS) reminders help improve appointment attendance and medication adherence, and shape behaviors for various healthcare issues39. By analyzing the consultation process and characteristics, an appointment queuing call system with a unified serial number database can be established for telephone appointments, short message service (SMS) appointments, online appointments and timely self-service appointment service at the hospital40. Innovative service models can be established and implemented through a series of measures such as building a 3D reservation service network, adopting an intelligent triage call system, and widely implementing self-service. That model fully integrated service counter functions, and promoted integrated outpatient clinics, which improved the environment and experience of visiting, and effectively improved outpatient service quality and efficiency41. Based on the above literature review, smart consultation covers the patients engaging in online and on-site appointments. Under the support of modern technology, most hospitals have established their own comprehensive information platforms equipped with self-service terminals, which can update the data in real time; meanwhile, they have also established calling systems interfaced with the appointment platform, for guiding the patients. Compared with the traditional consultation process, the Internet-based model can optimize all nodes of the consultation process, making it more rational, smarter, and more responsive to the needs of the patients. Five indicators were set under smart consultation, namely online waiting order, online call query, average waiting time, self-service registration, and triage call. Smart diagnosis and treatment Smart diagnosis and treatment refer to the process whereby patients can handle various procedures and enjoy services with modern technology tools. The implementation of the one-stop inpatient service model is conducive to more efficient service delivery and better patient experience, and the implementation of this model can effectively avoid the prolonged waiting time due to preparation problems42. Through combining this platform with the existing information system, the information construction was extended to the bedside, and the access to information was provided directly for the patients. The platform has greatly improved the hospitalization experience and the efficiency of medical care, and has provided data support for hospital management43. Based on the above literature review, smart diagnosis and treatment involves the initial consultation by the outpatient doctors, and the processes involved when the patient requires hospitalization and after admission. The patients’ perceptions of SM combining medical treatment with modern technology mainly involve information-based doctor–patient communications, notification of the test results necessary for diagnosis treatment, reasonable and flexible appointments for non-emergency surgery, rapid admission self-service, self-service hospitalization, a referral service, and convenient discharge. In this paper, the following indicators were set for smart diagnosis and treatment, namely doctor–patient communication, timely notification of test results, self-service appointment for surgery, self-service admission, self-service hospitalization, and self-service discharge. Smart nursing Smart nursing refers to the nursing of patients during hospitalization, and involves accurate identification of the patients with modern equipment, verbal and electronic notification provided by the doctors and nurses for the patients, as well as automatic alert settings based on multiple forms of monitoring with respect to infusion and injection management, bedside smart calling, and post-discharge follow-up. The application of 5G technology can build a digital ward comprehensive service platform consisting of a smartphone electronic medical record query system and a wireless infusion monitoring system based on infrared sensing technology44. The wireless infusion monitoring system based on infrared sensing technology can monitor the drip rate of infusion and its progress with the infrared sensing technology, which reduced the work of frequent rounds and improved efficiency. Traditional hand-written patient identification wristbands have many problems45. For example, medication may be given to the wrong patients due to poor identification, whereas personal digital assistants (PDA) could conveniently, smartly and safely identify the patients, and make the identity of inpatients tangible on barcode-printed wristbands, which could protect patient privacy and effectively ensure continuous legibility of the barcode information, without being affected by bathing and drugs. It could also optimize accountable nursing, ensure the safety of nursing, reduce the time for nursing, and improve patient satisfaction46. In this paper, five indicators were set under smart nursing, namely patient identification checks, inpatient medical order execution, infusion and injection management, patient bedside calling, and post-discharge follow-up. Smart settlement Smart settlement refers to the process whereby patients can perform recharging and settlement with medical insurance cards by medical self-service terminals in hospital lobbies and consultation rooms, or using bedside self-service platforms. There are various forms of recharging and settlement with medical insurance cards, including at least one of the current mainstream payment modes, such as WeChat, Alipay, and UnionPay. The “Internet+” settlement mode could facilitate hospitals’ adaptation to the development of Internet+ medicine, improve patient experience and settlement efficiency, and increase the level of hospital management47. The integrated “Internet+” medical platform could optimize appointments, registration, payments, and inquiries, and optimize the application of network technology for diagnosis and treatment, procurement, logistics, and follow-up visits; in addition, it could also realize real-time reimbursement by medical insurance companies and real-time claims for compensation by commercial insurance companies. One of the key measures for further improving the experience is to reduce settlement trips. The variety of payment methods should be added, and mature payment technologies should be adopted to make the payment more convenient. On the other hand, the advantages of Internet technology should be exploited to optimize and adjust the settlement process, set flexible payment pathways, and reduce settlement trips by digital methods while ensuring the safety of funds. Based on the above literature review, six indicators were set under smart settlement, namely real-name pre-deposit, medical insurance real-time network settlement, outpatient payment, payment methods, inpatient prepayment, and self-service printing. Smart healthcare Smart healthcare refers to the health services provided by the hospitals for the pre-treatment patients and treated patients, such as health education, health information pushing, the provision of health consultation channels, and the establishment of electronic health records for patients. Smart healthcare mainly serves people with health demands, and provides them with the required information about their health. Patients who are positive in their consultations would have a positive medical experience48. Health education on the relevant diseases for patients during outpatient infusion and treatment could effectively relieve the nervousness of both patients and their families, and improve their trust and cooperation, thus improving patient satisfaction49. Overall patient satisfaction increased by 3.65% after implementation of a “Health Edu-cation Mobile Course” program, which guided users to prevent or treat diseases such as colds by mastering basic medical knowledge50. Electronic health records can offer potential benefits for patients, physicians and public health51. In addition, physicians can make use of the information provided by other providers, realize remote access to medical records, remind patients of service needs, monitor electronic prescriptions and drug interactions, and make use of clinical information for research. As shown in Table 2, this paper determines a system for evaluating the quality of smart medical care, which includes six dimensions and 29 indicators of smart medical services. The given services are acceptable if the patients’ perceptions of service quality are equal to what they expect. The given services are ideal if the patients’ perceptions of service quality are higher than what they expect. However, the services are unacceptable if the patients’ perceptions of service quality are lower than what they expect, in which case the service quality should be improved from multiple dimensions and indicators to meet the needs of the patients and improve patient satisfaction.Table 2The smart medicine service quality evaluation system.DimensionDescriptionIndicatorsSourceSmart appointment (D1)Smart appointment refers to the process whereby a patient makes an appointment in advance to select the physician, schedules a visit, and locates the hospital by setting up a personal file based on his/her ID card or medical insurance card with a smart phone or PCReal name registration and archiving (C11)Internet appointment (C12)Appointment reminder (C13)Intelligent navigation (C14)6,36–38Smart consultation (D2)Smart consultation refers to the process whereby patients who have made an Internet appointment can check the waiting order, calling status and average waiting time prior to arrival at the hospital, while those who have not made an Internet appointment can make an on-site appointment with the multi-functional self-service terminalOnline waiting order (C21)Online calling inquiry (C22)Average waiting time (C23)Self-service registration (C24)Triage calling (C25)39–41Smart diagnosis and treatment (D3)Smart diagnosis and treatment refers to the process whereby patients can handle various procedures and enjoy services with modern technology toolsDoctor–patient communication (C31)Prompt notification of test results (C32)Self-service appointment for surgery (C33)Self-service admission (C34)Self-service hospitalization (C35)Self-service discharge (C36)42,43Smart nursing (D4)Smart nursing refers to nursing for patients during hospitalization, involving accurate identification of the patients with modern equipment, verbal and electronic notification provided by the doctors and nurses for the patients, as well as automatic alert settings based on multiple forms of monitoring with respect to infusion and injection management, bedside smart calling, and post-discharge follow-upPatient identification checking (C41)Medical advice execution (C42)Infusion and injection management (C43)Patient bedside call (C44)Discharge follow-up (C45)44–46Smart settlement (D5)Smart settlement refers to the process whereby patients can perform recharging and settlement with medical insurance cards by medical self-service terminals at hospital lobbies and consultation rooms or the bedside self-service platforms (except for window settlement)Real name deposit (C51)Medical insurance online real-timesettlement (C52)Various payment settlementmethods (C53)Payment methods (C54)Self-service printing (C55)46–48Smart healthcare (D6)Smart healthcare refers to the health services provided by hospitals for pre-treatment patients and treated patients, such as health education, health information pushing, provision of health consultation channels, and establishment of electronic health records for patients. Smart healthcare mainly serves people with health demands, and provides them with the required information about their healthHealth education (C61)Information pushing (C62)Health consultation response rate (C63)Electronic health record (C64)48–51 Table 2The smart medicine service quality evaluation system.DimensionDescriptionIndicatorsSourceSmart appointment (D1)Smart appointment refers to the process whereby a patient makes an appointment in advance to select the physician, schedules a visit, and locates the hospital by setting up a personal file based on his/her ID card or medical insurance card with a smart phone or PCReal name registration and archiving (C11)Internet appointment (C12)Appointment reminder (C13)Intelligent navigation (C14)6,36–38Smart consultation (D2)Smart consultation refers to the process whereby patients who have made an Internet appointment can check the waiting order, calling status and average waiting time prior to arrival at the hospital, while those who have not made an Internet appointment can make an on-site appointment with the multi-functional self-service terminalOnline waiting order (C21)Online calling inquiry (C22)Average waiting time (C23)Self-service registration (C24)Triage calling (C25)39–41Smart diagnosis and treatment (D3)Smart diagnosis and treatment refers to the process whereby patients can handle various procedures and enjoy services with modern technology toolsDoctor–patient communication (C31)Prompt notification of test results (C32)Self-service appointment for surgery (C33)Self-service admission (C34)Self-service hospitalization (C35)Self-service discharge (C36)42,43Smart nursing (D4)Smart nursing refers to nursing for patients during hospitalization, involving accurate identification of the patients with modern equipment, verbal and electronic notification provided by the doctors and nurses for the patients, as well as automatic alert settings based on multiple forms of monitoring with respect to infusion and injection management, bedside smart calling, and post-discharge follow-upPatient identification checking (C41)Medical advice execution (C42)Infusion and injection management (C43)Patient bedside call (C44)Discharge follow-up (C45)44–46Smart settlement (D5)Smart settlement refers to the process whereby patients can perform recharging and settlement with medical insurance cards by medical self-service terminals at hospital lobbies and consultation rooms or the bedside self-service platforms (except for window settlement)Real name deposit (C51)Medical insurance online real-timesettlement (C52)Various payment settlementmethods (C53)Payment methods (C54)Self-service printing (C55)46–48Smart healthcare (D6)Smart healthcare refers to the health services provided by hospitals for pre-treatment patients and treated patients, such as health education, health information pushing, provision of health consultation channels, and establishment of electronic health records for patients. Smart healthcare mainly serves people with health demands, and provides them with the required information about their healthHealth education (C61)Information pushing (C62)Health consultation response rate (C63)Electronic health record (C64)48–51 The smart medicine service quality evaluation system. Real name registration and archiving (C11) Internet appointment (C12) Appointment reminder (C13) Intelligent navigation (C14) Online waiting order (C21) Online calling inquiry (C22) Average waiting time (C23) Self-service registration (C24) Triage calling (C25) Doctor–patient communication (C31) Prompt notification of test results (C32) Self-service appointment for surgery (C33) Self-service admission (C34) Self-service hospitalization (C35) Self-service discharge (C36) Patient identification checking (C41) Medical advice execution (C42) Infusion and injection management (C43) Patient bedside call (C44) Discharge follow-up (C45) Real name deposit (C51) Medical insurance online real-time settlement (C52) Various payment settlement methods (C53) Payment methods (C54) Self-service printing (C55) Health education (C61) Information pushing (C62) Health consultation response rate (C63) Electronic health record (C64) Research method All methods were carried out in accordance with relevant guidelines and regulations. All experimental protocols were approved by Xiamen University of Technology. We obtained informed consent from all participants and/or their legal guardians. The BWM method The BWM method, as an MCDM method developed by Rezaei, can be used for calculating the index weight19. BWM is similar to AHP, and is also based on pairwise comparison; however, it is not the comparison of any two dimensions or criteria, but rather a structured way of comparison, with the specific steps described as follows: Step 1 Determine guidelines for decision-making about the set of criteria obtained from the literature review and experts’ opinions expressed as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{ {C_{1} ,C_{2} ,C_{3} ,...,C_{n} } \right\}$$\end{document}C1,C2,C3,...,Cn, indicating n criteria. Step 2 From the set of criteria \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left\{ {C_{1} ,C_{2} ,C_{3} ,...,C_{n} } \right\}$$\end{document}C1,C2,C3,...,Cn, select the optimal criterion CB and the worst criterion CW. Step 3 According to Table 3, mark within the range of 1–9 to determine the degree of preference of the optimal criterion over other criteria; determine the comparison vector \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$AB = \left\{ {a_{B1} ,a_{B2} , \ldots ,a_{Bn} } \right\}$$\end{document}AB=aB1,aB2,…,aBn, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a_{Bi}$$\end{document}aBi refers to the preference of the optimal criterion B over criterion i.Table 3BWM pairwise comparison language scale.Language variablesScoreEqually important1Between equally important and slightly important2Slightly important3Between slightly important and obviously important4Obviously important5Between obviously important and strongly important6Strongly important7Between strongly important and extremely important8Extremely important9 Table 3BWM pairwise comparison language scale.Language variablesScoreEqually important1Between equally important and slightly important2Slightly important3Between slightly important and obviously important4Obviously important5Between obviously important and strongly important6Strongly important7Between strongly important and extremely important8Extremely important9 BWM pairwise comparison language scale. Step 4 Determine the preference of other criteria over the worst criterion; determine the comparison vector \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A_{w} = \left\{ {a_{1w} ,a_{2w} ,...,a_{nw} } \right\}$$\end{document}Aw=a1w,a2w,...,anw, where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a_{iw}$$\end{document}aiw refers to the preference of criterion i over the worst criterion. Step 5 Determine a mathematical programming Eq. (1) and solve it to obtain the optimal weight \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left( {w_{1}^{*} ,w_{2}^{*} ,w_{3}^{*} ,...,w_{n}^{*} } \right)$$\end{document}w1∗,w2∗,w3∗,...,wn∗.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} & \min \;\max_{m} \left\{ {\left| {\frac{{w_{B} }}{{w_{m} }} - a_{Bm} } \right|,\left| {\frac{{w_{m} }}{{w_{W} }} - a_{mW} } \right|} \right\} \\ & \sum\limits_{{\text{m}}}^{{}} {w_{m} } { = }1 \\ & w_{{\text{m}}} \ge 0 \\ \end{aligned}$$\end{document}minmaxmwBwm-aBm,wmwW-amW∑mwm=1wm≥0where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$w_{{\text{m}}}$$\end{document}wm means the weight of a dimension or a criterion m, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a_{Bm}$$\end{document}aBm refers to the preference of the best criterion over criterion m, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a_{mW}$$\end{document}amW refers to criterion m to the worst criterion. Step 6 Calculate the CR value. CR is the consistency ratio, which can be calculated as follows:2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$CR = \frac{{x^{*} }}{CI}$$\end{document}CR=x∗CI The consistency index (CI) table is shown in Table 4.Table 4Consistency index.abw123456789CI00.4411.632.333.734.475.23 Table 4Consistency index.abw123456789CI00.4411.632.333.734.475.23 Consistency index. The VIKOR method The VIKOR method, proposed by Opricovic, is a compromise ranking method, which is mainly used to resolve conflicts between multiple indicators52. On the basis of “approaching the ideal solution and reaching a consensus through concession”, it combines the subjective preference of decision-makers with the maximum group benefits and minimum individual regrets of the alternatives, and establishes a compromise solution. The steps are as follows: Step 1 Assume the evaluation value of the alternative hospital under the evaluation index is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop C\nolimits_{j} \; \left( {j = 1,2,3 \ldots \ldots m} \right)$$\end{document}Cjj=1,2,3……m is \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{ij}$$\end{document}fij. Step 2 Calculate the average decision matrix.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$F = \frac{1}{k}\sum\limits_{k = 1}^{k} {\mathop F\nolimits_{k} }$$\end{document}F=1k∑k=1kFk Step 3 Calculate the optimal \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{*}$$\end{document}fj∗ and the inferior \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{ - }$$\end{document}fj- of all indicators, where j = 1, 2, … n.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{*} = Max\left( {\mathop f\nolimits_{ij} } \right)$$\end{document}fj∗=Maxfij5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{ - } = Min\left( {\mathop f\nolimits_{ij} } \right)$$\end{document}fj-=Minfijwhere \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{*}$$\end{document}fj∗ represents the positive ideal value and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop f\nolimits_{j}^{ - }$$\end{document}fj- represents the negative ideal value. Step 4 Calculate Si and Ri.6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop S\nolimits_{i} = \sum\limits_{j = 1}^{n} {\mathop W\nolimits_{j} \left[ {{{\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{ij} } \right)} \mathord{\left/ {\vphantom {{\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{ij} } \right)} {\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{j}^{ - } } \right)}}} \right. \kern-0pt} {\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{j}^{ - } } \right)}}} \right]}$$\end{document}Si=∑j=1nWjfj∗-fijfj∗-fijfj∗-fj-fj∗-fj-7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop R\nolimits_{i} = Max_{j} \left[ {\mathop W\nolimits_{j} {{\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{ij} } \right)} \mathord{\left/ {\vphantom {{\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{ij} } \right)} {\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{j}^{ - } } \right)}}} \right. \kern-0pt} {\left( {\mathop f\nolimits_{j}^{*} - \mathop f\nolimits_{j}^{ - } } \right)}}} \right]$$\end{document}Ri=MaxjWjfj∗-fijfj∗-fijfj∗-fj-fj∗-fj-where Si refers to the distance between the alternative scheme and the positive ideal value, Ri refers to the distance between the alternative scheme and the negative ideal value, and Wj refers to the global weight of the indicators by BWM. Step 5 Calculate Qi score.8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop Q\nolimits_{i} = \upsilon \left( {\frac{{\mathop S\nolimits_{i} - \mathop S\nolimits^{*} }}{{\mathop S\nolimits^{ - } - \mathop S\nolimits^{*} }}} \right) + \left( {1 - \upsilon } \right)\left( {\frac{{\mathop R\nolimits^{i} - \mathop R\nolimits^{*} }}{{\mathop R\nolimits^{ - } - \mathop R\nolimits^{*} }}} \right)$$\end{document}Qi=υSi-S∗S--S∗+1-υRi-R∗R--R∗where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop S\nolimits^{ - } = \mathop {Max}\nolimits_{i} \mathop S\nolimits_{i}$$\end{document}S-=MaxiSi, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop S\nolimits^{*} = \mathop {Min}\nolimits_{i} \mathop S\nolimits_{i}$$\end{document}S∗=MiniSi, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop R\nolimits^{ - } = \mathop {Max}\nolimits_{i} \mathop R\nolimits_{i}$$\end{document}R-=MaxiRi, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathop R\nolimits^{*} = \mathop {Min}\nolimits_{i} \mathop R\nolimits_{i}$$\end{document}R∗=MiniRi, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upsilon \in \left[ {0,1} \right]$$\end{document}υ∈0,1 is the decision mechanism coefficient, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upsilon > 0.5$$\end{document}υ>0.5 refers to decision making with the mechanism that maximizes group benefits, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upsilon < 0.5$$\end{document}υ<0.5 refers to decision making with the mechanism that minimizes individual regrets, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upsilon = 0.5$$\end{document}υ=0.5 refers to decision making with the compromise mechanism. Step 6 Assume the order of hospitals according to \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Q_{i}$$\end{document}Qi as \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$H_{1} ,H_{2} ,H_{3} , \ldots ,H_{n}$$\end{document}H1,H2,H3,…,Hn. If H1 satisfies both of the following conditions:9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{Condition 1:}}\quad Q\left( {H_{1} } \right) - Q\left( {H_{2} } \right) \ge {1 \mathord{\left/ {\vphantom {1 {\left( {n - 1} \right)}}} \right. \kern-0pt} {\left( {n - 1} \right)}}$$\end{document}Condition 1:QH1-QH2≥11n-1n-1where H1 is the alternative scheme ranking the second, and n is the number of total alternative schemes.10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{Condition 2:}}\;\;{\text{If}}\;\;S\left( {H^{1} } \right) = \mathop {Min}\limits_{j} S\left( {H^{j} } \right),\;\;R\left( {H^{1} } \right) = \mathop {Min}\limits_{j} R\left( {H^{j} } \right)$$\end{document}Condition 2:IfSH1=MinjSHj,RH1=MinjRHj H1 is the optimal scheme; if the above conditions are not satisfied, the corresponding compromise scheme can be obtained. Data analysis In this study, the MCDM method combining BWM with VIKOR was used to evaluate the SM service quality in three hospitals in Xiamen, Fujian Province, namely H1, H2 and H3. Xiamen is located on the southeast coast of China, and has a population of over 5 million. The living standard of its residents is ranked in the upper middle range of China, and its medical level is in the middle range of large and medium-sized cities in China. This gives Xiamen a certain degree of universality, so it can be used as a reference for promoting smart healthcare in China and even in other developing countries. These three hospitals are comprehensive public hospitals integrating medical treatment, teaching, and scientific research. They are committed to the construction of smart medicine in active response to the national and government medical informatization. In order to comprehensively assess the service quality of smart medicine, two questionnaires were designed. In the first questionnaire, the connotations of evaluation dimensions and indicators of the service quality of smart medicine were elaborated, and the filling method was introduced. A total of 10 valid expert questionnaires were collected over a period of three months from Dec. 2022 to Jan. 2023. The detailed backgrounds of these 10 experts are indicated in Table 5. The expert panel represents a comprehensive cross-section of the field, including professionals from various fields such as hospitals, academic institutions, and government agencies (including professionals from five hospitals, three academic institutions, and two government agencies). Each in-depth interview with the experts lasted about three hours, during which they were asked to use a structured questionnaire as a guide to evaluate the optimal and worst dimensions in their opinion, as well as the optimal and worst indicators under each dimension, in order to determine the relative importance of the dimensions and indicators. The comprehensive data collection process lasted for two months, ensuring thorough and in-depth exploration. This meticulous approach highlights the representativeness of the selected experts in smart healthcare.Table 5Background information of the ten experts.NoWork departmentJob titleEducational qualificationsWorking experience (years)1Public HospitalHospital directorPh.D252Public HospitalChief physician of internal medicineMaster’s203Public HospitalDeputy presidentPh.D224Public HospitalPhysician–patient management administratorMaster’s205Public HospitalDeputy presidentMaster’s186Technical departmentTechnical directorPh.D167Institute of Hospital ManagementProfessorPh.D228Smart Medical Research InstituteProfessorPh.D219Health Inspection BureauHospital business supervisorMaster’s1610Health BureauMedical quality supervisorMaster’s15 Table 5Background information of the ten experts.NoWork departmentJob titleEducational qualificationsWorking experience (years)1Public HospitalHospital directorPh.D252Public HospitalChief physician of internal medicineMaster’s203Public HospitalDeputy presidentPh.D224Public HospitalPhysician–patient management administratorMaster’s205Public HospitalDeputy presidentMaster’s186Technical departmentTechnical directorPh.D167Institute of Hospital ManagementProfessorPh.D228Smart Medical Research InstituteProfessorPh.D219Health Inspection BureauHospital business supervisorMaster’s1610Health BureauMedical quality supervisorMaster’s15 Background information of the ten experts. Patients who were seeking or who had recently sought medical advice in these hospitals were asked to fill in the second questionnaire. To ensure patient privacy was not compromised, the questionnaire did not include information about the patient's identity or condition, but only asked about the patient's satisfaction with the hospital based on these 29 indicators. The entire data collection process lasted nearly three months, and 152 valid questionnaires were collected. Calculation of dimension (indicator) weights Ten experts were expected to fully understand the specific connotations of each dimension and indicator, and were asked to select the best and worst dimensions from the six dimensions listed in Table 2; similarly, they were asked to select the best and worst indicators under each dimension. Table 6 lists the best and worst dimensions selected by the 10 experts. In terms of the best dimension, four of the experts chose smart appointment (D1) as the best, four chose smart diagnosis and treatment (D3), one chose smart consultation (D2), and one chose smart healthcare (D6). In terms of the selection of the worst dimension, five chose smart healthcare (D6), two chose smart diagnosis and treatment (D3), two chose smart settlement (D5), and one chose smart settlement (D4).Table 6The best and worst dimensions selected by the 10 experts.DimensionsExperts selecting the “best” dimensionExperts selecting the “worst” dimensionD11, 3, 4, 8D25D32, 7, 9, 104, 5D41D56, 7D662, 3, 8, 9, 10 Table 6The best and worst dimensions selected by the 10 experts.DimensionsExperts selecting the “best” dimensionExperts selecting the “worst” dimensionD11, 3, 4, 8D25D32, 7, 9, 104, 5D41D56, 7D662, 3, 8, 9, 10 The best and worst dimensions selected by the 10 experts. After selecting the best and worst dimensions and indicators, all experts should determine their preferences of the best and worst dimensions compared to other dimensions according to the language scale shown in Table 7. Their results are given in Tables 8 and 9.Table 7Language scale.Language variablesImportance levelLeast important1Relatively important2Strongly important3Very important4Extremely important5Table 8BO preference.Expert numberThe best dimensionD1D2D3D4D5D61D11539882D37513793D11654384D11387645D23154846D62457917D35314728D11353399D323156710D3421357Table 9OW preference.Expert numberThe worst dimensionD1D2D3D4D5D61D49571222D63597313D68345614D38612355D36814556D57642197D53475168D69757719D623567110D6467531 Table 7Language scale.Language variablesImportance levelLeast important1Relatively important2Strongly important3Very important4Extremely important5 Language scale. Table 8BO preference.Expert numberThe best dimensionD1D2D3D4D5D61D11539882D37513793D11654384D11387645D23154846D62457917D35314728D11353399D323156710D3421357 BO preference. Table 9OW preference.Expert numberThe worst dimensionD1D2D3D4D5D61D49571222D63597313D68345614D38612355D36814556D57642197D53475168D69757719D623567110D6467531 OW preference. After determining the preference rating of all dimensions and indicators, the weights of all dimensions and indicators were calculated using Eqs. (1) and (3), and their optimized weights and global weights are shown in Table 10.Table 10The weights of dimensions and indicators.DimensionsWeightIndicatorsLocal weightGlobal weightRankingSmart appointment (D1)0.265C110.2090.0554C120.4690.1251C130.2160.0573C140.1060.02817Smart consultation (D2)0.184C210.1680.03113C220.2110.0399C230.2060.03810C240.2840.0526C250.1310.02419Smart diagnosis and treatment (D3)0.208C310.3720.0782C320.2510.0525C330.1050.02222C340.1050.02223C350.1160.02418C360.0510.01127Smart nursing (D4)0.133C410.1700.02320C420.2280.03014C430.2440.03211C440.3080.0417C450.0500.00728Smart settlement (D5)0.095C510.1870.01825C520.3410.03212C530.1700.01626C540.2340.02221C550.0650.00629Smart healthcare (D6)0.116C610.2580.03015C620.1570.01824C630.3400.0398C640.2450.02816 Table 10The weights of dimensions and indicators.DimensionsWeightIndicatorsLocal weightGlobal weightRankingSmart appointment (D1)0.265C110.2090.0554C120.4690.1251C130.2160.0573C140.1060.02817Smart consultation (D2)0.184C210.1680.03113C220.2110.0399C230.2060.03810C240.2840.0526C250.1310.02419Smart diagnosis and treatment (D3)0.208C310.3720.0782C320.2510.0525C330.1050.02222C340.1050.02223C350.1160.02418C360.0510.01127Smart nursing (D4)0.133C410.1700.02320C420.2280.03014C430.2440.03211C440.3080.0417C450.0500.00728Smart settlement (D5)0.095C510.1870.01825C520.3410.03212C530.1700.01626C540.2340.02221C550.0650.00629Smart healthcare (D6)0.116C610.2580.03015C620.1570.01824C630.3400.0398C640.2450.02816 The weights of dimensions and indicators. In order to confirm the consistency of the expert questionnaires, the CR value of the questionnaire was calculated using Eq. (2), and the CR values of the 10 experts are shown in Table 11.Table 11CR values of the 10 expert questionnaires.ExpertsDimensionD1D2D3D4D5D6Expert 10.3310.2350.3310.3040.3040.3310.304Expert 20.3310.2410.2680.3310.3110.3110.253Expert 30.3110.4160.7670.2530.2680.3310.413Expert 40.3110.3590.3110.4160.3110.3110.253Expert 50.3110.4140.3080.3310.4070.3080.308Expert 60.3040.2530.3080.3110.2530.3110.338Expert 70.4130.8000.3380.3080.2680.4330.105Expert 80.3310.2680.2530.3110.3310.3110.311Expert 90.4130.2800.4330.2530.2800.3080.268Expert 100.3080.4350.2740.2800.2740.2800.200 Table 11CR values of the 10 expert questionnaires.ExpertsDimensionD1D2D3D4D5D6Expert 10.3310.2350.3310.3040.3040.3310.304Expert 20.3310.2410.2680.3310.3110.3110.253Expert 30.3110.4160.7670.2530.2680.3310.413Expert 40.3110.3590.3110.4160.3110.3110.253Expert 50.3110.4140.3080.3310.4070.3080.308Expert 60.3040.2530.3080.3110.2530.3110.338Expert 70.4130.8000.3380.3080.2680.4330.105Expert 80.3310.2680.2530.3110.3310.3110.311Expert 90.4130.2800.4330.2530.2800.3080.268Expert 100.3080.4350.2740.2800.2740.2800.200 CR values of the 10 expert questionnaires. As shown in Table 11, CR values of the 10 experts under overall dimension and other dimensions were 0–1, indicating good consistency. Ranking of alternative hospitals with the VIKOR method After obtaining the ranking of all weights and indicators, the VIKOR method was used to rank the alternative hospitals according to the weights of these dimensions and indicators. Equations (4)–(8) were used to calculate the Si, Ri and Qi values of the alternative hospitals, as shown in Table 12.Table 12Si, Ri and Qi values of the alternative hospitals.SiRiQiH10.6810.0780.676H20.2010.0520.000H30.6760.1250.995S− = 0.681R− = 0.125R− = 0.995S* = 0.201R* = 0.052Q* = 0.000 Table 12Si, Ri and Qi values of the alternative hospitals.SiRiQiH10.6810.0780.676H20.2010.0520.000H30.6760.1250.995S− = 0.681R− = 0.125R− = 0.995S* = 0.201R* = 0.052Q* = 0.000 Si, Ri and Qi values of the alternative hospitals. The hospitals were ranked according to the Si, Ri and Qi values, as shown in Table 13. The ranking of hospitals determined based on the Qi value was H2 > H1 > H3, where the hospital with the smallest Qi value was H2. This value can be used to determine whether the selected scheme meets Condition 1 and Condition 2 of Step 6 in “Research method” section. According to Eq. (8), \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$Q({H}_{2})-Q({H}_{1})=0.68>1/2$$\end{document}Q(H2)-Q(H1)=0.68>1/2, Condition 1 was met; H2 was the best selection based on Si and Ri, and it met Condition 2. Therefore, H2 was determined to be the best hospital during the evaluation of the SM service quality.Table 13Ranking of alternative schemes based on the Ri, Si and Qi values.SiRankingRiRankingQiRankingH10.68130.07820.6762H20.20110.05210.0001H30.67620.12530.9953 Table 13Ranking of alternative schemes based on the Ri, Si and Qi values.SiRankingRiRankingQiRankingH10.68130.07820.6762H20.20110.05210.0001H30.67620.12530.9953 Ranking of alternative schemes based on the Ri, Si and Qi values. Discussion Importance of the dimensions and indicators The weights of the dimensions and indicators for evaluating the service quality of smart medicine are shown in Table 10. The dimension weights and local weights of indicators are obtained by BWM analysis, and the global weights are obtained by multiplying the dimension weight by the local weight of each indicator. Table 8 indicates that the top three dimensions were smart appointment (D1, 0.265), smart diagnosis and treatment (D3, 0.208) and smart consultation (D2, 0.184). Smart appointment (D1, 0.265), with the highest weight, mainly involves real-name registration and archiving, Internet appointments, appointment reminders and intelligent navigation. Internet appointments (C12, 0.125) is the indicator with the highest weight in this dimension, and also the indicator with the highest weight among all indicators. It means that the making of an appointment is still the most anxiety-inducing issue for patients. Some studies have indicated that outpatient waiting time would affect the overall evaluation of the quality of medical services, and the long waiting time became one of the main factors resulting in patient dissatisfaction53. Hospitals are always in a state of overcrowding, and without an appointment, it would take a long time for patients to receive diagnosis and treatment. Internet appointments could help patients save waiting time, allowing them to receive a diagnosis and treatment within a short period of time after arriving at the hospital. Although there was no significant difference in overall satisfaction between outpatients making appointments and other patients, the outpatients making appointments were more satisfied with the physicians than other patients54. With the continuous development of informatization, the Internet has integrated into the medical field, and urban residents are more likely to make appointments via the Internet for non-emergency medical visits. To improve patient satisfaction, hospitals should establish their own Internet appointment platforms, improve the functions of those platforms, and ensure information security of the patients on the platforms. Within the period from appointment to arrival for consultation, hospitals should also make use of the improved platform functions to guide the patients to arrive on time. Hospitals may send appointment reminders to patients via the platform, or send SMS reminders to their cell phones one day or one hour in advance. At the same time, hospitals should indicate their specific locations, give full consideration of the modes of travel, and guide patients with platform-based map navigation and prominent signs in the hospitals. Smart diagnosis and treatment (D3, 0.208) ranked the second, and consisted of indicators such as doctor–patient communication, timely notification of test results, self-service appointment for surgery, self-service admission, self-service hospitalization, and self-service discharge. These evaluation indicators focus on the patients’ satisfaction with the services provided by the hospitals. Among the six indicators, doctor–patient communication (C31, 0.078) ranked the first, indicating that patients attach more importance to communication with their doctors. The overwork of doctors, the slow development of their humanistic diagnosis and treatment thinking, and the quality of medical and health services may induce doctor–patient conflicts. Among the factors constituting doctor–patient conflicts, the doctor–patient communication barrier is the main cause of doctor–patient conflicts55. It is suggested that a reasonable scale be established for assessing the quality of doctor–patient communication, which means that researchers should continue to conduct in-depth quantitative research on the affecting factors, quality, and evaluation of doctor–patient communication, as well as on the position of patients, influence, and communication skills during doctor–patient communication. Patients tend to select doctors who are willing to communicate with them, rather than silent doctors. Hospitals may strengthen the standard of the professional quality of doctors, conduct induction training, and build more humane hospitals. Smart consultation (D2, 0.184) ranked the third, and consisted of online waiting order (C21, 0.031), online calling inquiry (C22, 0.039), average waiting time (C23, 0.038), self-service registration (C24, 0.052) and triage calling (C25, 0.024). Self-service registration had the highest weight, indicating that hospital management should focus on convenient, fast, and accurate registration, and reasonable diversion of the peak registration crowd, which is also the key to the construction of smart medicine. Self-service registration, as an advanced method that can save the time cost and rationally allocate the resources, has greater appeal to patients. Self-service registration replaces manual operation, reduces cash transactions, prevents doctor–patient conflicts, and reduces the workload of outpatient registration staff, creating a win–win situation between hospital and patients56. Self-service registration has been developed to alleviate the current problem of “difficulty in seeking medical advice,” and it could realize reasonable triage, improve the medical environment, save time, relieve the pressure on the service counters, and improve the accuracy of registration information. Based on self-service, patients could have more choices of care. Self-service registration can shorten the non-medical time, and play an important role in relieving the pressure of queuing at manual service counters and extending the service hours of outpatient clinics. The hospitals should improve the self-service registration system, and strengthen the guidance of self-service registration, so as to improve the usage rate of the self-service registration system, and cultivate the habit of self-service registration, as it can improve patients’ satisfaction and save the labor costs in hospitals. In this paper, through summarizing the literature articles and expert questionnaires, a smart medicine service quality indicator system was initially established. With the general questionnaire, the service quality of three hospitals in Xiamen was investigated. The service performance of the three hospitals was ranked using the VIKOR method. The results of other researchers also showed that these three dimensions can significantly improve hospitals’ service quality and patients’ satisfaction. As suggested in this paper, hospitals should strengthen the construction of their appointment information platforms, strengthen the quality management of internal physician teams, and strengthen service information connection between hospital self-service terminals and information platforms, so as to improve the service quality of hospitals in the construction of smart medicine. In addition, smart medicine in China can also learn from the experience of developed countries. For example, in the United States, doctors are encouraged to issue electronic prescriptions to patients, digital medical records are shared among hospitals to improve diagnosis efficiency, and vital signs monitoring software is installed on mobile phones to track patients’ conditions51. Sensitivity analysis In order to verify the impact of indicator weights in the model, the sensitivity analysis method proposed by Gupta et al. was adopted in this study. The indicator with the highest weight (Internet appointment, C12) was selected from 29 indicators, the weight was adjusted from 0.1 to 0.9 (Table 14), and the weights of other indicators were also changed50.Table 14Changing of the weights of all indicators according to the changes of C12.IndicatorOverall weight123456789C110.0550.0570.0510.0440.0380.0320.0250.0190.0130.006C120.1250.1000.2000.3000.4000.5000.6000.7000.8000.900C130.0570.0590.0520.0460.0390.0330.0260.0200.0130.007C140.0280.0290.0260.0220.0190.0160.0130.0100.0060.003C210.0310.0320.0280.0250.0210.0180.0140.0110.0070.004C220.0390.0400.0350.0310.0270.0220.0180.0130.0090.004C230.0380.0390.0350.0300.0260.0220.0170.0130.0090.004C240.0520.0540.0480.0420.0360.0300.0240.0180.0120.006C250.0240.0250.0220.0190.0170.0140.0110.0080.0060.003C310.0780.0800.0710.0620.0530.0440.0350.0270.0180.009C320.0520.0540.0480.0420.0360.0300.0240.0180.0120.006C330.0220.0230.0200.0180.0150.0130.0100.0080.0050.003C340.0220.0220.0200.0170.0150.0120.0100.0070.0050.002C350.0240.0250.0220.0190.0170.0140.0110.0080.0060.003C360.0110.0110.0100.0080.0070.0060.0050.0040.0020.001C410.0230.0230.0210.0180.0150.0130.0100.0080.0050.003C420.0300.0310.0280.0240.0210.0170.0140.0100.0070.003C430.0320.0330.0300.0260.0220.0190.0150.0110.0070.004C440.0410.0420.0370.0330.0280.0230.0190.0140.0090.005C450.0070.0070.0060.0050.0050.0040.0030.0020.0020.001C510.0180.0180.0160.0140.0120.0100.0080.0060.0040.002C520.0320.0330.0290.0260.0220.0180.0150.0110.0070.004C530.0160.0160.0150.0130.0110.0090.0070.0050.0040.002C540.0220.0230.0200.0180.0150.0130.0100.0080.0050.003C550.0060.0060.0060.0050.0040.0040.0030.0020.0010.001C610.0300.0310.0270.0240.0200.0170.0140.0100.0070.003C620.0180.0190.0170.0150.0120.0100.0080.0060.0040.002C630.0390.0400.0360.0310.0270.0220.0180.0130.0090.004C640.0280.0290.0260.0230.0190.0160.0130. 0100.0060.003Total111111111 Table 14Changing of the weights of all indicators according to the changes of C12.IndicatorOverall weight123456789C110.0550.0570.0510.0440.0380.0320.0250.0190.0130.006C120.1250.1000.2000.3000.4000.5000.6000.7000.8000.900C130.0570.0590.0520.0460.0390.0330.0260.0200.0130.007C140.0280.0290.0260.0220.0190.0160.0130.0100.0060.003C210.0310.0320.0280.0250.0210.0180.0140.0110.0070.004C220.0390.0400.0350.0310.0270.0220.0180.0130.0090.004C230.0380.0390.0350.0300.0260.0220.0170.0130.0090.004C240.0520.0540.0480.0420.0360.0300.0240.0180.0120.006C250.0240.0250.0220.0190.0170.0140.0110.0080.0060.003C310.0780.0800.0710.0620.0530.0440.0350.0270.0180.009C320.0520.0540.0480.0420.0360.0300.0240.0180.0120.006C330.0220.0230.0200.0180.0150.0130.0100.0080.0050.003C340.0220.0220.0200.0170.0150.0120.0100.0070.0050.002C350.0240.0250.0220.0190.0170.0140.0110.0080.0060.003C360.0110.0110.0100.0080.0070.0060.0050.0040.0020.001C410.0230.0230.0210.0180.0150.0130.0100.0080.0050.003C420.0300.0310.0280.0240.0210.0170.0140.0100.0070.003C430.0320.0330.0300.0260.0220.0190.0150.0110.0070.004C440.0410.0420.0370.0330.0280.0230.0190.0140.0090.005C450.0070.0070.0060.0050.0050.0040.0030.0020.0020.001C510.0180.0180.0160.0140.0120.0100.0080.0060.0040.002C520.0320.0330.0290.0260.0220.0180.0150.0110.0070.004C530.0160.0160.0150.0130.0110.0090.0070.0050.0040.002C540.0220.0230.0200.0180.0150.0130.0100.0080.0050.003C550.0060.0060.0060.0050.0040.0040.0030.0020.0010.001C610.0300.0310.0270.0240.0200.0170.0140.0100.0070.003C620.0180.0190.0170.0150.0120.0100.0080.0060.0040.002C630.0390.0400.0360.0310.0270.0220.0180.0130.0090.004C640.0280.0290.0260.0230.0190.0160.0130. 0100.0060.003Total111111111 Changing of the weights of all indicators according to the changes of C12. The performance of these three alternative hospitals was ranked nine times with VIKOR, and the rankings were compared, as shown in Table 15.Table 15Sensitivity analysis of the three hospitals after nine rounds of ranking.HospitalRanking123456789H12222222222H21111111111H33333333333 Table 15Sensitivity analysis of the three hospitals after nine rounds of ranking.HospitalRanking123456789H12222222222H21111111111H33333333333 Sensitivity analysis of the three hospitals after nine rounds of ranking. The sensitivity results show that the model proposed in this paper has strong stability. Conclusion In this study, 29 indicators in six dimensions were proposed to evaluate the service quality of smart medicine, namely smart appointments, smart consultations, smart diagnosis and treatment, smart nursing, smart settlement, and smart healthcare. In order to construct a smart medicine service quality evaluation system, a mixed MCDM model was proposed, and the BWM method was used to calculate the weights of all the indicators; then, the VIKOR method was introduced to rank the service quality of the three hospitals. There are three theoretical contributions of this study. First, we constructed a system for evaluating smart medicine, including six dimensions: smart appointments, smart consultation, smart diagnosis and treatment, smart nursing, smart settlement and smart healthcare, and there were 29 criteria under these dimensions. Second, a hybrid model which combined the best and worst method (BWM) with TOPSIS was used to determine the weights of the dimensions and criteria for the service quality of smart hospitals. Third, a sensitivity analysis was applied to verify the robustness and stability of the approach. The findings of this study have implications for practitioners to achieve high service quality. First, hospitals can apply intelligent service software and hardware such as intelligent wearable devices, artificial intelligence-assisted diagnosis and treatment systems, and so on, to improve the intelligence and personalization of medical services. Second, medical institutions can provide online services such as intelligent guided triage, mobile payment, in-hospital navigation, inspection and test results, and mutual recognition of inspection and test results, and actively promote functions such as referral services, telemedicine, and drug delivery. The combination of online and offline services provides more systematic, safe and continuous medical services, and further improves the quality of medical services. Third, medical institutions can use Internet technology to continuously optimize the medical service process, provide patients with convenient services such as appointments for diagnosis and treatment, reminders, results inquiries, and information pushes. In this way, the medical efficiency can be improved and patients’ medical experience can be optimized. Lastly, taking patient experience and satisfaction as the core of medical service quality evaluation emphasizes the patient-centered medical service concept and promotes continuous improvement in service quality. This approach enhances patient engagement, improves the transparency and trust of medical institutions, and creates competitive advantages. It also helps optimize performance management and resource allocation, and reduces medical disputes and risks, thereby promoting the overall development and market competitiveness of medical institutions. In summary, hospitals implementing smart medical services face challenges such as technology integration, data security, user acceptance, regulatory compliance, cost control, and medical quality. Modular technology, strengthening data security measures, providing training and user education, collaborating with regulatory agencies to ensure compliance, exploring cooperation models to reduce costs, and establishing strict quality control processes can help hospitals overcome obstacles and effectively utilize smart medical technology to improve service efficiency and quality. There are certain limitations in this study. The hospitals are located in Xiamen, and the questionnaires were collected from the residents of Xiamen, but the results of service quality may change by region; therefore, the study results have some geographical limitations. Future studies should be comparative studies in two different regions with the same service quality evaluation indicators to verify the findings. Future studies can be conducted based on private hospitals, or can be performed to compare private with public hospitals. There is also a problem regarding the sample size. Since the MCDM method, which does not require a large sample size, was adopted in this study, longitudinal studies with a larger sample size can be conducted in the future to validate the results of this study. In addition, since there is inherent ambiguity in the dimensions and indicators finalized in this study, the fuzzy BWM and VIKOR may generate better results of ranking of these dimensions and indicators. How to make smart medical care benefit patients at the lower levels of society in China, or the elderly, or those who do not want to rely on technology, who cannot afford medical insurance, or who have less access to smart medical care is an urgent consideration for the development of smart medicine in China.