Adoption intention and influencing factors of users of intelligent diagnosis and treatment system: a study based on questionnaire data
TianzeGao1✉Phone(Phone: +86 18226587711Emailg18226587711@163.com
1School of ManagementNanjing University of Posts and TelecommunicationsNanjingJiangsu ProvinceChina
ABSTRACT
Background
The application of artificial intelligence technology in the medical industry has achieved significant phased results. As an important technological achievement of smart healthcare, intelligent diagnosis and treatment systems have become a key carrier for innovating and improving the quality and efficiency of medical and health services. The intention of users to adopt them is a key factor affecting the widespread use of intelligent diagnosis and treatment systems.
Objective
This study aims to investigate the factors influencing users’ adoption of intelligent diagnosis systems among patients and healthcare professionals.
Methods
Based on extended TAM, UTAUT, and TTF models, we construct targeted questionnaires, and the hypothesized pathways constructed in the study are analyzed and tested using software SPSS 27 and Smartpls 4, employing structural equation modeling (SEM) with bootstrapping (5,000 samples) to test hypotheses.
Results
A total of 430 valid questionnaires are collected. It is found that adoption intention is influenced by different factors, and the adoption drivers differ significantly between groups. For patients, performance expectancy (β = .273, p < .001), health literacy (β = .235, p = .001), and technology trust (β = .129, p = .01) significantly predict the adoption. For medical professionals, digital literacy (β = .119, p = .03), performance expectancy (β = .145, p = .001), and facility condition (β = .179, p = .001) are key drivers. Privacy concern negatively impacts patients (β = .205, p < .001), while human-machine collaboration shows no significant effect on professionals (p > .05).
Conclusions
Based on the above findings, the study provides tailored strategies guidance for developers of intelligent diagnosis and treatment systems, medical institutions, and policy makers, such as enhancing diagnostic accuracy and privacy for patients, and improving workflow integration for professionals, and points out future research directions.
Keywords:
Intelligent diagnosis and treatment system
Intention to adopt
TAM
UTAUT
TTF
SEM
1 Introduction
The application of artificial intelligence technology in the medical field has achieved remarkable phased results. As an important technological achievement of smart healthcare, intelligent diagnosis and treatment systems have become a key carrier for innovation to improve the quality and efficiency of medical and health services.
The intelligent diagnosis and treatment system integrates emerging technologies such as big data analysis, artificial intelligence and cloud computing, which can provide doctors with accurate diagnostic support and treatment suggestions, and provide patients with personalized medical services. Although the technical advantages and development space of intelligent diagnosis and treatment systems are significant, users’ intention to adopt intelligent diagnosis and treatment systems has not increased rapidly as scheduled. The intention to adopt intelligent diagnosis and treatment systems has become one of the biggest obstacles restricting the large-scale promotion and application of intelligent diagnosis and treatment systems. On the one hand, patients’ trust in intelligent diagnosis and treatment systems is uneven; on the other hand, doctors and researchers face many obstacles in using the system. Existing research suggests that users’ adoption of intelligent diagnosis and treatment systems is affected by many factors including system ease of use, usefulness, technical trust, and environmental support[1]. Especially in the medical industry, users’ intention to adopt is directly related to the promotion effect and practical application of the system. Therefore, it is of great theoretical and practical significance to explore the user’s intention to adopt the intelligent diagnosis and treatment system and its influencing factors.
2 Background
2.1 Development status of intelligent medical care in the context of digital intelligence
With the continuous application of digital intelligence technology in massive scenarios, the traditional industrial pattern is undergoing a profound and broad reshaping and transformation, and digital intelligence technology has become a key engine to promote industrial intelligence and the development of new quality productivity [2], and the high penetration of digital intelligence technology has triggered all-round changes in social innovation, and the form of social organization is constantly transforming to platform, sharing, and online [3]. As a product of the deep integration of information technology and medical and health services, intelligent medical care is greatly changing the pattern of the traditional medical industry, improving the quality of medical services, reducing the burden of medical care, and enhancing people’s health and well-being. Intelligent medical care refers to a new medical service model that uses artificial intelligence technology as a tool to provide systematic and precise medical services based on big data to realize intelligent, digital, and refined medical services in the fields of medical services, health management, and public health [4], with broad market space and good development prospects [5].
In this context, multiple application scenarios of intelligent medical care are constantly expanding, and Xiao Ma et al.[6] proposed a three-layer intelligent diagnosis and treatment system architecture to effectively manage and analyze medical data and provide intelligent medical services and applications by integrating key technologies. Wang et al. [7] analyzed the role of high-quality development of smart health care data in promoting the development of elderly care in our country, proposed three types of health data classification methods, and expounded the basic characteristics of smart health care data. In terms of model framework, Yu Haitao et al. [8] constructed an interpretable intelligent diagnosis theoretical model framework from the perspective of knowledge data two-wheel drive, and proposed to improve the interpretability of the model from three aspects: visualization of model results, transparency of operation logic, and knowledge of index selection. Chu et al. [9] pointed out that the AIGC-enabled smart health knowledge service model is a new knowledge service model, and designed and constructed the framework and model of the AIGC-enabled smart health knowledge service platform. In response to technology integration, Zhao et al. [10] emphasized that embodied intelligence combines artificial intelligence with physical forms, integrating sensors, machine learning, and natural language processing technologies, improving the professionalism and personalization of medical services, and further promoting the innovation and application of intelligent medical technology. Gu et al. [11] took WiNGPT as the research object to construct a multi-source knowledge system and a whole-process risk digital governance framework for large medical and health models, which provided new perspectives for ensuring the safe and effective application of large models in the medical and health field.
In addition, the research of Zhang[12], Wang[13], and Chen[14] provides theoretical support for the development of intelligent medicine from different perspectives. Zhang et al. [12] proposed a graph-driven online smart Q&A service process based on the theoretical characteristics of medical wisdom Q&A, which realized the organic combination of smart Q&A service theory and technology, and contributed new ideas and methods for the intelligent transformation of medical and health Q&A. Wang [13] pointed out three core themes and three modules covered by the knowledge base of existing medical AI research from the perspective of ethical governance. From the perspective of linguistics, Chen et al. [14] compared the real reply text of doctors in online medical scenarios with the text generated by ChatGPT, indicating that AIGC has certain language understanding and situational reasoning capabilities, and can provide certain decision support, but it still needs to be improved in terms of accuracy, coherence, and readability in language expression.
The rapid development of digital intelligence technology is profoundly changing the face of the medical industry, promoting it in the direction of intelligence, precision and efficiency. A study by Mullai Murugan MS et al. [15] showed that GPT-4 has great potential in enhancing healthcare provider support and patient access to complex pharmacogenomic information by using retrieval-augmented generation methods, combined with clinical pharmacogenetics implementation consortium data, to generate customized responses.
2.2 Reasearch status of the users’ adopt intention of intelligent diagnosis and treatment
In recent years, researchers have explored the factors affecting users’ intention to adopt from multiple dimensions, providing a theoretical basis for the optimization of intelligent diagnosis and treatment platforms[16,17]. Su et al. [16] constructed a hierarchical diagnosis and treatment simulation model based on multiple agents, accurately depicting patient behavior, simulating the actual system, and proposing four targeted strategies to effectively guide patients to seek medical treatment reasonably.
When discussing the influencing factors of user adoption intention, Liao et al. [18] constructed a research model of user satisfaction and adoption intention of smart elderly care services from the perspective of perceived quality, and regarded interaction quality and result quality as the two core factors affecting users’ satisfaction and adoption intention. Wu et al. [19] studied the influencing factors of digital health APP users’ continuous adoption intention from the perspective of perceived value, and proposed three effective configuration paths, which provided a reference for APP improvement. Zhu et al. [20] used meta-analysis methods to construct a comprehensive model of mobile health users' intention to adopt, and found that factors such as perceived usefulness and attitude have a significant positive impact on adoption intention. In addition, Chen et al. [21] analyzed the influencing factors of health information adoption behavior of mobile UGC community users based on the information adoption model and planned behavior theory, enriching the relevant research. Zhu et al.[22] reviewed 61 empirical research papers on mHealth user adoption behavior published from 2010 to 2019, and proposed that based on classical information technology adoption models such as the technology acceptance model, the influencing factors can be classified from four aspects: individual, technology, environment, and service content, which provides a reference for subsequent research and promotion practice.
At the same time, Mo et al. [23] found that factors such as platform ease of use and doctors' professionalism have a positive impact on adoption intention through perceived usefulness. Han et al. [24] explored the influence of online medical review information characteristics on users' intention to adopt based on the fine processing possibility model and trust transfer theory, and pointed out that specific and detailed review content can improve users' intention to adopt. Zheng [25] discussed the particularity of artificial intelligence in medical damage liability and the special challenges and coping strategies in the determination of causality, which provided important theoretical support for the legal and ethical nature of intelligent diagnosis and treatment systems.
2.3 Summary
With the rapid development of digital intelligence technology, the development of the intelligent medical industry is changing with each passing day. Technology has not only improved the level and efficiency of medical services, but also raised concerns about data and privacy security, ethical security and other aspects in the field of scientific research. Users' intention to accept intelligent diagnosis and treatment systems is affected by the quality of system interaction, perceived usefulness, and perceived ease of use, and these influencing factors are directly related to the interpretability and transparency of technology. Existing policies create a favorable environment for the development of the industry by promoting the construction of medical informatization and optimizing payment policies, but the development of the intelligent medical industry also faces the problem of trade-offs between technology and ethics.
Existing studies have made some progress in technology optimization, improving users’ adoption intention and ethical governance, but there are still certain limitations. In summary, this study takes the user's intention and influencing factors of intelligent diagnosis and treatment system as the research object, explores the psychological and behavioral motivation of users, reveals the internal mechanism of different types of users' adoption intention, and proposes corresponding optimization strategies, so as to provide theoretical support and practical guidance for the improvement and promotion of intelligent diagnosis and treatment system.
3 Methods
3.1 Theoretical basis
Based on the Technology Acceptance Model (TAM), Unified Theory of Acceptance and Use of Technology (UTAUT) and Task-Technology Fit (TTF), this study constructed a theoretical framework suitable for the study of users' intention to adopt and the influencing factors of intelligent diagnosis and treatment systems.
3.1.1 TAM
TAM was proposed by Davis in 1989 to study users' intention to accept information systems using rational behavior theory. TAM is one of the mainstream theories in the field of information systems, which is used to explain and predict user acceptance and use of new technologies. The core idea of TAM is that users' use of technology is mainly influenced by two key factors: perceived usefulness and perceived ease of use, which refers to the degree to which users subjectively believe that using a particular technology can improve their work performance, and perceived ease of use refers to the subjective perception of the difficulty of using a particular system[26].
In the scenario of the intelligent diagnosis and treatment system, patients will be enhanced when they believe that the intelligent diagnosis and treatment system can provide correct diagnostic advice, personalized health management plans, and improve their condition. Patients will think that the system is simple and easy to use when the system interface is user-friendly, the process is simple, the interaction logic is clear, and there are fewer problems encountered during use. Similarly, when the intelligent diagnosis and treatment system can help medical staff quickly and accurately access medical records, provide correct treatment suggestions, and assist them in making clinical decisions, medical staff will find the system very helpful. When the system can seamlessly connect with the existing workflow and information system of the hospital, is simple and easy to operate, and has low learning cost, it will increase their perceived ease of use evaluation. The limitation of TAM is that it does not fully consider the moderating effect of external environmental factors and individual user differences on the technology acceptance process. Therefore, the study introduced UTAUT to more comprehensively explain the users’ intention to adopt the intelligent diagnosis and treatment system on the basis of using TAM to provide a framework for users' initial acceptance behavior of the intelligent diagnosis and treatment system.
3.1.2 UTAUT
UTAUT was proposed by Venkatesh et al. in 2003 to combine the advantages of previous technology acceptance models to more comprehensively explain users' technology acceptance behavior, including four core elements: performance expectations, effort expectations, social impact, and convenience conditions.
Performance expectations are similar to perceived usefulness in TAM, which refers to the user's expectation of improving job performance through the use of technology, but emphasizes the direct contribution of technology to individual performance. Effort expectation is related to the perceived ease of use in TAM, which refers to the user's expectation of the effort required to use the technology, but further expands the connotation of ease of use. Social impact refers to the subjective influence of the user's social environment on the use of technology. Convenience conditions refer to the objective conditions such as technical support and resource guarantee faced by users when using technology. UTAUT believes that these four factors together affect users' behavioral intention and actual usage behavior [1,27].
The performance expectation in the context of the intelligent diagnosis and treatment system refers to the patient's expectation of the accuracy of the system's diagnosis and the expectation of the medical staff for the effectiveness of the system's assisted clinical decision-making. The effort expectation is reflected in the patient's ease of learning the system operation process and the ease of medical staff integrating the system with the existing workflow. Social impact refers to the recommendation of the leaders of the medical institution and the sharing of the patient's relatives and friends on the use of the system. Convenience conditions include whether the hardware equipment required for system operation is complete, whether the network is stable, and other factors. UTAUT integrates several different theoretical perspectives, breaks through the limitations of TAM and provides a solution for a further understanding of the adoption behavior of users of intelligent diagnosis and treatment systems.
3.1.3 TTF
TTF was proposed by Goodhue and Thompson in 1995, and the basic idea of TTF is that the value of technology to users is determined by the matching degree between technical attributes and task needs, and if users think that technology can meet their needs, it will increase their usage expectations and generate intention to use[28].
In the field of intelligent diagnosis and treatment systems, for patients, when the functions provided by the system match the needs of their health management tasks, patients think that the system matches the tasks and will be more accepting of the system. For medical staff, when the functions provided by the system are more in line with the clinical diagnosis and treatment tasks of medical staff, and the system is integrated with the existing medical workflow, they will also be inclined to accept the system. TTF provides a theoretical framework for measuring the match between the intelligent diagnosis and treatment system and the users’ task, which is helpful to understand the cognitive process of the users’ acceptance of the system. In this study, TTF was introduced to supplement TAM and UTAUT from the perspective of task matching, and the users’ adoption intention and influence mechanism of the intelligent diagnosis and treatment system are comprehensively analyzed.
3.2 Hypothesis mechanism
In view of the characteristics of the intelligent diagnosis and treatment system, the study modified and expanded the original variables. Based on TAM and UTATUT, the factors that may affect users' intention to adopt (BI) are divided into health literacy variables and technology perception variables of the patient module, work integration variables of the medical staff module, and data algorithm variables of the researcher module based on the perspectives of patients, medical staff, and scientific researchers, respectively. Based on UTATUT, considering the data-sensitive characteristics of the medical field, a risk perception variable is added to the patient module to represent the impact of users' risk perception on personal medical information leakage. On the basis of TTF, considering the possible misdiagnosis and misuse of AI systems, professional identity variables and scientific research scenario variables are added to the medical staff module and researcher module respectively to reveal the matching and adoption mechanism of users' diagnosis accuracy and ease of use of the system.
3.3 Hypotheses construction
3.3.1 Common hypotheses based on UTAUT
In this study, the performance expectation (PE), effort expectation (EE), social impact (SI), and facility condition (FC) variables in the UTAUT model were used.Previous studies have shown that these four factors positively affect users' intention to act, and Zhang et al.[1]have shown that doctors' performance expectations and effort expectations positively affect their intention to adopt intelligent diagnosis and treatment systems. Zhang et al.[26]found that the level of communication convenience positively affects patients' perception of the usefulness of Internet medical care. Since the three types of user modules in this study all contained the above four types of variables, three common hypotheses were proposed for the three types of modules in this study.
H1a: Performance expectation (PE) positively affects users’ adopt intention (BI).
H1b: Effort Expectation (EE) positively affects users’ adopt intention (BI).
H1c: Social impact (SI) positively affects users’ adopt intention (BI).
H1d: Facility conditions (FC) positively affects users’ adopt intention(BI).
3.3.2 Hypotheses of personal literacy factors
In this study, health literacy (HK) and digital literacy (DL) variables that depend on users' own factors were respectively proposed for patients and healthcare workers. A more complete and convenient system may increase patients’ confidence in operating the system and reduce the difficulty of operation, which in turn leads to increased patients’ intention to adopt the system. Similarly, when medical staff’s digital skills improve, the learning cost required for their operating systems may decrease, leading to increased adoption of the system by medical staff[18,19]. Among them, patients' health literacy (HK) plays a mediating role in the influence path of facility condition (FC) on users’ adopt intention (BI), and effort expectation (EE) plays a mediating role in the influence path of digital literacy (DL) on users’ adopt intention (BI). Based on this, the following hypotheses were proposed in this study.
H2a: Health literacy (HK) positively affects patients' adopt intention(BI).
H2b: Facility condition (FC) positively affects patients' health literacy (HK).
H3a: Digital literacy (DL) positively affects medical staff’s adopt intention (BI).
H3b: Digital literacy (DL) positively affecst medical staff’s effort expectation (EE).
3.3.3 Hypotheses in medical scenario
In medical scenario, privacy protection is a key factor affecting users' adopt intention. In order to measure the patient group's concerns about the risk of medical information leakage, reflect the concerns of patients and medical staff about the effectiveness of technical safeguards, and emphasize the fit between system functions and users’ needs, the study added privacy concern (PC), technical trust (TT) and task-technology fit (TTF) variables of the patient module, as well as technical trust (TT) and human-machine collaboration (HC) variables of the medical staff module. When the system pays more attention to privacy protection, adopts more advanced technical models, and improves the accuracy of system diagnosis to match the specific needs of users, it may lead to an increase in users' favorability, which in turn may increase their expectations for the system and increase their intention to adopt the system[27,29]. Among them, patients' performance expectation (PE) play a mediating role in the influence path of technical trust (TT) on user adoption intention (BI), and performance expectation (PE) plays a mediating role in the influence path of task-technology fit (TTF) on patients’ adopt intention (BI). Based on this, the following hypotheses were proposed in this study.
H4a: Privacy concern (PC) negatively affects patients' adopt intention (BI).
H4b: Technical trust (TT) positively affects patients' adopt intention (BI).
H4c: Technical trust (TT) positively affects patients' performance expectation (PE).
H4d: Task-technology fit (TTF) positively affects patients' adopt intention (BI).
H4e: Task-technology fit (TTF) positively affects patients' performance expectation (PE).
H5a: Technical trust (TT) positively affects medical staff’s adopt intention (BI).
H5b: Human-machine collaboration (HC) positively affects medical staff’s adopt intention (BI).
3.3.4 Hypotheses in scientific scenario
In scientific scenario, researchers are most concerned about data-driven scientific research efficiency, algorithm transparency, and credibility of results, and their intention to adopt them is highly dependent on scientific research ethics and the credibility of academic achievements. As the core means of production, the quality and operability of data will directly affect the perceived value of the system by researchers, and algorithm transparency is used as a measure to reduce the difficulty of technology use, and the role of perceived ease of use in TAM on adoption intention is strengthened[30]. Based on this, the following hypotheses were proposed in this study.
H6a: Data availability (DA) positively affects researchers’ performance expectation (PE) .
H6b: Algorithm transparency (AT) positively affects researchers' effort expectation (EE).
H6c: Research compliance (RC) positively affects researchers’ adopt intention (BI).
H6d: Data bias (DB) positively affects researchers' adopt intention (BI).
By integrating the cognitive evaluation mechanism of TAM and the situational regulation mechanism of UTAUT, and combining the TTF to concretize performance expectations, this study continued the explanation of technical acceptance behavior in previous studies, and meets the special needs of medical situations through the expansion of variables, providing theoretical support for further empirical research.
3.4 Research model
In this study, TAM, UTAUT and TTF are organically combined to construct a theoretical framework suitable for the study of users' adopt intention and affecting factors of intelligent diagnosis and treatment systems. By integrating the perceived usefulness and perceived ease of use in TAM, the performance expectation, effort expectation, social impact and facility condition in UTAUT, and the task-technology fit concept in TTF, this study analyzed the intention and influence mechanism of different user groups to adopt intelligent diagnosis and treatment systems, which provides a solid theoretical support for in-depth revelation of the complex logic behind the adoption behavior of users of intelligent diagnosis and treatment systems. Based on this, the study constructed a research model of the influencing factors of users' intention to adopt the intelligent diagnosis and treatment system as shown in Fig. 1.
Fig. 1
Theoretical model diagram.
3.5 Research design and data collection
3.5.1 Variable definition and measurement
Based on the research hypothesis of 3.2, this study designed the questionnaire around 15 factor type variables and user adoption intention variables in three types of user modules, see Additional File 1. The involved questions were designed using a 5-point Likert scale, with each option corresponding to "very inconsistent" to "very consistent", and assigned a score of 1 to 5 respectively to measure the degree of consistency between the expression of the item and the respondents' feelings.
3.5.2 Questionnaire design
The questionnaire designed by this study consisted of four parts, and the specific structure is as follows:
The first part is the title and introduction of the questionnaire, introducing the content and purpose of the survey to the respondents, and introducing the estimated response time and privacy confidentiality of the questionnaire.
The second part is the identity confirmation part of the respondents, which first classifies the identity of the respondents and collects their frequency and preliminary intention to use the intelligent diagnosis and treatment system. Polygraph questions are added to this part to test the authenticity of the respondents' questionnaires, improve the credibility of the questionnaire results, and avoid duplicate surveys caused by untrue answers.
The third part is the measurement of core influencing factors, based on the identification of respondents in the second part, which is divided into three modules: patients, medical staff and scientific researchers. Each respondent answers different questions of the module, collects the influencing factors and other relevant information that may affect the intention to adopt the system, and finally uses a scale to count the respondents' intention to adopt the intelligent diagnosis and treatment system.
The fourth part is demographic information, which collects basic information such as gender, age and education level of respondents and expresses gratitude to the respondents.
3.5.3 Data collection
This study used the Questionnaire Star platform to implement the distribution and collection of questionnaires. In order to ensure the quality of the questionnaire and the representativeness of the sample, this study chose WeChat, QQ and other platforms as the main channels to expand the coverage through the social circle of researchers and some respondents, including medical students and medical industry practitioners, and set up an open link to the Questionnaire Star platform to obtain random traffic samples.
The questionnaire was collected from April to May 2025, and a total of 497 questionnaires were collected, and 67 invalid questionnaires were eliminated based on the questionnaire lie detection questions and other indicators, and finally 430 valid questionnaires were retained, with an effective recovery rate of 86.52%. The questionnaire samples were distributed in 21 provincial-level administrative regions across China, and the distribution of urban and rural areas was relatively balanced, providing a diversified sample for subsequent analysis.
3.6 Descriptive statistical analysis
Based on the 430 valid questionnaires collected, see Additional File 2, the basic information statistics are made as Table S1.The proportion of male and female genders in the sample is roughly the same, and the gender structure is relatively balanced. The age distribution of the samples shows a clear trend of getting younger: The proportion of the population under 25 years old is the highest, followed by the middle-aged and young groups aged 40–55 and 25–40, while the population over 55 years old only accounts for 2.33%. This trend indirectly confirms the phenomenon of the gap between technological acceptance and age, that is, the elderly may be reluctant to try intelligent diagnosis and treatment systems due to fear of difficulties or physical inconvenience. The relatively high distribution of the sample among the 40–55 age group might be due to the demand of middle-aged people for the prevention and treatment of chronic diseases, which indirectly reflects the potential value of intelligent diagnosis and treatment systems in the middle-aged population. The sample population of the survey is mainly composed of people with higher education. Nearly 70% of them have a bachelor's degree or above, and 18.6% have a master's degree or above. To some extent, this phenomenon is related to the social networks of researchers and respondents, but it may also reflect that there may be certain usage thresholds for intelligent diagnosis and treatment systems. Complex professional terms and operational procedures may pose obstacles for users with lower educational attainment. The stronger trust and desire to explore technology among those with higher education may also be the internal driving force for them to become the main user group.
Statistics | Category | Number | Rate |
|---|---|---|---|
Gender | Male | 209 | 48.60% |
Female | 221 | 51.40% | |
Age(years) | Under 25 | 183 | 42.56% |
25–40 | 115 | 26.74% | |
40–55 | 122 | 28.37% | |
Over 55 | 10 | 2.33% | |
Educational level | Junior high school and below | 14 | 3.26% |
High school | 38 | 8.84% | |
Bachelor | 298 | 69.30% | |
Master | 67 | 15.58% | |
Doctor and above | 13 | 3.02% | |
Having used intelligent diagnosis and treatment system | Yes | 236 | 54.88% |
No | 194 | 45.12% | |
Willing to use the intelligent diagnosis and treatment system | Yes | 388 | 90.23% |
No | 42 | 9.77% | |
Identity using the intelligent diagnosis and treatment system | Patient | 208 | 48.37% |
Medical Staff | 218 | 50.70% | |
Scientific reseacher | 4 | 0.93% |
More than half of the respondents said they had come into contact with intelligent diagnosis and treatment systems, while nearly half said they had never used such systems. 90.23% of the respondents are willing to use the intelligent diagnosis and treatment system, while 9.77% of the respondents hold a negative attitude towards it. The gap between high usage intention and low usage rate indicates that users may be concerned about issues such as the mismatch between system functions and their own needs, privacy leakage, or complex operation. In conclusion, the respondents of this study tended to be younger and more highly educated. The core functions of the intelligent diagnosis and treatment system mainly serve patients and medical staff. Although the social acceptance is relatively high, the gap between the actual usage rate and the intention, as well as the proportion of low-frequency users, all indicate that there is considerable room for popularization and functional matching of the system.
4 Results
4.1 Reliability analysis
After the collection and preprocessing of questionnaire data, this study used SPSS 27 to conduct reliability analysis on the 209 collected questionnaire data. Due to the small sample size of the questionnaire for scientific researchers, only the samples of patients and medical staff were analyzed. The results are shown in Tables S2 and S3. The test results show that the Cronbach coefficients of all variables are greater than 0.75, and the overall Cronbach coefficient of the questionnaire is greater than 0.9. Moreover, the Cronbach coefficients of each variable after deleting the item combinations are all lower than the original values, indicating that the overall reliability of the questionnaire is good and further validity analysis can be conducted.
Variable | Coding | Cronbach’s α | Number |
|---|---|---|---|
Health literacy | HK | .895 | 4 |
Performance expectation | PE | .910 | 5 |
Effort expectation | EE | .895 | 4 |
Social impact | SI | .870 | 3 |
Facility condition | FC | .881 | 4 |
Privacy concern | PC | .791 | 4 |
Technical trust | TT | .768 | 3 |
Task-technology fit | TTF | .872 | 3 |
Adopt intention | BI | .924 | 6 |
Total | .934 | 32 |
Variable | Coding | Cronbach’s α | Number |
|---|---|---|---|
Digital literacy | DL | .886 | 4 |
Performance expectation | PE | .859 | 3 |
Effort expectation | EE | .844 | 3 |
Social impact | SI | .835 | 3 |
Facility condition | FC | .845 | 3 |
Technical trust | TT | .830 | 3 |
Human-machine collaboration | HC | .869 | 4 |
Adopt intention | BI | .917 | 6 |
Total | .951 | 29 |
4.2 Validity analysis and hypothesis testing
4.2.1 Validity analysis
Since the variable dimension division proposed in this study mainly came from the mature theoretical models commonly used in the academic community, and based on the research path proposed in Section 3.3, this study used Smartpls 4 to construct the structural equation model, and uses the Bootstrapping and PLS-SEM algorithms to conduct validity analysis on each variable dimension respectively.
As shown in Table 1, the combined validity (CR) of each variable in the scale is all greater than 0.8, with a minimum value of 0.898, indicating that the scale has good internal consistency. According to Table 2 and Figs. 2 and 3, it can be known that the AVE values of all variables are above 0.7, and the external loads of all variables are greater than 0.75. The discriminant validity of the scales in the patient and medical staff modules of this study is good.
Variable (patients) | CR | AVE | Variable (medical staff) | CR | AVE |
|---|---|---|---|---|---|
BI | .94 | .723 | BI | .935 | .706 |
EE | .927 | .761 | DL | .921 | .746 |
FC | .918 | .738 | EE | .906 | .762 |
HK | .927 | .76 | FC | .907 | .765 |
PC | .925 | .754 | HC | .911 | .719 |
PE | .933 | .735 | PE | .914 | .78 |
SI | .92 | .794 | SI | .901 | .753 |
TT | .919 | .792 | TT | .898 | .746 |
TTF | .922 | .798 |
Fig. 2
Operation result of patient model.
Fig. 3
Operation result of medical staff model.
Meanwhile, the data in Tables 2 and 3 show that the square root of the AVE value of each variable is greater than the correlation coefficient between the other variables, indicating that the colinearity between the variables in the scale is small[31], and it has good validity for discrimination.
BI | EE | FC | HK | PC | PE | SI | TT | |
|---|---|---|---|---|---|---|---|---|
BI | .851 | |||||||
EE | .805 | .872 | ||||||
FC | .799 | .798 | .859 | |||||
HK | .827 | .611 | .807 | .872 | ||||
PC | .818 | .788 | .796 | .758 | .868 | |||
PE | .729 | .808 | .507 | .811 | .809 | .857 | ||
SI | .687 | .787 | .682 | .788 | .785 | .794 | .891 | |
TT | .786 | .662 | .776 | .781 | .862 | .669 | .839 | .89 |
BI | DL | EE | FC | HC | PE | SI | TT | |
|---|---|---|---|---|---|---|---|---|
BI | .84 | |||||||
DL | .792 | .864 | ||||||
EE | .689 | .86 | .873 | |||||
FC | .796 | .858 | .855 | .874 | ||||
HC | .71 | .618 | .863 | .871 | .848 | |||
PE | .677 | .739 | .829 | .836 | .848 | .883 | ||
SI | .761 | .635 | .819 | .806 | .845 | .819 | .868 | |
TT | .785 | .771 | .86 | .851 | .811 | .841 | .846 | .864 |
4.2.2 Hypothesis testing
In this study, the PLS-SEM algorithm in Smartpls 4 was used to estimate the path parameters of the research hypothesis model. Subsequently, the Bootstrapping algorithm was used to test the significance of the hypothesis path coefficients, and the sampling number was set at 5000 times. It can be seen from Table 4 that in the patient population model, the significance P values of the hypotheses H1b, H1c, H1d and H4d are greater than 0.05, and the significance test has not been passed, indicating that the hypotheses are not valid. The P values of the remaining hypotheses are all less than 0.05. The hypotheses are valid by significance detection. It is indicated that among the patient population, their expectations of the efforts required to use the intelligent diagnosis and treatment system have not significantly affected their intention to adopt it. This indicates that patients in medical scenarios are more concerned about the actual effectiveness of the system rather than the convenience during the usage process. Patients tend to rely more on their own judgment and the actual performance of the system in medical decision-making. External social pressure or recommendations have a relatively small impact on them. At the same time, there are problems such as patients' low perception of the matching degree of system functions, or the functional design of the system failing to fully meet the diverse needs of patients.
Hypothesis | Path | Coefficient | T-statistic | P-value | Established |
|---|---|---|---|---|---|
H1a | PE ->BI | .273 | 3.845 | < .001 | True |
H1b | EE ->BI | .077 | 1.417 | .16 | False |
H1c | SI ->BI | .047 | 0.881 | .38 | False |
H1d | FC ->BI | − .002 | 0.03 | .98 | False |
H2a | HK ->BI | .235 | 3.434 | .001 | True |
H2b | FC ->HK | .907 | 6.199 | < .001 | True |
H4a | PC ->BI | .205 | 3.509 | < .001 | True |
H4b | TT ->BI | .129 | 2.477 | .013 | True |
H4c | TT ->PE | .396 | 7.003 | < .001 | True |
H4d | TTF ->BI | .036 | 0.572 | .567 | False |
H4e | TTF ->BE | .556 | 1.935 | < .001 | True |
As shown in Table 5, in the model of medical staff, the p-value of hypothesis H5b is greater than 0.05, which is not significant. The P values of the remaining hypotheses are all less than 0.05. Through significance testing, the hypotheses are valid, that is, the level of human-machine collaboration, such as the manual review interface provided by the system and the ability to handle complex cases, does not significantly affect the intention of medical staff to adopt. The remaining P values are less than 0.05, and the hypotheses are established through the significance test, that is, the man-machine collaboration level, such as the manual review interface provided by the system and the ability to deal with complex cases, does not significantly affect the adoption intention of medical staff. It indicates that in clinical practice, medical staff are more inclined to retain the initiative in diagnosis and treatment, and hold a cautious attitude towards the auxiliary functions of the system. They may think that the system's suggestions can only serve as a reference, and the final decision still depends on their own professional knowledge and clinical experience. Medical staff may lack trust in the system, worrying that misdiagnosis or suggestions from the system will interfere with their clinical judgment, thereby affecting their willingness to adopt it.
Hypothesis | Path | Coefficient | T-statistic | P-value | Established |
|---|---|---|---|---|---|
H1a | PE ->BI | .145 | 3.295 | .001 | True |
H1b | EE ->BI | .148 | 2.851 | .004 | True |
H1c | SI ->BI | .1 | 2.269 | .02 | True |
H1d | FC ->BI | .179 | 3.399 | .001 | True |
H3a | DL ->BI | .119 | 2.154 | .03 | True |
H3b | DL ->EE | .86 | 2.112 | < .001 | True |
H5a | TT ->BI | .086 | 1.651 | .1 | False |
H5b | HC ->BI | .243 | 4.401 | < .001 | True |
5 Discussion
5.1 Research conclusions
Based on the Technology Acceptance Model (TAM), integrated Unified Theory of Acceptance and Use of Technology (UTAUT) and Task-Technology Fit (TTF), this study constructed the influencing factors model of user intention of intelligent healthcare system according to the characteristics of medical context, and empirically tested the differences in the influence paths of two user groups, patients and medical staff. It is found that the intention of the two types of user groups to adopt the intelligent diagnosis and treatment system are affected by technical characteristics, individual literacy, situational factors and ethical risks, but there is a significant differentiation between the factors between the groups, and the influence path and path coefficient of the final model are shown in Fig. 4 and Fig. 5.
5.1.1 Affecting path of patients’ adopt intention
Fig. 4
Adjusted path diagram of patient model.
For the patient population, performance expectation (PE), health literacy (HK), technical trust (TT), and privacy concern (PC) are the key direct factors influencing the adopt intention. Facility condition (FC) and task-technology fit (TTF) indirectly affect the adopt intention of patients by acting on mediating variables. The research results show that performance expectation (PE) has the most significant positive impact on the adopt intention. In the medical scenario, the main needs of patients are to obtain accurate diagnosis and effective treatment, while saving time and cost as much as possible. Whether the system can accurately identify symptoms, shorten the time for seeking medical treatment and provide personalized health management plans is of vital importance. Only by accurately identifying symptoms can an intelligent diagnosis and treatment system provide patients with more accurate diagnostic results, assist doctors in formulating more reasonable treatment plans, and improve treatment outcomes. Technical trust (TT) indirectly positively affects the intention to adopt by enhancing the mediating role of performance expectation, while privacy concern (PC) directly positively affects the intention to adopt, reflecting patients' sensitivity to the risk of data leakage. Patients' trust in intelligent diagnosis and treatment systems is based on the system's performance. When patients trust the system, they will be more willing to believe the diagnostic and treatment suggestions provided by the system, and their expectations for the system's performance will be raised. Facility condition (FC) affects patients' health literacy (HK) by reducing the difficulty and threshold of using the system, indirectly affecting patients' attention to adopt. The failure of the effort expectation (EE) and social impact (SI) to pass the significance test indicates that patients pay more attention to the practical utility of the system rather than external recommendations.
5.1.2 Affecting path of medical staff’s adopt intention
Fig. 5
Adjusted path diagram of medical staff model.
A
At the level of medical staff, digital literacy (DL), performance expectation (PE), and facility condition (FC) have a significant positive impact on the intention to adopt. Digital literacy (DL) indirectly enhances the willingness to adopt by increasing the effort expectation (EE). The technical proficiency of medical staff in operating the system determines their perception of the system's ease of use. Medical staff with high digital literacy can skillfully use the intelligent diagnosis and treatment system, quickly master the system's functions and operation methods, and improve work efficiency and quality. The direct impact of performance expectation (PE) on users' intention to adopt highlights the value of the system in assisting clinical decision-making. intelligent diagnosis and treatment systems can quickly extract and analyze medical record information, providing diagnostic references and treatment suggestions for medical staff, reducing repetitive work and improving work efficiency. The path coefficients of human-machine collaboration (HC) and facility condition (FC) are significant, indicating that medical staff prefer systems with convenient and easy-to-use functions that can be integrated with existing workflows. Medical staff tend to view systems as auxiliary tools and maintain the dominant position of manual review and decision-making. Medical staff are the ultimate decision-makers of diagnostic behaviors. They need to review and judge the diagnostic and treatment suggestions of the intelligent diagnosis and treatment system. The clinical experience of medical staff is also irreplaceable by the system. They can adjust and optimize the system suggestions according to the actual situation of the patient to achieve the best effect of human-machine collaboration.5.1.3 Current situation of the integration of intelligent diagnosis and treatment system in scientific researches
For the group of researchers who do not have sufficient data to support the research, this study searched and integrated the current status and progress of the use of intelligent diagnosis and treatment systems in the field of scientific research. At present, the open use of data in intelligent diagnosis and treatment systems has formed a multi-level practical exploration. The construction and application of the National Cancer Clinical Diagnosis and Treatment Database in China have provided researchers with a large amount of clinical diagnosis and treatment data on tumors, strongly supporting scientific research related to tumors. The "Intelligent Management System for Severe Tumor Information" independently developed by the Cancer Hospital of Tianjin Medical University uses data from monitors, ventilators and other equipment as well as hospital information systems to build a localized hierarchical storage architecture for all-round severe tumor cases. These data are not only used for clinical decision-making assistance but also open to research teams to support the construction of tumor prognosis models and the optimization of treatment plans. It has solved the problems of large sample heterogeneity and short cycle in the research of severe tumors. Infervision Medical Technology has collaborated with over a thousand medical institutions in more than 30 countries around the world to build an "AI Medical Data Factory" based on high-quality clinical data. It has opened up the desensitized CT image data of the pulmonary nodule detection system for research teams to optimize the sensitivity and specificity of the model.With the continuous advancement of technology and the gradual improvement of relevant standards, it is expected to better solve the existing problems such as different data collection standards, different data structures, and missing or ambiguous semantic information, promote the open sharing of medical data, promote the wider application of intelligent diagnosis and treatment systems in the field of scientific research, and realize the innovative development of medical research.
5.2 Management Implications
This study revealed the differentiated driving mechanism of users' intention to adopt intelligent diagnosis and treatment systems, providing multi-dimensional practical guidance for the in-depth application and continuous promotion of intelligent diagnosis and treatment systems.
5.2.1 Feasible strategies for enhancing patients’ adopt intention
(1) Improve the accuracy of core functions
The system should enhance symptom recognition and diagnostic suggestions, leveraging cutting-edge artificial intelligence and data analysis technologies to conduct extensive medical data mining and learning. This will enable the system to have a better ability to identify vague symptoms and provide accurate diagnostic suggestions for patients. Developers can collect and analyze case data in collaboration with multiple top-tier cooperative hospitals to train the system's diagnostic capabilities for common and rare diseases, enabling the system to better identify patient symptoms and provide diagnostic suggestions.
The system can provide personalized health management for patients based on their health conditions, basic medical history, living habits and other information, such as dietary guidance, exercise guidance, medication reminders and other contents. Take blood glucose monitoring as an example, the system can automatically adjust the patient's blood glucose control plan based on the patient's blood glucose measurement values and information such as the patient's diet and exercise, assisting the patient in controlling blood glucose levels. Through the linkage with wearable devices, it can monitor the patient's health data in real time, promptly detect hidden dangers, and issue early warnings to the patient.
(2) Establish a transparent and efficient privacy protection mechanism
During the process of data transmission and storage, advanced encryption technology should be used to protect patients' sensitive data, preventing data leakage and illegal acquisition. Before the application of the intelligent diagnosis and treatment system, the purpose, scope and method of data usage should be informed to patients through user agreements, privacy agreements, etc., and their consent should be obtained. The scope of data usage should be strictly controlled, and the data should only be used for medical diagnosis, treatment and health management, and must not be used for commercial purposes such as advertising push and third-party data sale, to enhance patients' trust in the system.
(3) Establish an endorsement mechanism with authoritative medical institutions
By leveraging the endorsement effect of authoritative medical institutions and obtaining recommendations and recognition from well-known medical institutions, patients' technical trust in the intelligent diagnosis and treatment system can be enhanced. System developers and users can establish cooperation with well-known domestic tertiary hospitals and authoritative medical research institutions to jointly develop, test and promote intelligent diagnosis and treatment systems. By leveraging the authority and professionalism of large hospitals, they can enhance patients' trust in intelligent diagnosis and treatment systems. During the research and development and promotion of the system, it can cooperate with medical institutions to carry out large-scale clinical validation, effect evaluation and other projects, scientifically argue and verify the diagnosis rate, effectiveness rate, safety and other aspects of the intelligent diagnosis and treatment system, and publish the diagnosis results to increase patients' trust in and willingness to try the intelligent diagnosis and treatment system.
5.2.2 Feasible strategies for enhancing medical staff’s adopt intention
(1) Enhance the clinical decision support capability of the system
System developers should further improve the system function of case data extraction, analysis and mining, so that the system can extract the patient's past medical history data in the hospital's electronic medical record system faster and more accurately for analysis and mining, provide medical staff with the most comprehensive and accurate patient condition analysis and diagnosis report, and help medical staff understand the patient's condition faster and provide medical staff with more reasonable treatment plans. By integrating the latest global medical research achievements, clinical guideline materials, and expert experience, a medical knowledge base is established and combined with a systematic artificial intelligence diagnostic engine to provide medical staff with real-time diagnostic analysis, real-time diagnosis and treatment reference suggestions, helping medical staff improve diagnostic accuracy and therapeutic effects. By conducting real-time detection and analysis of the diagnosis and treatment process of medical staff through the system, it helps medical staff identify potential problems and hidden dangers, and provides more timely diagnostic suggestions for them.
(2) Optimize the integration of the system and workflow
Developers should have an in-depth understanding of the workflow and usage habits of medical staff, cooperate with professional user interface design teams and system target users, optimize the user interface of the intelligent diagnosis and treatment system, make it more in line with the operating habits and work needs of medical staff, and improve the usability and operational efficiency of the system. Strengthen the integration and consolidation with hospital information systems, achieve seamless connection and data sharing between intelligent diagnosis and treatment systems and hospital electronic medical record systems, laboratory information systems, imaging information systems, etc., reduce the workload of medical staff in switching between different systems and data entry, improve work efficiency and data accuracy, realize data collaborative sharing among multiple departments, and enhance the compatibility of the system with the workflow of medical staff.。
Medical institutions can set up demonstration wards or pilot departments of intelligent diagnosis and treatment systems in the hospital, demonstrating the functions and advantages of the system through practical applications, so that medical staff and patients can experience the convenience and effectiveness of the system firsthand, thereby driving the promotion and application of the whole hospital. According to the functional characteristics of the intelligent diagnosis and treatment system, institutions can make appropriate adjustments to the workflow and division of responsibilities of the medical team, clarify the role and positioning of the system in clinical work, give full play to the auxiliary decision-making role of the system, and achieve the best effect of human-machine collaboration。
(3) Strengthen digital skills training and support
In order to improve the popularity and acceptance of the intelligent diagnosis and treatment system in the whole society, in view of the problem of uneven digital skill levels of system users, medical institutions and social organizations need to actively carry out digital skills training in various scopes to help patients and medical staff master the functions and operating procedures of the intelligent diagnosis and treatment system, reduce the cost of system learning, and then improve the skill literacy of the patient group operating system and the efforts and expectations of medical staff for the use of the system, and improve their operational proficiency and application ability of the intelligent diagnosis and treatment system. Lower the threshold for the use of intelligent diagnosis and treatment systems from the root.
Developers should also establish a professional technical support team to provide timely and effective support and solutions for technical problems encountered by users during the use of the intelligent diagnosis and treatment system. They provide comprehensive technical support services through various means such as on-site technical support to ensure that users can smoothly use the system to fulfill their needs.
5.2.3 Feasible strategies for enhancing scientific researchers’ adopt intention
(1) Improve the quality and availability of data resources
System developers need to formulate a strict data quality control system, strictly control the quality of scientific research data from data collection, entry, storage, control and other links, strengthen data cleaning, sorting, annotation and other work, improve data quality and availability, and provide high-quality data services for scientific researchers; Integrate multi-modal data such as clinical data, imaging data, and electronic medical records to provide all-round and multi-angle data support for researchers, meet the needs of researchers required to carry out complex medical research, and promote the in-depth development of medical research.
(2) Promote the sharing and openness of medical data
Led by the government, a unified medical big data sharing center should be established to integrate medical big data resources from medical institutions at all levels, achieve the sharing and exchange of big data resources, and promote data sharing and big data collaboration among medical institutions by formulating big data sharing norms and standards, clarifying the mechanism and process of big data sharing, and providing big data sources for the research and application of intelligent diagnosis and treatment systems. To encourage big data sharing among medical institutions, relevant departments can formulate reward measures for big data contributors, such as providing certain economic rewards and research support to them, to enhance the enthusiasm of medical institutions for big data sharing and promote the maximum utilization of big data resources in medical institutions.
(3) Establish a result-driven positive feedback mechanism
By leveraging the power of academic achievements, various institutions and researchers can publish research results involving intelligent diagnosis and treatment systems in high-level medical journals, demonstrating the effectiveness, safety and reliability of the systems in clinical applications. This will promote the synergistic improvement of adoption willingness among different user groups and accelerate the wide application of intelligent diagnosis and treatment systems.
5.3 Outlook
This study has certain limitations in terms of the age and educational distribution range of the sample population, mainly focusing on young people and those with a higher level of education. Future research should expand the sample range and pay more attention to the willingness and influencing factors of the elderly and low-education groups to use intelligent diagnosis and treatment systems, so as to achieve comprehensive coverage of different groups and provide a broader theoretical basis for the promotion and application of intelligent diagnosis and treatment systems.
The application of intelligent diagnosis and treatment systems involves multiple disciplinary fields. In the future, interdisciplinary teams can be organized to carry out data verification work. Starting from the knowledge and methods of multiple disciplines such as medicine, computer science, psychology, and sociology, the influence mechanism of users' intention to adopt can be explored from a multi-disciplinary perspective, theoretical models can be improved, and more comprehensive and specific guidance can be provided for the improvement of intelligent diagnosis and treatment systems. Subsequent research can introduce moderating variables such as cultural background, disease type and accessibility of medical resources to explore the differences in user adoption mechanisms under different backgrounds. The cultural acceptance of intelligent diagnosis and treatment systems may vary in different regions, and patients with certain disease types may have more specific requirements for system functions. Compared with regions rich in resources, users in areas with scarce medical resources may also have different demands and expectations for this system. By introducing these moderating variables, the influencing factors of the intention to adopt intelligent diagnosis and treatment systems can be identified more accurately, providing certain references for the promotion of intelligent diagnosis and treatment systems in different scenarios.
This study is a cross-sectional study, which can be combined with dynamic data to analyze the long-term evolution of users' behavior in the future. By collecting user data in different periods, observing the change trend of their adoption intention over time, and obtaining user adoption intention data in different periods based on system performance upgrades, technological improvements, government policies, etc., and observing the changes of user adoption intention in the time dimension, it provides a more scientific basis for the continuous update and improvement of intelligent diagnosis and treatment systems.
List of abbreviations
AIGC
artificial intelligence penerated content
APP
application
TAM
Technology Acceptance Model
TTF
Task—Technology Fit
UTAUT
Unified Theory of Acceptance and Use of Technology
Declarations
Ethics approval and consent to participate
The requirement of ethics approval is exempted by the Academic Ethics and Morality Committee of the Academic Committee of Nanjing University of Posts and Telecommunications.
The study adheres to the Declaration of Helsinki to this effect.
The study has obtained the informed consent of all research participants.
Consent for publication
Not applicable.
A
Data Availability
All data generated or analysed during this study are included in this published article and its supplementary information files.
A
Funding
Not acceptable.
A
Author Contribution
GT Independently completed the research process and the manuscript writing.
Acknowledgments
First, I would like to express my heartfelt thanks to my mentor Yihan Zhang, who helped me successfully complete the writing of my thesis with her rigorous attitude towards learning, profound academic attainments and careful guidance and clarification. From determining the topic selection to sorting out ideas, from questionnaire design to data analysis, to the final thesis writing and revision, her guidance and encouragement in each stage have given me great help. Her strict requirements for every detail have allowed me to develop rigorous and serious study habits, and her unique insights into academics have allowed me to continue to innovate in my research.
Further, I want to thank everyone involved in my thesis writing process. Each of their questionnaires formed the cornerstone of my research; Each of their journal articles constitutes a beacon of my research; Each of their criticisms and suggestions polished my research even more.
Authors' information
Author: Tianze Gao (Nanjing University of Posts and Telecommunications, School of Management, Nanjing, Jiangsu Province, China)
Corresponding Author: Tianze Gao (Phone: +86 18226587711, e-mail: g18226587711@163.com)
Electronic Supplementary Material
Below is the link to the electronic supplementary material
References
1. Zhang Z, Luo C, Jiang Z, Geng N. Research on the Willingness of Doctors to Adopt Artificial Intelligence Aided Diagnosis and Treatment System Based on the UTAUT Model. Chinese Hospital Management. 2024;44:79–83.
2. Yin X, Wu P, Qian Y, Chen J. Context-driven digital intelligence innovation enabling new quality productive forces:Theoretical logic and practical approach. China Soft Science. 2024;(10):18–31.
3. Chen K, Xue Z, Zhang C. Social innovation in the digital era:Theoretical framework and policy implications. Studies in Science of Science. 2025;43:775–786.
4. Feng J, et al. Full-dimension Data and Intelligence Medical Care:Prospects and Challenges. Bulletin of National Natural Science Foundation of China. 2021;35:73–80.
5. Ye X, Ma S, Wang Y, Li J, Zhang J. The Evolution of Intelligent Medical Industry from the Scenario-driven Perspective. Science & Technology Progress and Policy. 2023;40:20–30.
6. Ma X, Wang Z, Zhou S, Wen H, Zhang Y. Intelligent healthcare systems assisted by data analytics and mobile computing. Wireless Communications and Mobile Computing. 2018;(1):3928080.
7. Wang W, et al. Reflection on High-Quality Development of Smart Health Data. Strategic Study of Chinese Academy of Engineerng. 2024;26:32–42.
8. Yu H, Mou D, Wang C. Research on Interpretability Strategy of Intelligent Diagnostic Model. Journal of Modern Information. 2024 http://kns.cnki.net/kcms/detail/22.1182.G3.20241223.0907.002.html. Accessed 23 Dec 2024.
9. Chu J, Fan X, Liu B. Intelligent Health Knowledge Service Platform Enabled by AIGC. Library Tribune. 2024;45:126–132.
10. Zhao Y, Zhang J. Research on the Human-Intelligence Interactive Experience Based on Embodied Intelligence:Theory,Application and Prospect. Information Studies:Theory & Application. 2025;48:52–62,72.
11. Gu D, Huang Z, Zhu K, et al. Medical Health Large Language Models Construction of Knowledge System, Service Applications, and Synergetic Governance of Risk Management. Information Science. 2024 http://kns.cnki.net/kcms/detail/22.1264.G2. 20240925.0948.002.html. Accessed 25 Sep 2024.
12. Zhang J, Liu J, Deng J, Liu Y, Huang Q. Research on the Framework of Online Medical Health Wisdom Q&A Services Driven by Knowledge Graph and Large Language Model. Journal of Modern Information. 2025;45:164–176.
13. Wang C. Research on Medical Artificial Intelligence for Ethical Governance:Development Trends,Key Issues,and Future Prospects. Science and Technology Management Research. 2024;44:217–224.
14. Chen J, et al. Comparative Study of Artificial Intelligence Generated Content and User Generated Content from a Linguistic Perspective: Taking Online Healthcare Services as an Example. Information Studies:Theory & Application. 2024;47:192–201.
15. Murugan M, et al. Empowering personalized pharmacogenomics with generative AI solutions. Journal of the American Medical Informatics Association. 2024;31:1356–1366.
16. Su Q, Shen X, Hu Y. Research on Simulation and Optimization of Multi-Agent Hierarchical Medical System. Industrial Engineering and Management. 2020;25:10–18.
17. Zheng X, Liu C, Liu Q. Benefiting the Public or Pursuing Profit? Consumer Perception and Response to AI Drug. Nankai Business Review. 2025;28:4–15.
18. Liao C, Chen J. Research on the Influencing Factors of User Adoption Intention of Smart Elderly Care Services under the Background of Comprehensive Health Industry:based on the Perspective of Perceived Quality. Modern Management Science. 2021;(05):109–120.
19. Wu D, Hu X, Zhao Y, Ai W. Research on Influencing Factors and Path of Users' Continuous Adoption Intention of Digital Health Apps from the Perspective of Perceived Value:Based on fsQCA. Library and Information Service. 2021;65:93–104.
20. Zhu Z, Liu Y, Cao X. Influencing Factors of Mobile Health Users'Adoption Intention:A Meta-Analysis. Journal of Systems & Management. 2020;29:49–60.
21. Chen Y, Zhou X, Yue L, Yu X. Factors Influencing Adoption Willingness of the Health Information in the Mobile UGC Community. Document,Informaiton & Knowledge. 2022;39: 82–95.
22. Zhu Z, Liu Y, Liu J. Review of Empirical Study on Adoption Behavior of Mobile Medical Users. Science and Technology Management Research. 2020;40:206–213.
23. Mo M, Kuang Y, Zhu Q, Li X, Yue Q. Research on the Influencing Factors of Users' Intention to Adopt Online Medical Consultation Information. Journal of Modern Information. 2022;42: 57–68.
24. Han X, Jiang P, Chen S, Han W. The Influence Mechanism of Review Features on Users'Willingness to Adopt Online Physician Review Information——Based on ELM Model and Trust Transfer Theory. Journal of Modern Information. 2023;43:36–50.
25. Zheng Z. Medical Damage Liability for Artificial Intelligence in Medical Diagnosis. China Legal Science. 2023;(01):203–221.
26. Zhang X, Zhang T. Internet Medical Acceptance Study Based on TAM. Science and Technology & Innovation. 2021;(06):1–3.
27. Deng Z, Li H, Ge M, Qi B. Research on Users' Behavioral Intention of Mobile Medical Services Based on UTAUT Model. Chinese Hospital Management. 2022;42:68–72.
28. Yan W, Wu J, Yuan Q. Task-Technology Fit Theory and Its Application and Prospect in the Field of Information System Research. Journal of Modern Information. 2024;44:147–154.
29. Nie L, Ren W, Lv J, Zang C. Analysis of influencing factors of the elderly's intention to use mobile medical services and the moderating effect of individual characteristics. Chinese Hospitals. 2023;27:65–68.
30. Du T, Li J, Li N. Empirical Study on the Willingness to Use Remote Consultation Among Doctors in Yan'an Medical Alliance Based on the Extended UTAUT Model. Medicine and Society. 2023;36:131–137.
31. Fornell C, Larcker DF. Evaluating structural equation models with unobservable variables and measurement error. Journal of marketing research. 1981;18:39–50.
Additional File 1
File format: .doc
Title of data: Questionnaire on the user's willingness to adopt the intelligent diagnosis and treatment system and the influencing factors
Description of data: The attachment contains the questionnaire template designed and developed for this research that was used in the questionnaire survey process.
Additional File 2
File format: .xls
Title of data: Original data of the questionnaire
Description of data: The data of 430 valid questionnaires collected through the distribution of questionnaires.