In this study, we conducted a survey to identify the perceptions and demands of health big data analysis and use among workers in health care–related fields.

Among the survey panels held by South Korean research, URLs were sent via email and mobile phone to those who met the criteria for this study, and snowball sampling was conducted by sending URLs through social media, mainly to acquaintances of researchers who were engaged in health care–related fields. The specific criteria for selecting participants were as follows: (1) currently working in South Korea as a doctor, nurse, medical technician, pharmacist, hospital employee, pharmaceutical company or medical equipment company employee, health care–related department professor, or health care–related field researcher and (2) those who voluntarily agreed to participate. The number of participants was calculated as 384, which satisfies the confidence level of 95% and the maximum sampling error of 5%, and finally, 400 people were surveyed in consideration of the dropout rate [16-18].

We prepared the questionnaire based on the health insurance big data usage experience and data opening demand survey conducted by the National Health Insurance Corporation [19] and based on the questionnaire used in Baek et al’s [20] study titled “Recognition and Use of Cancer Big Data in South Korea.” To ensure the appropriateness of the questionnaire, we recruited 2 university professors and 2 experts from the health care and information and communication domains to review it. We revised and completed the questionnaire based on the feedback from these reviewers. The 29-question survey was divided into 5 categories, namely, basic information of respondents, awareness and use of health big data, experiences of health big data collection and analysis, demand for health big data analysis and use, and some conditional questions (Figure 1). To gauge respondents’ understanding of health big data, we asked the following questions: “Have you ever witnessed the use of health big data?” and “Have you ever felt the need to use health big data?” In relation to respondents’ current or potential use of health big data, we asked, “Do you have any experience using health big data?” We also asked users whether the situation necessitated the use of health big data. We used the 5-point Likert scale to score the questions on health big data collection and analysis experience. There were questions that measured the degree of difficulty experienced in the process of collecting and analyzing health big data from 1 “very difficult” to 5 “very easy,” meaning that the lower the score, the greater was the difficulty. Finally, in relation to the development of a system to aid in the health big data analysis process, we included a subjective question requesting respondents’ suggestions. To meet the demand for health big data analysis and use, it is necessary to develop a big data system.

We conducted a web-based survey to collect data from May 8, 2022, to May 18, 2022, after obtaining approval from the bioethics committee of Keimyung University. South Korea Research (a polling company) commissioned the web-based survey. The company sent emails to health care workers in its survey panels. It distributed URLs through emails and simple notification services using the Naver form. We received 400 responses. However, we deleted 10 questionnaires with incomplete responses. This filtration yielded a final sample of 390 questionnaires. We surveyed participants who provided consent to participate after understanding the study’s purpose and their rights to anonymity, confidentiality, and withdrawal. We also gave participants a gift as a token of gratitude for participation.

We analyzed data using SPSS 23.0 (IBM Corp). We used Fisher exact test and analysis of variance to determine the differences among occupations. We expressed the results of the analysis by item in frequency and percentage and expressed the difficulties in the health big data analysis process in mean and standard deviation.

This study was approved by the institutional review board of Keimyung University (40525-202112-HR-083-04). The research purpose, methods, and participants’ rights, including that they could cease participation at any point without penalty, were explained. All questionnaires were completed anonymously, and participants were told that the results would not be used for anything other than research purposes. Participation was voluntary, and all respondents provided written informed consent. Those who participated in the web-based survey were given an e-voucher worth US $10.

Table 1 shows the demographic characteristics of the respondents. Of the 390 respondents, 178 (45.6%) were males and 212 (54.4%) were females. Regarding age, 76 (19.5%), 141 (36.2%), 119 (30.5%), and 43 (11%) participants were in their 20s, 30s, 40s, and 50s, respectively. With regard to relevant practical work experience, 118 (30.3%), 102 (26.2%), 85 (21.8%), and 85 (21.8%) had experience of 1-2 years, 3-5 years, 6-10 years, and more than 10 years, respectively. Regarding occupations, 34 (8.7%), 47 (2.1%), 79 (20.3%), 24 (6.2%), 55 (14.1%), 44 (11.4%), and 107 (27.4%) respondents were doctors, nurses, medical technicians, hospital administration and computer staff, pharmaceutical and medical company staff, professors and researchers, and personnel from other health care fields, respectively. Regarding respondents’ responsibilities, 156 (40%), 78 (20%), 26 (6.7%), and 12 (3.1%) were working as direct health care providers, administrative and management personnel, marketing personnel, and data processing and information personnel, respectively.

Table 2 shows the perceptions of health big data among the respondents. In total, 47.9% (187/390) and 86.4% (337/390) of the respondents stated that they had witnessed health big data use and that they felt the need to use health big data, respectively. Concerning the current use of health big data, 13.3% (52/390) and 24.9% (97/390) of the respondents stated that they had very low and somewhat low use, respectively, while 38.2% (149/390) of the respondents expressed the need to use health big data in the future, and 65.6% (256/390) and 29.5% (115/390) of the respondents gave full consent and agreed to the future need for big data use, respectively.

Table 3 shows respondents’ use of health big data. Among 65.6% (256/390) of the respondents who never used health big data, 39.6% (153/386) attributed their nonuse to the lack of awareness about the source of the desired data (the most cited reason for nonuse). Among the users, the most cited purpose for using big data was for collecting and utilizing work-related data (62/242, 25.6%), followed by the purposes of predicting and diagnosing diseases (52/242, 21.5%), streamlining other care services (45/242, 18.6%), and academics and research (35/242, 14.5%). Concerning the types of big data used, the most used type of health big data was public institution data (75/249, 30.1%), followed by patient care data (57/249, 22.9%) and clinical study data (44/249, 17.7%). By comparing the types of health big data used by occupation, we found that pharmaceutical or medical companies made substantially less use of patient care data (Figure 2).

Table 4 shows respondents’ experiences with health big data collection and analysis. Experience “integrating different data and matching expression units” received the lowest score of 2.39 points out of 5 points. Experience “processing missing values and removing noise from collected data” and experience “removing and converting data into easily understandable formats” scored 2.50 points and 2.61 points, respectively. Experience “converting collected data into structural forms and storing and managing data” and experience “choosing appropriate analysis techniques” scored low at 2.71 points and 2.78 points, respectively. These findings confirmed that participants faced difficulties in data preconditioning during the health big data analysis process.

We compared the difficulties faced by different professionals during the big data analysis process. Professionals from 5 health care fields had the lowest score during the process. Among them, doctors had difficulties integrating different data and matching units of expression. This finding confirmed that participants experienced common difficulties, regardless of occupation. Among all the professionals, nurses and pharmaceutical company employees had the lowest scores (Figure 3).

Table 5 shows respondents’ demand for the analysis and use of health big data, with 91.8% (358/390) expressing the need for a system that can facilitate big data analysis and use. This need followed the need to develop health care services. Concerning the application of health big data, 21.1% (214/1016), 12.5% (127/1016), 14.8% (150/1016), 13.2% (134/1016), 12.5% (127/1016), and 12.3% (125/1016) of the respondents stated that health big data should be used for health care service development, individual disease prediction, chronic disease prediction, working process improvement, health care product development, and disease surveillance system development in communities and companies, respectively.

We received 210 answers to the subjective questions regarding the development of a system that helps analyze big data. The respondents who highlighted the need to improve access to personal health and medical care data and to develop a system to support data collection of these data accounted for 18.1% (38/210). This was followed by the need for professional guidance on the selection and configuration of appropriate data (26/210, 12%), providing advice on questions arising during the overall process (34/210, 16.1%) and easier data analysis module development (23/210, 10.9%). Other recommendations included the need to compare analytical techniques and provide guidance on appropriate analytical techniques (18/210, 8.5%); increase the ease of use and access (18/210, 8.5%); and provide training and guidance manuals for program use (15/210, 7.1%), guidance (10/210, 4.7%), and simple procedure data integration systems (8/210, 3.8%). In relation to increasing the usability of health big data, respondents highlighted the need to provide education on and promote such data (17/210, 8.1%). This was followed by opinions on the need to consider views on developing measures for protecting personal information (10/210, 7.5%). Few respondents also provided opinions on the need to develop a data classification system, provide expert feedback on the interpretation of analyzed data, and revise law for convenient procedures.

This study was conducted to understand the perception of and demand for health big data analysis and use among domestic health care–related workers in South Korea. According to our survey of 390 health care–related workers, 86.4% (337/390) of the respondents felt the need to use health big data and 95.1% (371/390) recognized the need to use big data in the future. However, only 23.6% (92/390) of the respondents stated that health big data are being satisfactorily used and that it is yet to reach its full potential. Only 47.9% (187/390) of the respondents had seen the use of health big data, which reflects the low awareness and knowledge of health big data use among health care professionals. Among the 65.6% (256/390) of the respondents who had never used health big data, 39.6% (153/386) attributed their nonuse to the lack of awareness about the source of the desired data (the most cited reason for nonuse).

Despite various attempts to increase health big data use, such as government-level efforts, legal reforms, and the establishment of health big data systems by public institutions, individuals still have difficulty accessing health big data. Currently, a service that guides the introduction and access procedure of public big data that are also available on the government-built health care big data platform is provided. However, even this can be considered difficult for beginners with no analysis experience; thus, a system that enables immediate question-and-answer with a more specific explanation should be combined. Many of those who have experience using health big data have experienced significant difficulties in tasks such as “defects in other data sources” and “missing value processing and noise removal,” which are part of the data preprocessing stage during the big data analysis process. However, these data integration and purification processes are closely related to the quality of the data. As the quality of data plays a significant role in the accuracy of the analysis results, it is important to solve these issues [21]. Although patient treatment data were highly utilized in hospitals by doctors, nurses, and administrative and computer workers, their use was low in other industries such as pharmaceutical and medical instrument companies. These findings imply that nonhospital workers may find it challenging to use health big data, as clinical data are difficult to access and use due to issues of data standardization (ie, data structure and governance is unique to each hospital). Hence, it is urgent to develop a platform to expand accessibility and ease of use by standardizing and constructing hospital data through systems such as the electronic medical records certification system. The government is currently promoting the development of the Common Data Model [22].

In this study, many respondents wanted to use the health big data actively to develop useful programs or products that contribute to health care and promotion in the future. To date, health big data use has primarily been limited to the collection or utilization of work-related data or the purpose of academic research. In addition to the innovative development of digital technology after COVID-19, various digital health care services are rapidly emerging in the health and medical field [23,24]. If accurate predictions and evidence-based judgments based on health big data are applied to the development of health care services or products developed in various health-related fields such as diet, beauty, and exercise, more precise services can be provided to users [25,26]. In the case of health care fields, professionals can develop prediction models via data mining technologies. This can help them analyze large data sets of patient data and identify different clusters from and correlations between data sets [27]. The use of health big data with artificial intelligence technologies such as machine learning and deep learning can further facilitate innovation in health care services, such as individual clinical decision support systems and real-time precision diagnosis [28,29]. Organizations can use big data based on artificial intelligence to address cost and efficiency problems emerging during drug trials [9]. Therefore, supporting the use of health big data for various purposes in various fields can lead to a paradigm shift in the health care field.

In this study, one of the main factors hindering the use of health big data was the difficulty in the analysis process, and most of the respondents expressed wanting a support system to help with the big data analysis process. In particular, many respondents noted the need for services that provide expert guidance or advice in the overall process of data collection, preprocessing, analysis, and interpretation of analysis results. However, it is not easy to establish a support system that provides real-time communication with experts in the current situation, wherein there is a substantial shortage of experts in big data analysis. Therefore, it is expected that users’ difficulty can be largely addressed if an interactive system is developed to enable them to communicate with real-time users on behalf of such expert roles. In addition, with the increasing demand for non–face-to-face services in the post–COVID-19 era, an intact system can be more useful [30]. However, users’ level of experience with big data analysis varies widely from beginners to experts; thus, the developed support system should be designed to meet the level and needs of users [31]. Further, more specialized support should be provided for the specific challenges that arise at each stage during the entire process of big data analysis [32]. Recently, the scope of the user-centered design methodology in industrial design has been expanded and applied not only to the development of products with external shapes but also to the development of technologies or services [33]. This approach enables customized design by specifically identifying user requirements and problem situations and reflecting them in the service design [34]. Therefore, in developing a support system that helps analyze and utilize health big data, reflecting the empirical characteristics of the participant should be considered first.

Currently, various national policies and projects are being promoted for the effective use of health big data based on large public health data holdings and advanced digital technology, and as a representative example, a health care big data platform has been developed and provided. However, these achievements developed with substantial financial input have not led to the active use of health big data in the actual health care field. Therefore, to ensure that new health care data–related projects that are currently or continuously promoted in the future do not merely remain as tangible results, it is urgent to specifically identify and resolve the barriers to entry to health big data for end users. Based on this study, we propose a study on the development of customized health big data analysis and utilization support services tailored to users’ needs, and ahead of that, a follow-up study is needed to identify the specific needs of health big data users.

This study has several limitations. First, to check the answers of the participants related to health big data collection and analysis, we selected some participants through snowball sampling, starting with those who had analyzed health big data; however, this is not a probability sample for health care workers, and the investigation is limited to the situation in South Korea. Second, as the questionnaire has not been verified for objective reliability and validity, there may be limitations in interpreting the results. Third, this study was conducted over a relatively short period of time, and the contents confirmed in the study are the general experiences and requirements of various workers in the health care field; therefore, it is difficult to identify specific needs subdivided by the participants. In the future, a follow-up study is needed to subdivide the participants and divide them by type to confirm participants’ in-depth experience by type.

In this study, respondents strongly perceived the need for health big data. However, the actual use was low in the related field. They attributed the reasons for low utilization to difficulty getting access to data and difficulty in the analysis process. They highlighted the need to develop a system that can assist them at each stage of the data analysis process. This is expected to increase big data use. Therefore, this study attempts to identify the main issues hindering the use of health big data by investigating the perceptions of the health big data and the experiences and needs related to big data analysis among various health care workers. This study is meaningful as the first attempt to identify critical failure factors that are easily overlooked in health big data utilization, and it is hoped that various attempts and studies will continue to discover and solve more specific and detailed utilization problems in the health big data area in the future.