Keywords

Introduction

The National Health Insurance (NHI), the single-payer system, not only protects the right to access to health care for Taiwan’s residents, but also centralizes and coordinates resources and regulates their wide range of applications. The system has won international praise for its advantages of low administrative costs, fairness and consistency of insurance premiums and has become a reference model for other countries when establishing or reforming their health insurance systems. In recent times, the NHI has been undergoing a digital transformation by combining digital technology and the concepts of data governance. It is hoped that the NHI will provide high-quality medical care for the public under the premise of limited resources and continuously move toward the ultimate goal of Universal Health Coverage (UHC) proposed by the World Health Organization (WHO).

Issues: People’s Consent to Data Authorization Collection [1]

In 2012, the Taiwan Association for Human Rights filed a lawsuit against the Health Insurance Administration (NHIA), arguing that the medical records of the NHI cannot be provided to researchers and academic organizations without people’s consent and explicit authorization by law. Although the NHIA won the lawsuit because of public interests, this argument is under review by the Constitutional Court.

The NHIA has a huge database, with a collection of more than 70 billion records of medical data accumulated over the past 27 years. Since 2018, by collecting medical images, including computed tomography (CT) and magnetic resonance imaging (MRI), and the analysis and value-added applications of the NHI big data, the NHI database has become the cornerstone for developing the precision medicine and biotechnology industry.

The NHIA has great concerns about the legitimacy of the secondary use of the NHI data and personal information protection. It is essential that the release of the NHI data is legitimate and that these de-identified data are pseudonymized or anonymized.

People tend to regard pseudonymized data as anonymized data. However, according to the US Health Insurance Portability and Accountability Act, 18 items of personal identification fields (identifier) must be removed to achieve de-identification. Thus, the NHIA still has a long way to go.

The EU General Data Protection Regulation (GDPR) states that pseudonymized personal data still constitute one kind of personal data. In Taiwan, medical data and insurance data were uploaded to the NHI system, and personal identification fields were encrypted. Although completely removing personal data may avoid data being identified, there is still a possibility of data leakage. If too many personal data are deleted, it may not be easy to use academic research and value-added applications.

According to the Personal Data Protection Act (PDPA), personal data, including medical records, medical services, genetics, and health examinations, shall not be collected, processed, or used. However, this Act also points out that government agencies and academic institutions are not limited to health care, public health, and crime prevention. It should be guaranteed that the specific individuals cannot be identified by data processing.

The NHIA has developed the My Health Bank system for people and the National Health Insurance MediCloud System (NHI MediCloud System) for physicians and pharmacists to facilitate data use. More than seven million people have used the My Health Bank system to check up on personal medical records.

The NHIA has repeatedly stated that the rationale of the secondary use of the NHI data is based on the public interests and the use must be in accordance with the law. In fact, it is essential for people to be fully informed and then the NHIA must obtain their consent. Therefore, one questionnaire in the National Health Insurance Mobile Easy Access mobile application (NHI Express app) is provided for users to express their willingness to consent to the use of their own personal medical records.

Taiwan’s NHI adopts a single-payer system. The claims data collected over the past 27 years are among the country’s precious assets. Facing the issues of human rights protection, the point of balance must be reached between innovations and securing people’s rights. Further, making good and proper use of the NHI data cannot be avoided in the face of public interests.

National Health Insurance Open Data and Personal Data Protection—Challenges and Prospects of the Secondary Use of Health Data [2]

In 2020, the NHIA revised its rules on the NHI Database Application Service to include commercial access to de-identified data of 3.5 million deceased individuals. This plan to unlock data has sparked controversy. The NHIA has consequently suspended this policy implementation to wait for social consensus.

It has been argued that deceased individuals are not natural persons to be protected by the data protection law and that the public benefits of open data should be recognized. However, it remains arguable whether there are sufficient legal grounds for the secondary use of the deceased’s health data beyond its primary purpose. In Taiwan, the original purpose of collecting NHI data is for premium auditing only. As the deceased individuals had not had a chance to consent to data reuse while they were alive, opening up their health data for secondary purposes requires a lawful basis.

According to Paragraph 4, as a proviso of Article 6 of the PDPA in Taiwan, “public service agencies” or “academic research institutions” may use de-identified data for health or medical purposes. This rule sets restrictions on a limited set of applicant entities so that commercial entities need to rely on a public–private collaboration to apply for access to NHI data.

Even though the NHIA had once revised its internal rules to expand data access to commercial entities, such administrative rules are not up to the legislation level. In addition, even if there is a requirement for approval from institutional review boards (IRBs) for data access, it remains arguable whether it is appropriate to rely on the amendment of the administrative rules to bypass the PDPA. Overall, relying on internal rules would not be a long-term solution to the NHI Open Database.

Furthermore, the EU GDPR stipulates that pseudonymized data remain personal data to be protected. According to the GDPR, the process of de-identification is regarded as an appropriate safeguard for data processing that cannot “replace consent” as a sufficient condition or a lawful basis for the reuse of data. In practice, the operating standard for evaluating how the data have been processed to be “non-identifiable” remains to be established in Taiwan. If the data are not truly “anonymous,” according to the EU standards, they are still personal data and fall within the scope of the GDPR.

Thus, are there any possible ways to expand access to the secondary use of the NHI data? The advice can be mapped out in the following five directions:

  1. 1.

    Incorporate a dynamic consent model in the My Health Bank app to allow individuals to consent to the secondary use of their health data. It includes marking their preferences for different types of data, sharing scenarios, applying entities, and access purposes that shall be managed through the sophisticated IT system and modified at any time.

  2. 2.

    Stipulate special laws or allow individuals to opt out of their data being used for research or commercial purposes if made accessible. For example, the Health and Social Care Act 2012 is the legal basis for National Health Service (NHS) Digital’s legal right to collect health data in England. This bill is supplemented by a National Data Opt-Out Programme that allows patients to opt out of their confidential information to be used for research and planning.

  3. 3.

    Refer to the EU GDPR to amend the PDPA in Taiwan. It is advised to set restrictions on “data usage purposes” rather than “applicant entities” to allow private entities to access de-identified data so long as the data processing and use is for the public interest, or for scientific or statistical purposes.

  4. 4.

    Set up a data access committee for data access control and management. It is advised to classify risks associated with different data types and carry out privacy impact assessment and independent ethical review prior to granting access to data.

  5. 5.

    Refer to benefit-sharing arrangements adopted by the NHS on its public-private partnerships. It is advised to require a certain percentage of commercial benefits to be returned to the NHI Fund to balance private interests and public benefits. The purpose is to gain more social trust when considering expanding access to NHI data for commercial use.

Prospects of the Use of NHI Data

The NHI database has so far accumulated more than 70 billion medical records and 3.4 billion medical images. This immense medical database is a precious national asset. The NHI has improved people’s health and wellbeing; its valuable database will help precision medicine and digital health and related industries to thrive. However, although confronted with the conflicts among innovative values, the interests of using medical big data, and human rights issues, the NHIA will find a balance between pioneering and conservative policies more pragmatically and create an environment that is more conducive to the health and happiness of the next generation.

In order to have an open communication with the public, the NHIA has started a “use and sharing the NHI data” survey in “My Health Bank,” which has nearly seven million users, to collect public opinions on using the NHI data for academic, research, and commercial purposes. From 9 July 2021 to 18 October 2021, more than 80,000 people filled out the survey. Among them, 92% support the use of encrypted NHI data for academic and research purposes, and 85% support expanding the data access to commercial entities, hoping to facilitate the development of digital technology.

In order to be in line with the EU GDRP, the NHIA has conducted the “Personal Data Use Notification Consent” trial survey. More than 50,000 people expressed their willingness to make decisions on their personal NHI data. If options are given in the future, more than 82% of the people are willing to provide their personal health data for academic use; 83% of the respondents agree that the NHIA should open access to personal data for commercial entities under the condition that personal data are completely de-identified and it should be impossible to identify individuals.

The preliminary survey found that 80% of the respondents have a positive attitude toward the application of the NHI data, which shows that people have confidence in the importance of the NHI data for industry development. The NHIA will continue the public communication for the policy of opening up the NHI data to strengthen social trust. We look forward to driving the digital transformation of Taiwan’s healthcare industry through public-private partnership and cross-domain cooperation to enhance the health and well-being of the people.

Academic Applications of the National Health Insurance Database

The medical claims data have been stored in the medical data warehouse built by the NHIA since 1995. The data accumulated over 27 years include the medical information of all insured people in Taiwan. Unlike Medicare and Medicaid in the United States, which only include subsets of the population, or the NHS in the United Kingdom, in which people can opt-out of the centralized database, the NHI’s data system is a medical database of the whole population, representing the real-world data of people receiving medical services in Taiwan. To promote the development of medical and health care, the NHIA offers encrypted and de-identified NHI data to academic institutions and healthcare researchers to apply, while protecting people’s privacy rights, data security, and reasonable use. Many academic research results have been published in academic journals. Furthermore, to promote the sharing of NHI data and reduce the duplication of resources invested, the NHIA has compiled 6550 international journal articles published from 2005 to 2021 that utilized Taiwan’s NHI data and has established a search engine for these articles. They are classified into ten categories, including “biochemistry, genetics, and molecular biology,” “dentistry,” “health professions, immunology, and microbiology,” “medicine,” “neuroscience,” “nursing,” “pharmacology,” “toxicology and pharmaceutics,” “psychology,” and “social sciences. ” The search engine is designed in a user-friendly and intuitive way. Users can directly apply filters for keywords, journal, article title, author, publication period or subject and easily search for journal articles using Taiwan’s NHI data on the official website of the NHIA.

In addition to displaying bibliographic information, the query results also direct the users to the PubMed website for the full text so that researchers or those who are interested in using the NHI database can quickly learn from the experience of previous researchers. The previous studies are “the shoulders of giants” that we can stand on to improve the effectiveness of our studies.

Development and Planning of Decision Aid Tools

According to the WHO Digital Health Guideline released in 2019, the intervention of digital technologies, such as mobile phones, tablets, and computers, improves people’s health and strengthens necessary services, which are critical components in achieving the goal of UHC. Because Taiwan’s NHI database has rare and valuable evidence-based data, the NHIA launched the “NHI Big Data Digital Application Project” to introduce artificial intelligence (AI) into the system and develop decision aid tools through information and communication technology and the integration of mobility, cloud computing, and big data.

For healthcare providers, the clinical decision support system enables physicians to make more precise decisions regarding clinical treatment, providing a solid foundation for the development of precision medicine and holistic health care. For example, physicians can evaluate follow-up medication more accurately by reviewing the monitoring record of patients’ personal blood international normalized ratio (INR). By tracking personal health status, patient decision aids (PDA) serve as a reference for patients or the general public when they make self-care decisions, such as blood sugar control during pregnancy and suggestions on how to deal with COVID-19-induced emotions for adolescents.

Under the core value of data governance, the NHIA combines the NHI database and emerging technologies to establish smart medical care services, hoping to address health issues that are of concern to medical providers and people and provide people with more convenient and efficient services with digital technology. In response to the core concept of smart government development, the NHIA continues to improve public services with open data and implement “open data transparency and maximized value-added applications” to achieve goals of continuous quality improvement of medical care services and more efficient use of medical resources.

Developing Virtual NHI Cards to Meet the Needs of Atypical Medical Fields

The Virtual NHI Card falls under the big project of Smart Government, which supports the development of “Digital Nation.” In line with the national digital transformation and in response to the need for digital healthcare, the NHIA has virtualized the NHI cards and expanded the virtual NHI Card Pilot Project in 2021. People can now use the virtual NHI card when seeking medical treatment in atypical medical fields, such as “home-based medical care,” “teleconsultations,” and “telecommunications.” (Fig. 12.1).

Fig. 12.1
figure 1

The virtual NHI card supports the development of “Digital Nation”

By the end of 2019, the COVID-19 pandemic deeply impacted the healthcare and medical systems in many countries. The healthcare industry has introduced Information Technology to upgrade their service models and provide solutions for safer and easier ways of delivering medical care. The virtual NHI card is an excellent example of how physical contact can be reduced when combining telecommunications and medical care to improve convenience and reduce the risk for people seeking medical care.

The use of the virtual NHI card through the entire outpatient visit flow, including registration, consultation, and medicine pick-up, has been fully tested. The NHIA will continue to refine the service process to improve connectivity and comprehension. Multiple virtual NHI card reading methods will be developed, such as self-scanning and scanning by mobile device camera, to decrease the necessity of a QR code scanner, hoping to establish a more convenient and safer service model.

To optimize the convenience of home-based medical care, the NHIA added the virtual NHI card QR code function to the “home-based medical care Bluetooth App” in August 2021. Before leaving their facilities to provide home-based medical care services, medical professionals can complete the authentication of the medical institution card and medical staff card first. At the patient’s home, they only have to scan a patient’s virtual NHI card’s QR code through the Bluetooth App to query the NHI MediCloud System, give a consultation, and upload the data. In this way, the equipment that medical professionals need to carry for home-based medical care will be reduced and the process will be simplified.

In the future, the NHIA will leverage the full-scale roll-out of the virtual NHI cards for “home-based medical care” to meet the medical needs of the public and medical institutions and gradually expand the implementation to cover all services that the NHI provides. In the long term, the use of the virtual NHI card will be expanded to outpatient, emergency, and inpatient settings to comprehensively promote their utilization for medical treatment.

The development of the virtual NHI card protects the right to receive medical treatment of people residing in remote areas or areas with insufficient medical resources. It also plays an important role in epidemic prevention by meeting the medical needs during the epidemic. In the post-epidemic era, the development of a digital medical system can reduce the risk of unnecessary exposure during medical treatment and reduce the burden on the medical system. We believe that the virtual NHI card will be a useful tool for identity verification and will become a powerful platform for public services.

Artificial Intelligence Application Pilot Project of National Health Insurance Data

From 2018 to 2020, the NHIA and the National Development Council jointly implemented the Asia Silicon Valley Development Plan to promote the application of AI in medical imaging. This project, which was divided into three phases, gradually opens the NHI data up to external organizations for value-added applications through an industry–academia cooperation mechanism. Therefore, the NHIA started the “NHI Data AI Application Services Pilot Project” in June 2019 to embody the core value of the plan, promoting the NHI data sharing and social innovation services while protecting personal health privacy. The plan allows academic and research units to team up with industry and apply for de-identified medical images, such as CT and MRI, to test or develop AI application models.

The purpose, rationality, and relevance of each team’s proposal must be viewed by the NHI Data AI Application Review Committee, and the de-identification mechanism must be confirmed to guarantee that no individual can be recognized before the team can access the NHI image data in the designated locations. A total of 15 teams participated, including 9 medical organizations and 6 academic institutions, and 10 were collaborations between industry and academia.

The research designs of the teams mainly included AI model verification and construction. The former is for research teams to verify the validity of the AI model developed with single data sources with real-world data or adjust the model parameters through the NHI database to broaden the model’s applicability. The latter uses the NHI big data to develop AI models. The main research topics include a disease prediction reference model, auxiliary verification, and correction of medical image location and lesion interpretation, establishment of an early disease prediction model, enhancement of computer image organ segmentation and labelling model validation, postoperative risk prediction model data expansion, as well as prediction of the incidence of lifestyle-related diseases.

The AI models developed or verified by the participating teams have achieved outstanding results. Among them, the world’s first AI pancreas diagnosis auxiliary tool, PANCREASaver, which provides early warnings of pancreatic cancer, was published in a top international journal—The Lancet Digital Health. One medical institution has introduced AI models and coordinated them with the green channel mechanism to arrange inspections in hospitals. With the assistance of AI, the patient’s brain metastases can be accurately identified early. Moreover, by largely reducing the patient’s waiting time from initial diagnosis examination to return visits for reports, doctors can utilize the golden treatment period and achieve better outcomes. Another successful example is the “Chest X-Ray Image Assisted Research System for COVID-19,” which the NHIA and National Cheng Kung University developed jointly. The system uses de-identified X-ray images of COVID-19 in the NHI database to balance the performance of the images from different disease stages. The system can detect COVID-19 with an accuracy rate of 92%. In the future, the NHIA will integrate an image AI service on the virtual private network platform for medical institutions to upload chest X-ray images and leverage the power of the “Image AI National Team.” Such a “scientific and technological anti-epidemic sentinel” mechanism adds value to the epidemic prevention technology developed.

With the ageing of the population and the changing of lifestyles, the application of innovative technologies, such as information communication and AI, to change the delivery of medical services is one of the most crucial issues internationally. In response to expectations from the public, the NHIA opened the second phase of the “NHI Data AI Application Service Center Pilot Project” in March 2021. By mid-September 2021, three teams had submitted applications and successively participated. To provide more convenient medical services, the NHIA plans to discuss the important results with the research team and evaluate the feasibility of introducing their technology to the NHIA. This will improve the effectiveness of the NHI medical claim reviews and enable the NHIA to provide AI-assisted models on public platforms for all medical institutions to use so that the whole population can receive better medical care.

Conclusion

The importance of the NHI big data and its research contribution is non-negligible. The claims data collected in the past 27 years are among the country’s precious assets. Facing the issues of human rights protection, the point of balance must be reached between innovations and securing people’s rights. Further, making good and proper use of the NHI data cannot be avoided in the fact of the public interest. The NHIA will continue to find a balance to open up NHI data for research use, aiming to drive the development of related industries and promote the growth of innovative and accurate smart medical systems and health aid tools. By adding value to health data, collective statistical information can have application value. Through resource sharing, activation, and reuse, we can maximize the value of health data. The NHI will utilize medical AI lawfully and carefully to launch a new era of smart medical care that benefits all people.