The 4631-2 Project, the guidance of health informatization in China, consisted of 3 phases. The first phase was to improve health information technology (IT) hardware infrastructure and the application systems of routine businesses and set up the information standards and security systems. The second phase was to construct the information platform levels according to region levels. The hierarchy and mechanism enabled data exchange between different regions and applications. The third phase was to apply big data technologies, artificial intelligence (AI), and Internet Plus (China’s plan to manage its information superhighway) proposed in a government work report from March 5, 2015, that refers to the application of the internet and other IT in conventional industries [11] and enables advanced applications (eg, predictive modeling, clinical decision support, disease or safety surveillance, public health, and research) [2,12].

The various medical information systems, such as EHR systems, were the fundamental data resources. The continuously improved EHR systems and 3 DHIDs converged the data and gradually became the central resources. The NHFPC had promoted nationwide deployment of EHR systems and regular data synchronization into the EHRD to compile a complete EHR for each citizen regardless of where the original data were generated. Each region’s local HFPC was responsible for the implementation of the EHRD. The survey showed that 86% (18/22) of provincial HFPCs and 75.6% (232/307) of municipal HFPCs had established the EHRD as of March 2017. Meanwhile, 71% (15/21) of provincial HFPCs and 70.4% (216/307) of municipal HFPCs had EHRs for 70% population in the corresponding regions. In addition, 16.19% (3,880,879/23,956,343) of provincial and 14.31% (271,920/1,900,209) of municipal HFPC EHRs were built in 2016 alone.

Data from 6 types of applications and related systems were organized into the other 2 core databases, the DID and EMRD. The survey showed that 96% (22/23) of provincial HFPCs had built DIDs and 48% (10/21) of provincial and 31.6% (97/307) of municipal HFPCs had started constructing EMRDs. The NPHIDP13 had explicitly required the data from 6 applications and other related systems to be available to 3 DHIDs. It had also required that the 3 DHIDs should be integrated following the people-centered principle. Meanwhile, NPHIDP13 had explicit criteria about system consistency, accuracy, and integrity.

Centralized medical data had significant potential for applications and research. Enabling information sharing within and across regions had been attempted by a few countries and was demonstrated to be difficult for various reasons [13-17]. The 4631-2 Project supported regional informatization that collected data from 6 primary types of applications, 3 population health information databases (PHIDs), and other public health systems. Systematic informatization also promoted sharing data within and across regions. As the development of 4-level information platforms was the core task of regional informatization, the platforms’ construction had been one of the key tasks in the Instructions for Accelerating Population Health Informatization [1]. In the 4-level information platforms, most data were generated at the provincial and municipal platforms, so these 2 levels of the platforms naturally became the focuses of the survey. As of March 2017, 32% (7/22) of provincial and 35.2% (108/307) of municipal HFPCs had built the information platforms. All provincial and all but 17 (6%) of the municipal HFPCs had started establishing information platforms. More importantly, all 32 provincial platforms had been connected with the national platforms.

The 4-level information platforms offered 2 advantages. The first was centralization of storing and sharing health information. The 4-level information platforms also connected public health information systems with medical and health service institutions within regions. The integration and sharing within the information platforms would directly improve the management efficiency of health administrations, for example, by providing panoramic information of the entire country during an epidemic breakout. The second was to unify management for EHRs. Based on the information platforms, unified management could facilitate various nationwide standardization.

Advanced applications based on the converged data and technologies like big data, AI, and Internet Plus were considered to have great potential to benefit the people and the health care industry [18-20]. The NPHIDP13 plan had been designed with this consideration. The survey showed that 67% (16/24) of provincial and 31.8% (100/314) of municipal HFPCs and 32.35% (647/2000) of hospitals had implemented various big data applications. Business intelligence reports, medical quality control warnings, clinical medical records, and related data retrieval had been well received by the users. A poll within the survey showed that clinical quality control and chronic disease management based on big data were most appreciated by medical professionals. At the same time, the poll showed that 1180 (59.00%), 1062 (53.10%), and 944 (47.20%) of the 2000 hospitals expected more funds, consistent standards, and interinstitution cooperation, respectively, regarding big data research and application development.

The combination of big data, AI, and Internet Plus still needed to overcome technical and technological challenges to reveal the true potential and support the most demanded applications (eg, precision medicine and AI-assisted clinical decision support systems) [21-23]. One of the issues was the collection and processing of multimodel data from various sources. The NPHIDP13 had tried to structure the data as much as possible. The Big Data Technology in Health Informatization Questionnaire showed that 45.10% (902/2000) of hospitals had structured health big data, 41.40% (828/2000) had semistructured electronic medical record health big data, 28.50% (570/2000) had unstructured medical image data, and 6.00% (120/2000) had unstructured omics data. For regional HFPCs, 24% (5/21) of provincial and 15.9% (50/314) of municipal HFPCs had unstructured medical image data, and 48% (10/21) of provincial and 1.9% (6/314) of municipal HFPCs had unstructured omics data. Meanwhile, the unstructured data, which could be 80% or 85% of overall data, needed more professional curating staff and technologies like natural language processing and image processing to automatically process it [24,25].

The national health informatization project was extremely complex and expensive, but the Chinese government played an essential and supportive role in the project: increases were not seen in local medical service prices and data privacy and usability were better assured. As of March 2017, the 32 provincial platforms received ¥1.87 billion (US $276.6 million) from the central government (not including funds from local governments). During the 12th Five-Year Plan, the central government invested more than ¥10 billion (US $1.5 billion) into health informatization [1,3,6].

Since health care reform in 2009, the number of doctors (approximately 1,730,000 in 2015) has increased by 3.3% annually. The annual rate of new nurses is even larger (9.9%, approximately 472,000 in 2015). The increasing rate of pharmacists is 3.1% (approximately 109,000 in 2015) [26-28]. In comparison with the increasing number of medical professionals, the number of IT professionals is relatively low. Often, the responsibility of the IT departments must be limited in coordinating the needs with the health IT service providers due to a limited number of IT staff. According to the survey, 19% (4/21) of provincial and 85.0% (267/314) of municipal HFPCs had fewer than 9 IT staff members. For the hospitals, only 7.00% (140/2000) had more than 20 full-time IT staff members; the number of IT staff was not enough to meet the actual work needs. The survey showed that 76.00% (1520/2000) of medical institutions expressed the lack of IT staff and the difficulty of recruitment.

Standards, motivation, and credibility were 3 keys to effective interoperability of health information technology [29]. Standards provided the possibility of efficient communications, motivation promoted the usability and consequently created more data, and credibility ensured a trustworthy environment for using data. China’s information standardization framework comprises 3 parts: data standards, technology standards, and security and privacy standards (Figure 2). Since 2010, 283 national health informatization standardization projects had been approved and initiated, ranging from data acquisition and exchange to information management, storage, cataloging, and security, etc. Among the approved standards, 33 have been adopted by consortiums. Of 250 planned industry standards, 209 have been published, 16 will be published soon, another 16 have passed preliminary assessment, and 10 are under development.

Informatization standards not only cover hardware-related issues such as data acquisition, storage, and transmission, called hard standards, but also software-related issues such as metadata standards, information usage standards, supervision and certification, called soft standards. Generally, soft standards are relatively immature compared with hard standards. Furthermore, hard and soft standards should be better integrated within the information standardization framework (Figure 2). Many existing information standards still play separate roles, forming the biggest barriers for interoperability. With awareness of these issues, the NHFPC is making efforts to improve the synergy of health informatization standards to enhance interoperability.

To help the regional HFPCs and hospitals stay informed about standardization and better implementation of standards, a series of certification programs has been launched by the NHFPC. The National Medical Health Information System’s Interconnection and Interoperability Standardization Certification was started in 2013. The certification covers a wide range of data standards, document sharing, platform functions, hardware infrastructure, information security, business applications, etc [30]. As shown by the survey, 9.1% (24/263) of municipal platforms participated in the certification in 2017. Because the program for provincial platforms just started, only one provincial platform has participated. Meanwhile, 46% (8/17) of provincial and 31.3% (77/246) of municipal HFPCs have a clear timeline for taking part in the program. Figure 3 shows the regional HFPCs’ planned schedule of participating the program.

Compared with 59.9% of eligible nonprofit hospitals that get funding in the United States [31], almost all major hospitals in China are public and nonprofit. The Chinese government is responsible for funding support and therefore plays a more active role in health informatization.

One of the key achievements in recent years was the development of nationwide 4-level information platforms. It was an ambitious and challenging task that established a certain level of connectivity between HFPCs in the country. This might enhance information sharing and collaboration and may directly support allocating health care resources according to regional differences. For medical professionals, it largely improved information accessibility—for example, to be able to make rapid diagnoses based on more complete information or reduce the number of reexamination cases. For patients, it improved the accessibility of health information and enabled interregional medical health and insurance services. For health administrations, it delivered real-time intelligence of dynamic situations about health service resources to support scientific management, such as inspecting and maintaining reasonable medical expenses. The potential of converged data was still far from being fully exploited and many more data applications were still under development. The survey showed that 71% of provincial HFPCs regarded developing and improving applications as a main task for the future.

Another more serious problem is low data quality. According the analysis of the survey, empty or invalid data, disorders of disease description language, lack of retrieval tools, lack of metadata management, and inability to handle the text data of electronic medical records are prominent problems in China’s health informatization. Only 66.8% of municipal NFPCs implements or plans to implement data quality management system standards. The problem is serious and may cause unexpected barriers for the usability of the 3 databases and the platform. Without high-quality local data, centralized medical data on the 4-level information platforms won’t play a real role and advanced applications based on converged data and technologies like big data and AI will not have a solid foundation. The survey also reported that 74% of provincial HFPCs put this at the top of the list of problems to be solved. To summarize, we list the best practices and corresponding barriers and successes of the national health informatization implementation in Table 1.

Health informatization in China started in the relative recent years and has developed at a very rapid pace. Its objective is to enable advanced clinical research and applications at the beginning while the IT infrastructure is built. Initially, informatization aimed to improve efficiency and administration of finances. Gradually, informatization transited from finance-centered to people-centered in order to bring full potential in terms of health outcomes. The transition was naturally occurring while the goal of informatization was explicitly set to bring more benefits for every patient in NPHIDP13.

According to NPHIDP13, personalized health care services and products based on perception technology will start being applied and promoted by 2020; DHIDs storing individual lifelong records for the entire population will be eventually built; and novel technologies, which might bring revolutionary outcomes in health care and research, were considered while designing the current systems and planning future advanced applications. Generally, NPHIDP13 planned the projects for the period of 2016 to 2020 to solve remaining problems. They include upgrading current systems with new technologies, continuing development of applications of key businesses, improving coverage of the population in 3 DHIDs, and improving interconnection and interoperability between the 4-level platforms.