The Catalyst for Continuous Healthcare Innovation: A Five-Dimensional Exploration of Learning Health Systems

Over the past few decades, modern medicine has made remarkable progress. Our understanding of diseases has deepened, and a variety of new treatment methods have emerged, bringing new hope to patients. However, in the broad field of medicine, due to the complexity of health conditions and the heterogeneity of patient characteristics, experimental studies—such as randomized controlled trials (RCTs) that employ strict exclusion criteria—are costly and often not applicable to the patients encountered in clinical practice.

In this context, transformation in the healthcare industry is imminent. In light of this, research institutions worldwide are actively investigating an emerging paradigm known as the “Learning Health System.”Institute for Health and Society, Newcastle University, UK(IHS) is an academic research institution with extensive expertise in healthcare services, public health, epidemiology, professional practice, and medical technology research.

Recently, the organization released its research findings funded by The Health Foundation: “Report on the Potential of Learning Health Systems.” The report centers on the development and potential of learning health systems, providing a detailed analysis of their background, constituent modules, use cases, industry impact, and future prospects. VCBeat (WeChat ID: vcbeat) has compiled the main body of the report for your reference; see below for details.

WithMedical Treatment MethodsandClinical Research DataWith the explosive growth of information, today’s medical practitioners can often do little more than barely keep up with the latest developments in their own narrow subspecialties. Due to the sheer volume and complexity of research, even the application of evidence-based medicine approaches—such as conducting systematic reviews—can only partially address this challenge. As a result, many medical practices still rely heavily on researchers’ intuition and biases.

In addition to the rapid expansion of the evidence base, the healthcare industry faces challenges such as population growth and aging, rising levels of chronic diseases, budget constraints, health inequities, and a surge in high-cost interventions and technologies, all of which have led to diminishing returns from health improvements. Furthermore, various unjustified variations in medical practice are becoming increasingly unacceptable.

Meanwhile, in other industries, the Internet and big data analytics have already begun to bring about transformative changes. When thisWhen combined with prognostic improvement measurement and system behavior change techniques, a “Learning Health System” can be established. According to the definition by the Institute of Medicine (IOM), a Learning Health System is “a system in which science, informatics, incentives, and culture are aligned for continuous improvement and innovation, with best practices seamlessly embedded in the delivery process and new knowledge captured as an integral byproduct of the delivery experience.”

Learning healthcare systems can take various forms, but all follow a similar ecosystem chain that includes compiling, analyzing, and interpreting data, as well as feeding it back into practice to drive change.

Regular collection within the healthcare systemData is the driving force behind the movement of learning healthcare systems, but how to translate these data intoCan be processed electronicallyCollecting data in a standardized format remains a significant challenge. Typically, data recorded by clinicians are often incomplete or of poor quality, and coding practices frequently vary across different healthcare organizations, hindering interoperability among systems. Furthermore, attention must be paid to technical and information management issues related to the methods and locations of data storage.

In addition to data within the healthcare system, fromData Outside the Healthcare SystemThe number is also increasing: the emergence of new wearable technologies and online platforms enables patients to take greater control of their own health data. Of course, the importance of such technologies remains to be validated.

Prognostic Improvement Measurement TechnologyThe role of learning health systems will be demonstrated. This measurement is reflected not only in patient mortality rates but also encompasses various levels that significantly influence patient outcomes. Novel automated data collection methods will reduce the cost of organizing such information.

Finally,Behavior Changeis a critical indicator of the success of a learning health system, as true value is realized only when the behaviors of clinicians and patients change. Current advances in behavior change research have laid the groundwork for identifying comprehensive, evidence-based approaches to systemic behavior change, enabling the integration of such approaches into the implementation of learning health systems.

Public concern and anxiety over medical data sharing serve as an early warning of the controversies that may accompany the implementation of learning health systems. Under current ethical frameworks, clinical practice and research are strictly delineated, making it difficult to adapt to learning health systems. In response, a new framework has emerged that assigns moral obligations to patients, clinicians, and researchers.

Learning healthcare systems have been applied in the following six areas: intelligent automation, comparative effectiveness research, positive deviants, real-time monitoring systems, predictive models, and clinical decision support systems.

Intelligent AutomationIt will reduce the burden on clinicians and improve healthcare services by “doing the right things, the easy things.” Intelligent automation includes automating routine care, pre-filling appointments and clinical records, and summarizing case notes.

Learning Health Systems Have the Capacity to TransformComparative Effectiveness ResearchCurrent Status of Comparative Effectiveness Research (CER): The application scope of observational studies and pragmatic randomized controlled trials (RCTs) will expand, enabling faster and more cost-effective filling of evidence gaps. Ultimately, this will generate insights into the effectiveness of different treatments for specific patients. Traditional RCTs will, of course, continue: learning healthcare systems can be utilized to identify eligible patients for trials and streamline the data collection process.

Improved prognostic data will provide a common benchmark for different healthcare providers. With this benchmark in place, provider-level improvements can be achieved through a method known as “positive deviation.” Individuals will identify the “Positive Deviants(i.e., exceptionally outstanding suppliers) and use them as the subject of study to derive successful strategies for “positive deviance,” which can then be disseminated to other relevant organizations.

In DevelopmentReal-time Monitoring SystemIt can be used to track epidemiological phenomena and adverse events associated with new treatment methods. The system utilizes routinely recorded data, enabling more timely machine learning applications.

The quality and necessity of nursing care have long been challenging issues in the delivery of healthcare services. Currently, the applicationPrediction ModelThis enables the identification of instances of low-quality care or unnecessarily expensive care. The impact model developed on this basis can also help determine which cases are most likely to experience alleviation of relevant symptoms.

Clinical Decision Support SystemIt can be used to assist clinicians in managing unfamiliar domains or high-risk situations. The system operates based on machine-readable guidelines and can be integrated into electronic health records (EHRs).

Learning healthcare systems will have a significant impact on the workforce within the industry. While these systems certainly cannot replace clinicians, they will, over time, alter the skill sets required of clinicians and may also affect the types and numbers of physicians needed. Similarly, learning healthcare systems will exert a profound influence on researchers and informaticians.

Furthermore, because this system will be based on or dependent upon the cooperation of healthcare providers, it must be acceptable to them. This means that a learning healthcare system must align with providers’ goals and also secure the support of clinicians, administrators, and board members.

Learning Health Systems will support and evaluateNovel Delivery Modelplay a crucial role. It will provide quality-of-care data that was previously unavailable. In learning healthcare systems,Quality SupervisionIt will differ from the past; the acquisition of more and better data means that people can conduct more targeted examinations and make more timely judgments. It also increases the likelihood that regulators can identify risk factors before adverse care events actually occur.

Learning healthcare systems are alsoValue-Based Healthcare Delivery Agendamatch. In addition to improving the quality of medical services, this system can help address various practical challenges in healthcare, such as the cost crisis, by enabling early diagnosis, personalized treatment, fewer errors, and less burdensome research methodologies. However, given the substantial costs associated with building societal technical infrastructure, no robust economic evaluations have been conducted to date.

Currently, although early learning healthcare systems have emerged, the field remains in a nascent stage of development, with many providers still documenting patient experiences in paper records. As technology continues to advance and mindsets shift, significant progress is expected over the next five years, leading to widespread adoption of learning healthcare systems both within and across large organizations.

Outcome-based reimbursement models may help realize this visionThe development of integrated platforms, standards, and technical infrastructure can enable small organizations to build and deploy learning healthcare systems at very low costs, significantly increasing participation across various types of organizations and promoting the adoption of these systems. Notably, vendors of these platforms may wield substantial power, necessitating regulatory interventions to oversee their activities.

With the strengthening of various foundational elements, the emergence of a broad developer ecosystem, and the growing public and professional awareness of the concept of Learning Health Systems, the timeframe for the “Ten-Year Plan” for Learning Health Systems may be shortened. Ultimately, Learning Health Systems have the potential to transform the delivery of healthcare services.

This “Report” presents a positive outlook on the potential of learning health systems and categorizes it into three themes.

Topic I: Digital Healthcare

Globally, the “digitalization” of healthcare is advancing vigorously. In the United States, incentives such as the Affordable Care Act (ACA) and “Meaningful Use” have accelerated the adoption of electronic health records (EHRs) across relevant institutions. In England, EHRs have been widely accepted at the primary care level. The UK’s National Health Service (NHS) has also set “paperless care” as a 2020 target for secondary care; developments inconsistent with this goal have been incorporated into commissioning frameworks and quality inspection regimes, making compliance with the target a high priority.

Although the aforementioned developments are essential for a learning healthcare system, they are insufficient to launch one. Merely converting paper medical records into electronic versions is not only costly but also fails to create a truly sociotechnical learning healthcare system, thereby preventing the realization of the various use cases mentioned earlier. A learning healthcare system must enable the collection, aggregation, and processing of more relevant data and medical knowledge in digital formats in a more efficient manner. Achieving this is no overnight feat; it will require substantial efforts to ensure that patients and the public are informed about and engaged in the system’s development. Furthermore, human resources will also be a significant constraining factor affecting the system’s development.

Theme II: Development and Deployment of Use Cases and Platforms for Learning Healthcare Systems

The Report previously outlined six use cases already implemented by healthcare organizations. Supporting these use cases requires substantial infrastructure investment, while the development of reusable platform components offers a pathway for the continuous evolution and expanded deployment of learning health systems. However, due to the significant potential of such platforms, their vendors also face considerable risks.

To facilitate broader deployment of these use cases and promote a better understanding of platforms under development, the Report recommends: conducting further independent research to evaluate the potential uses of patient-generated data; identifying additional smart automation scenarios to improve care quality and reduce workload; achieving greater impact potential by increasing the proportion of funding allocated to observational studies and pragmatic trials within learning health systems; having traditional randomized controlled trials (RCTs) consider whether costs can be reduced through automated data collection and whether patient recruitment quality can be improved by using electronic health record (EHR) data to identify potential participants; encouraging vendors to leverage available efficacy data to identify positive deviators; enhancing the utilization of existing surveillance networks; applying predictive models to improve healthcare delivery or operational efficiency; and conducting independent research to determine the types of platforms currently under development, the breadth of their potential applications, and any emerging market issues.

Theme III: Ensuring Learning Health Systems Create Positive Change

A learning health system is not merely about IT and informatics; standalone technological solutions, journal articles, or guidelines alone cannot improve healthcare outcomes at the speed currently required. Improvement is only achievable when the behaviors of patients, clinicians, providers, government regulators, and other stakeholders change within the system’s application. Therefore, any learning health system should adhere to evidence-based theories of behavior change.

In this regard, the Report also provides detailed recommendations, such as conducting further independent studies to evaluate the effectiveness of different types of medical applications; assessing existing research to ensure the validity and reliability of innovative methods for collecting efficacy data; encouraging the application of evidence-based behavior change theories within learning healthcare systems and using them as criteria for funding decisions; adopting the Wilson-Jungner screening criteria, or modifications thereof, when deciding whether to implement predictive models; and conducting comprehensive economic evaluations of both existing and planned components of learning healthcare systems.