More than ever, rapid changes in healthcare call for research to enhance the understanding of ways to accelerate innovation, improve health outcomes, and enhance the sustainability of health systems. The global COVID-19 pandemic has dramatically elevated the urgency of better science to generate a deeper understanding of the complex issues pertaining to disease progression, the determining role of multi-disease, and non-traditional factors outside healthcare, such as social, economic, ethnic, cultural, and environmental dimensions on health and well-being.

For researchers, this means that a broader, more expansive view of the complex factors that impact health is required in their efforts to inform better decision-making and health interventions. Such a broader, more integrated view is the aspiration of Human Data Science, a relatively new and emerging discipline. Human Data Science combines insights from human science, incorporating the understanding of disease and biology, prevention and treatment, with enabling technologies and data science capabilities, including AI, machine learning, and predictive analytics, and applies those insights to understanding and advancing human health, including the consideration of social, environmental, and behavioral health elements.

With this backdrop, the IQVIA Institute for Human Data Science convened the IQVIA Research Forum on October 18-22, 2021 – a virtual event that gathered several hundred academic researchers and stakeholders from governmental and public health institutions. The theme for the Research Forum was, Expanding the Aperture to Advance Healthcare by Applying Human Data Science.

It was very exciting and thought-provoking to convene and lead the IQVIA Research Forum this year. We were honored to have many outstanding speakers and panelists join us for intriguing presentations and discussions highlighting both the opportunities and the challenges in applying the methodologies, novel technologies, and diverse, complex data sources required to take on this more expansive and disruptive approach to healthcare-related research.

Key themes in the discussions

The forum focused on a range of current and evolving topics in health policy, population health, and healthcare delivery,  and the implications for research endeavours:

• U.S. health policy initiatives and their implications for stakeholders and research
• Artificial intelligence and machine learning and the development, validation, and adoption of applications in this field, including the application in real-time clinical care and the importance of getting clinicians involved
• Real-world evidence, a rapidly expanding and morphing area incorporating registries and evolving into more agile data innovation hubs and learning health systems
• Public health research, data and surveillance, and the public health infrastructure, and how to get underneath the issues of diversity, inclusion and racism
• Health services research and the role it plays in informing decision-making across the health system

Key takeaways

The Research Forum reinforced the exciting opportunities for taking research down the new paths of a more open landscape of forces, factors, and determinants, but it also emphasized the unique challenges posed when embarking on applying new methodologies, novel research enablers, and using data science when integrating and linking human biology with complex social and behavioral elements.

As stated by one of the speakers: “It’s not about the technology, it’s about sociology.”

Health policy – stalemate for now

The discussion about the Biden administration’s health policy priorities highlighted the challenges in developing and enacting policies around health insurance - including the public option, expansion of Medicare, and drug pricing, caused by the political stalemate in the U.S. Congress. The question was raised whether policy changes can happen through regulation by government agencies instead of legislation. While there are expectations around the positive role Centers for Medicare and Medicaid Innovation (CMMI) can play as drivers of innovative change, for example, the outlook for sustained changes in policies around value-based purchasing generally looks dim.

The challenges around implementing health policies were also reinforced in the discussion about the implications these might have on stakeholders, including the uncertainties about unintended consequences, and the potentially negative impact policy initiatives in one area may have on other areas. As an example, concerns were expressed about the negative impact on providers and reimbursement rates from efforts to increase tax-payer funded, lower cost insurance options.

The current health policy landscape elevates the urgency of evidence-based research that looks more holistically at how changes impact multiple stakeholders and the broader healthcare ecosystem, rather than being focused on one narrow element or specific stakeholder.

Adoption of AI and ML in healthcare – keeping physicians involved

While there are high expectations about the game-changing role of AI and ML in healthcare, progress is required to validate the evidence, deployment, and adoption of these advanced capabilities in mainstream medicine. The forum provided very compelling use cases and real successes using AI and machine learning to identify low-value care, improve patient safety, and the diagnostic performance of algorithms to identify retinopathy or abnormal mammograms or sepsis, as well as to improve efficiency. But we also heard about barriers that are pervasive and perennial; for example, alarm fatigue.

Scaling matters, and the fact that EMRs have not been built for real-time predictive analytics as the primary goal, represents a barrier. There was a very compelling anecdote about a COVID algorithm that not only didn’t work well across other institutions but, in fact, stopped working well within the very institution where it was developed because core standards and patterns of care for COVID-19 patients changed so much that the algorithm quickly became obsolete.

We also heard a lot about the importance of getting clinicians involved from the ground up to prevent errors and ensure quality of outcomes from using AI and ML to guide clinical care. “Keeping physicians involved in AI and ML is not a nice-to-have, but a must,” as someone put it.

Finally, there was an important discussion about bias in AI and algorithms – racial bias, for example – and how it leads to research outcomes that overlook health disparities, racial issues, and exclude underserved populations.

The discussion also emphasized the importance of healthcare provider systems taking a more structured and comprehensive approach to the development and implementation of AI and ML in both research and clinical care, for example, by establishing formulary committees for algorithms and having a systematic approach to manage and supervise the development and implementation of these. AI is still in its infancy in healthcare, and researchers will play a critical role in validating further progress, tracking adoptions, and evaluating deployment at scale.

Advancing RWE and learning health systems – “facts vs. poetry”

Real-world evidence (RWE) is gaining prominence in the research landscape and is rapidly evolving as new approaches and capabilities emerge. RWE is evolving and morphing together with registries, from traditional to agile, real-world data integration hubs, and the application of RWE to learning health systems. We had speakers from academic medical centers and clinical research and provider networks contributing their experiences working with RWE and registry programs.

The discussion pointed to opportunities for integrating clinical care workflows with clinical study workflows to enable improved quality, reduce redundancies, increase physician and patient participation, and lower costs. Use of Fast Healthcare Interoperability Resources (FHIR) standards have the potential to enable interoperability, joining disparate data systems and enhancing information sharing. However, there are challenges using FHIR as its profiles are not used consistently across development teams, even within the same integrated healthcare delivery system. Furthermore, when talking to different EHR systems, variabilities will exist. Semantic interoperability is hard.  It was articulated as the challenge around “facts vs. poetry,” and the many different ways to “legally” represent the same information and patient data.

There was also a discussion about a common problem facing most real-world data projects at academic medical centers in larger hospitals: They often don’t serve underserved populations; for example, indigent people in an urban area. That role is usually played by a public county hospital, which is typically underfunded and often not very involved or resourced to participate in research.

RWE will continue to rise in importance as part of the evolution of more agile registry programs and learning health systems, but more research efforts are required to advance the constistency, linkage and semantic operability of datasets as well as the inclusion of underserved populations.

Public Health Research – bringing public health into the cloud

The pandemic has brought heightened interest and, in some cases, concern about our public health systems, data and research. Therefore, the forum focused on the innovative approaches that are being adopted and the infrastructure needed to advance our public health disease surveillance system. The Institute also focused on the issues of disparities and inclusion in our public health system and the case for equity-centered care. We had speakers from the Centers for Disease Control (CDC), the American Medical Assocation (AMA), the Office of the Assistant Secretary for Planning and Evaluation, U.S. Department of Health and Human Services, and the White House Office of Science and Technology.

The discussion centered around innovative approaches to government inter-agency collaboration, linking of multiple data sources, and the urgency of sharing information during a pandemic across federal agencies, states and healthcare organizations.

The session also provided updates on new efforts to strengthen the biomedical ecosystem in the U.S. through the establishment of the new Advanced Research Projects Agency for Health (ARPA-H), as well as advances in public health surveillance and intervention through data modernization and the development of a new CDC data management hub, which is intended to “bring public health into the cloud” by creating a digital infrastructure and using advanced analytics to enhance public health surveillance.

The intersection of public health research with digital technology and advanced analytics and the efforts to enhance innovative interagency data sharing and collaboration provide promising opportunities for researchers to pursue public health science.

Disparities and inequities - advancing radical statistics

The issues around advancing research to address disparities and inequities in research played a central role throughout many of the sessions.

Several speakers centered their comments around efforts to apply the power of data and evidence in public health research to address the issues associated with health disparities and racial inequities – spanning racial bias in algorithms, the exclusion of underserved populations in RWE and health registries, and closing the gaps in health services research.

The point was made forcefully that while leading medical journals have had a remarkable increase in the number of published articles mentioning the term “racism,” fewer than 10% of all articles provide scientific evidence of health disparities and racial inequities. There is a need for what was called “radical statistics,” i.e., research that employs statistical analysis, public health tools, and epidemiology not only to describe inequities, but to change and overcome them.

The volume of research regarding health disparities and racial inequities is rapidly expanding, but there is an urgency to ensure studies focus on the fundamental underlying drivers of health disparities, delivering evidence for the root causes, and the validation of tangible interventions that address these factors.

Health Services Research – reframing the focus, refocusing the frame

Looking back on the 1990s, the dominant frame for health services research was cost, quality, and access.  Now, equity and population health are central themes in health services research.

Regarding equity, health services research is focused on the documentation of disparities in health and healthcare, and the the need to look more broadly at a fuller range of populations. But one of the challenges in health services research is how to get beyond the elementary use of race as a variable to truly understand how to study the impact of race and racism, whether that’s interpersonal, organizational or structural, and the role it plays in health and health outcomes.

Population health research is now centered around the delivery of care to populations and how to deliver the best care to the right people at the right time by the right providers, and includes re-examining the role of primary care that often delivers high-value preventative care at a lower cost, and moving away from a system that is designed to focus on specialist care.

Interdicisplinary and multistakeholder collaboration – essential for high impact research

During the forum, many speakers spoke to the importance of multi-stakeholder collaboration. One speaker made the comment, “We don’t need clinicians who are data scientists.  We need clinicians who treat patients working with data scientists.”  This underscores the point that the collective is smarter than the individual. While this has been understood for decades, tremendous fragmentation remains throughout our healthcare system - whether it is about the importance of improved care coordination and the role of primary care vs. specialty care, real world research and learning healthcare systems and the importance of silo-busting to address pervasive small areas or variations that have proved so vexing in many different settings, or combining clinical research with AI and machine learning.

Therefore, the future lies in both technological and team-focused interventions that help to break down these silos and to improve the collaboration and coordination across different teams and groups of people and systems of care. This is in line with the importance of multi-stakeholder research when looking at policy-related initiatives, so that we’re not only looking at those from the perspective of a single stakeholder. The challenge to the research community is to reflect on how much of the current research is done with that single stakeholder lens, and which doesn’t consider and evaluate the impact of policy changes or other kinds of interventions on other stakeholders. And if those other stakeholders and how they may be affected by a policy decision are not factored in, then the research is really not all that useful.

Broadening the aperture for research – a clarion call

Ultimately, the common, overarching perspective at the IQVIA Research Forum this year is a clarion call for broadening the aperture for research to validate and drive real change in healthcare:

• In health policy research, ensuring that the approach taken is a multi-stakeholder one and combines a multidisciplinary approach with a comprehensive effort looking at all relevant stakeholder audiences and the impact across populations
• In validating and adopting applications of AI and ML, advancing the collaboration between data scientists and clinicians
• In RWE, pursuing registry programs and data hubs integrating clinical research and clinical care workflows
• In public health and health services research, accelerating the exchange of data and the collaboration between public health agencies, public health researchers and data scientists.

Broadening the aperture for research opens up the entire landscape of data, technologies, and methodologies connecting multi-stakeholders to enact collective transformation in health and healthcare.