Data, Data Everywhere: The Solutions

Restaurants , retail and financial services use data analytics a lot. Sick care? Not so much. So, what's the problem?

Medical data and population health scientists, researchers and clinicians face a lot of challenges making sense of medical data. The problems will only grow as we place more emphasis on using data to lower costs and improve outcomes and making The Big Fix-moving sickcare to healthcare. There are a many bumps along the digital health innovation road. 

The COVID pandemic has accelerated the need for data interoperability and transparency. Why is data-sharing and transparency among sickcare stakeholders more important than ever?

• To prevent spread and optimize space, staff and stuff for the sickest patients in healthcare settings, screening and triage apps need to be connected to existing patient data systems.
• Ever-evolving guidelines and emerging evidence requires the ability to quickly update standards of screening and care.
• A fully interoperable healthcare system is key to ensuring the right care is provided to the right patient at the right time. It’s also crucial to evolving the healthcare industry by establishing a foundation for value, with data that is easily available and accessible by patients, providers, and health plans.
• Data interoperability paired with predictive analytics improves health outcomes for vulnerable populations by more effectively and proactively identifying these groups at the point of care.
• Data integrity and analytics-fueled collaboration offer tremendous potential to strengthen the quality of telehealth visits and improve outcomes while reducing costs.
• Creating a single source of truth speeds the pathway to achieving the quintuple aim. 
• Data sharing speeds the development of higher value and validated algorithms
• Interoperability helps us create the tools to change patient and physician behavior and create appropriate evidence based messages e.g. to address vaccine alarmism
• Combining clinical, demographic and financial data might improve the patient and doctor insurance and cost IQ

So, where do we turn for solutions and how do we fill the gaps?

Start by focusing on the 5V's

Data and Informatics. Create and sustain investigators access to high-value data assets that are discoverable, accessible, and reliable to facilitate research in data sciences and discovery by providing informatics expertise and tools in: (1) data management and curation, (2) data discovery, (3) data integration and person level linkages, (4) federated data sharing methods, (5) archival of data repositories, metadata, analytic code, software, and results for data reuse, and (6) support of open science and research reproducibility.

Democratize Analytics To make data science more strategic and democratic in your company, take the following steps. First, focus on problems or opportunities with the highest level of strategic benefit. Second, develop “citizen data scientists” across the organization. Third, reprioritize data science efforts and reassign your data scientists. Finally, develop and communicate a broad vision of data science.

 Innovative Analytics. Build a multidisciplinary, collaborative analytic and computational research environment that includes medical/health scientists, mathematicians/statisticians, computer/data scientists, informaticists, and implementation scientists. This environment will catalyze research resulting in (1) development of novel analytic and computational approaches, methods, and tools, (2) innovative applications of complex analytic approaches, and (3) clinical applications which utilize big health data to create actionable knowledge.

 Patient and Public Engagement. Ensure the success and productivity of the value of research to human health and health care systems by using innovative methods to engage patients and the public in ongoing discussions to address the ethical, legal and social implications of big health data research. Public engagement is key to development of a trustworthy governance framework as well as for the identification of high priority problems and the creation of feasible, usable solutions.

 Transformative Data Governance to facilitate the use of digital health data and other relevant data to benefit individuals and society by addressing concerns and competing interests such as individual/community privacy, business competition, public-private collaborations, and national and local regulations.

Teach health informatics through innovative education, training and mentorship programs for faculty, fellows, and students to develop (1) successful independent future computational (PhD) and medical sciences (MD) investigators and (2) successful non-academic data and computer scientists with special knowledge and understanding of health data issues and applications.

Transform interoperable data into actionable information that is safe, accessible and useful.

Here are three ways to address the medical data scientist training gap

Increase the number of universities offering data science degrees: To accelerate this, employers can partner with universities to help design and fund the creation of these programs.

Offer data science programs for undergraduates: Enrollment in data science programs must radically increase — cohort sizes of 23 students simply will not close the talent gap fast enough. Offering data science degree programs to undergraduates as well as graduate students can significantly expand enrollment. Universities also should consider offering courses online and during evenings and weekends in an effort to reach non-traditional students.

Launch programs that train analysts to become data scientists: Many organizations are flush with analysts who typically have some data science skills in key areas such as statistics. 

Dissemination of the knowledge and scientific products to stakeholders, including to catalyzing successful academic and academic-industry partnerships to transfer data science knowledge and technologies to broad real-world applications.

The future of health data science is bright. The mission is to use data to create high value care and many of us at the Anschutz Medical Campus at the University of Colorado are working to meet the goal.

The vision of D2V@CU is to make the University of Colorado School of Medicine a national and global leader in the development, implementation and dissemination of person-centered, high-value health care by advancing innovations in data and health systems science to improve the lives of patients, families, and communities. 

To achieve this goal, D2V@CU will:

1. Accelerate rigorous cutting-edge data science methods and applications to address pressing health problems
2. Advance the science of person-centered high value health care across the care continuum by integrating different stakeholder perspectives including the UC Health System as a learning and innovations partner.
3. Train the next generation of scientists who can advance and integrate the methods of data science and health systems innovation. 

As noted in a recent McKinsey article, Leaderpreneurs have 8 tools at their disposal to make data work for them. "After all, performance—not pristine data sets, interesting patterns, or killer algorithms—is ultimately the point. Advanced data analytics is a means to an end. It’s a discriminating tool to identify, and then implement, a value-driving answer. And you’re much likelier to land on a meaningful one if you’re clear on the purpose of your data (which we address in this article’s first four principles) and the uses you’ll be putting your data to (our focus in the next four)."

Here are the core components of a data science strategy for sick care institutions.

Here's how Kaiser is treating the 1% who account for 29% of their spending.

Rather than seeking to use AI to deliver care, experts believe that IA or “information augmentation” is the proper first step in using emerging AI capabilities. Using IA-driven applications, administrators, clinical staff and other decisionmakers can access critical information when it is needed, presented in a format that is easy to digest.

IA systems comb through the available data, identify what information within the data is most important, and then deliver it in dynamic visualizations and dashboards that help the user see and understand the message within the data. With continued use, IA systems learn what is important in the data, providing users with relevant, actionable insights.

Informaticists have much work to do to understand how AI can be applied in care delivery. AI should initially be used to narrow the data presented to clinicians to only the most important information, rather than take over the task of directing clinical care.

However, it is a new academic and practice discipline that will have growing pains. It is up to us to be sure it has a proper upbringing.

Lisa Schilling, MD, MSPH is a Professor of Medicine at the University of Colorado School of Medicine.

 Arlen Meyers, MD, MBA is the President and CEO of the Society of Physician Entrepreneurs

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.