The 2019 Healthcare Analytics Summit: Thursday Recap

Thursday – MORNING GENERAL SESSION

19 – Day in Review

Paul Horstmeier – Chief Operating Officer, Health Catalyst

Paul Horstmeier opened day two of the 2019 Healthcare Analytics Summit™ (HAS) conference with a recap of day one. Health Catalyst Analytics SVP Dan Lowder shared photos from Wednesday’s fun run and a light-hearted review of the “rules”: sleeves are required, no excessive celebration, stay off the grass, there are no winners (and no losers), and lastly, no interpretive dance. Analytics Vice President Emily Tew said that analysts have data hunches, but they don’t know if they’re right or wrong, until they look at the data.

She shared some of analytic highlights from HAS 19 so far, including that many attendees predicted Wednesday’s session would end on time; in fact, the session ended two whole minutes early! Lastly, Mr. Horstmeier shared the best #SocksofHAS, which featured Bob Ross, peanut butter and jelly, unicorns and hearts, skeleton feet, and girl superpower.

20 – Healthcare AI: Are We There Yet?

Marianne Slight – Healthcare Analytics and Machine Learning – Product Management Executive, Google

When it comes to healthcare AI, Marianne Slight asks the question, “Are we there yet?” The answer, of course, is “not yet,” and the distance we still need to travel to get there varies greatly from system to system. While many people envision hospitals as futuristic technology hubs, Slight said many are still a “mix of clunky non-interoperable desktops and fax machines.” Technology is not where it should be, even in leading hospitals.

AI can answer many of the questions that consumers are asking. Marianne walked the audience through Google’s five-step journey of applying AI to healthcare data:

• Step 1. Identify which problems you want to solve, but more importantly, the change in the decision making that you want to enable.
• Step 2. Define your starting point as a healthcare organization. Organizations tend to be behind the curve with technology, clearly impeding the end goal of seamless data transferability.
• Step 3. Address interoperability and infrastructure. People are particularly interested in how to leverage existing AI, but the challenge lies in interoperability and lack of appropriate infrastructure to support AI.
• Step 4. Applying AI adds another a layer of complexity; for example, the struggle to map and transfer data between different systems.
• Step 5. Work toward data harmony. Google has also used AI to improve health search results, including better curating content and testing algorithms with data farms.

Marianne wrapped up her address with an AI update at Google—they are not “there yet” but have seen more accurate models emerge throughout the clinical research, demonstrating they are making progress in applying AI to healthcare data.

21 – Digital Innovators Panel: A New Era of Pharma Collaboration—Pharma’s Evolution from Products to Outcomes

Jessica Federer, MPH – Former Chief Digital Officer, Bayer

When discussing pharma collaboration, Jessica Federer likes to start at the end so she can remember the ultimate goal: that a learning healthcare system improves with every patient interaction.

Throughout her career in pharma, she had the opportunity to utilize data in every aspect. However, as CDO of Bayer, she quickly learned that a digital transformation wasn’t about technology but people. She said that if employees understand the latest innovative technology, they will adopt it on their own. Educating team members and providing them with the insights they need allows team members to take the wheel. Digital transformation will follow.

According to Federer, the secret of working with pharma is process. Her experience taught her that process is the only way pharma can exist. Process allows pharma to take a molecule and deliver a medication to someone’s medicine cabinet over a ten-year period. Therefore, the only way for healthcare organizations and providers to work more closely with pharma, and even influence pharma, is through process.

The following are the ABCs of influencing pharma processes:

• Authorities: if you want to see how fast pharma can change their process, bring in the healthcare authorities.
• Barriers: changing the barriers, guidelines, and standards helps pharma adapt accordingly.
• Cost: cost is a powerful force for collaboration with pharma.
• Data: pharma can now partner for direct, near real-time, auto-syncing integrated data; keep innovating in data interoperability and usability.
• Ecosystem: intentionally disrupt the ecosystem pharma has created and change their influence.
• Future: Pharma plans decades ahead, making future projections critical to resource allocation; health systems can gain influence if they create the future and commit to a learning healthcare system.

Federer closed by emphasizing each audience member’s ability to personally influence pharma, but only through the application of the ABCs.

21 – Digital Innovators Panel (Continued): Machine Learning, Data Democratization and My Generation’s Future

Justin Aronson – Junior in High School and Founder of variantexplorer.org

“This is what standing in front of a room full of people who can change the world looks like,” said Justin Aronson, the first-ever high school keynote presenter at HAS. However, he went on to say that his generation is inheriting a lot of problems from HAS attendees: climate change, social justice issues, and of course, many problems in healthcare. Aronson became interested in the possibilities and capabilities of machine learning to help solve some of healthcare’s most pressing issues. He focused on one issue in particular, the discrepancies in genetic testing between laboratories, which can have life-and-death consequences. He created the website variantexplorer.org to display and share his findings of the discrepancies in ClinVar (NIH’s repository website that aggregates information about genomic variation and its relationship to human health).

Aronson told attendees there are two things his generation needs to help improve healthcare are:

1. Roadmaps: act as guides to help people understand the role of machine learning in improving societal issues; without a proper framework or roadmap, it is difficult to determine the optimal path to achieve an end goal.
2. Data democratization: protect and promote the ability to openly share data to tackle social issues, such as deadly diseases.

Aronson hopes that today’s data mindset of data monetization will eventually shift to data altruism, in which people will use data to overcome society’s most pressing issues. According to Aronson, the collective brain power of data scientists can help overcome these data challenges, making data more accessible so people can learn how to use the tools and better protect peoples’ data. He finished with a plea to the group that decisions they make today, about data in particular, will influence everything in the future.

Thursday – Breakout Sessions

Wave 3- 22 – AI for Healthcare Leaders: The New AI Frontier for Improved Leadership Decision Making

Jason Jones, PhD, Chief Data Scientist, Health Catalyst

AI has historically been known as “artificial intelligence”—but Jason prefers the term “augmented intelligence.” While AI has focused first on the models needed, “augmented intelligence” shifts to focus first on an organization’s values and objectives, then augment how we will achieve those objectives using AI tools.

AI gets a lot of attention for contributions at the point of care/point of service; however, there is also a significant opportunity to help leaders achieve focus, allocate resources, and be accountable for decisions.

What do analysts and data scientists need to do to affect change at the leadership level? First, separate signal from noise while focusing on the future. Analysts can interpret the data and provide forecasting that shows where a specific area is seeing improvement—and where improvement is still needed.

One unexpected application of AI for leaders is to focus on quality. The IOM (captured by AHRQ) measures healthcare quality by six domains: safe, effective, person-centered, timely, efficient, and equitable. Although equitable is listed last, it provides a significant opportunity to support leadership with data. Equitable care doesn’t vary based on gender, ethnicity, geography, and socioeconomic status. Using AI tools, analysts can assess whether specific areas of care are equitable—and improve care for those populations.

Using AI tools, analysts can support leaders by providing focus—by identifying inequity, resourcing opportunities, and more. Leaders can encourage these attempts and remove fear, while building trust in the data.

Wave 3- 23 – The Super-Powered Medical Group: How Data Powers the Connected System of Health (Innovation, Clinical, Financial, Course Level: Intermediate)

Scott Shreeve, MD, Chief Executive Officer, Crossover Health

Dr. Shreeve of Crossover Health parallels his vision for the connected system of health with power generation—paralleling how the Glen Canyon Dam uses water, starting with a single drop, to power cities and beyond with how each health data point powers better care. The connected system of health, he says, is the framework for achieving healthcare’s Triple Aim, with a focus on employer health activists. Large employers’ healthcare costs are rising, and 85 percent of their spending is not on primary care. These health activist employers can achieve significant savings with more effective, data-enabled management.

Crossover Health is leveraging multiple diverse data streams to create a large data lake that powers next-generation primary care. Moving from a reactive, visit-based, sick care model to a proactive and predictive care model built for the digital health. A trusted national medical group with advanced, said Shreeve, has superpowers.

Wave 3 – 24 – More Impact, Less Burden: How Partners HealthCare Is Advancing PCMH Through Next-Generation Analytics

Colleen Blanchette, Director, Population Health, Partners HealthCare

Salina Bakshi, MD, Internal Medicine, Massachusetts General Hospital

What is a patient-centered medical home (PCMH), what benefits does it aim for, and what can your organization learn from the PCMH journey of Partners HealthCare?

In this informative session, Colleen Blanchette and Salina Bakshi began by providing useful background on PCMH and NCQA accreditation for PCMH. They described how Partners established PCMH and achieved National Committee for Quality Assurance (NCQA) accreditation—but ultimately elected to implement an Advanced Primary Care Model, which replaced the NCQA process with a less-burdensome and more aligned structure for their primary care practices.

Endorsed by numerous professional organizations (AAFP, ACP, AAP) and supported by research, PCMH supports better care, cost, and experience outcomes by increasing access to care, focusing on prevention and care coordination, and taking a “whole person” approach. PCMH strengthens the larger health system and is part of payment reforms.

NCQA’s PCMH accreditation program provides a set of PCMH criteria and requirements for participation. These criteria guide practices undergoing PCMH transformations—and standards are critical for ensuring clinical quality. Initially, Partners used the NCQA definitions for their PCMH standards.

Can they do better?With most (96 percent) of Partners’ primary care physicians practicing in a formally recognized PCMH, the organization has shifted away from an NCQA focus and toward the Advanced Primary Care Model. This model reduces the administrative burden associated with NCQA and aims to leverage advanced analytics to drive focus on primary care priorities.

As Partners’ multiyear journey continues, Colleen Blanchette and Salina Bakshi offered session participants these lessons for other organizations seeking to transform primary care:

• Measure what matters most.
• Take a phased approach.
• Don’t expect it to be easy, but don’t give up.

Wave 3- 25 – Improvement Science and Analytics: Keys to Improving Population Health
Heather Schoonover, MN, ARNP-CNS, PHCNS-BC, FCNS, Vice President, Customer Success, Health Catalyst

There’s an exponential amount of growth in healthcare data. In this breakout session, Heather Schoonover explored how to effectively use all this data to improve outcomes and decrease costs.

Although the healthcare industry has made progress in this area, unwarranted clinical variation is still a huge opportunity for most healthcare organizations. There are two main types of variation in healthcare: common cause and special cause variation. Common cause variation are causes inherent in a system over time, while special cause variation arises because of specific circumstances. In Improvement work, we try to cause special cause variation, said Schoonover. Session attendees also participated in a group exercise to understand variation.

Schoonover reviewed how to use change concepts to generate improvement interventions, including the following:

• Eliminate waste: eliminate activities or resources that don’t add value to an external customer.
• Change the work environment: consider a high-leverage opportunity for making other process changes more effective.
• Manage time: reduce time to develop new products, wait times for services, and cycle times for functions.
• Error proof the system: redesign the system to make it less likely for people in the system to make errors.
• Reduce variation: reducing variation improves the predictability of outcomes and helps reduce the frequency of poor results.
• Improve the workflow: improving the flow of work in processes improves the quality of goods and services.

Throughout the breakout presentation, Schoonover explored case studies across healthcare organizations to illustrate how the science of improvement and analytics can be used to tackle complex problems, saving millions of dollars, and improve population health.

Wave 3 – 26 – How IT Can Leverage Consumer Trends to Get Health Systems on the Modern Digital Playing Field (Strategy, Innovation, Course Level: Intermediate)

Ryan Smith, SVP and Executive Advisor at Health Catalyst

Ryan Smith opened his session by sharing a personal experience with his daughter at a busy urgent care. While they were waiting for care, he overheard a patient say, “Why is it so hard to get the care you need?” This is a frustrating, all-too-common sentiment, but the words inspired Smith to remember the “why” behind his work in healthcare.

First, Smith discussed digital transformation in healthcare and consumer expectations. Although most companies who also operate in a service-oriented industry digitally deliver their products and information to meet consumer expectations, healthcare is far behind.

A few reasons for this lag are the complexity and nuanced reality of healthcare lack of sophisticated technology infrastructure to support this new digital direction. However, Smith advised that healthcare systems can apply strategies to overcome technology barriers, ride the digital transformation wave, and meet consumer expectations.

Smith’s five key recommendations to leverage IT and keep up with consumer trends include:

1. Define “digital” in your organization.
2. Develop strategic guiding principles for enabling digital.
3. Position around portfolios and govern accordingly.
4. Develop an analogy.
5. Strategically select vendor partners.

Smith reminded that consumers expect and demand a digital experience—whether it’s ordering fast food on a McDonald’s app or seeing a provider. As organizations commit to a digital future and apply Smith’s five key recommendations, enabling digital success will become a reality.

Wave 4- 27 – Panel: The Digitized Patient Experience: How Novel Digital Therapeutics (DTx) Data Will Reshape Your Analytics and Remote Patient Care Programs (Innovation, Life Sciences; Course Level-Intermediate)

Chris Hogg, MBA, Chief Commercial Officer, Propeller Health

Carlos Rodarte, Life Sciences Strategy Business Development, Health Catalyst

Matt Omernick, Chief Creative Officer

Mette Dyhrberg, CEO, Mymee Inc.

Owen McCarthy, MBA, Co-founder and President, MedRhythmns, Inc.

Digital therapeutics (DTx) is a fast-evolving domain focused on creating specialized software to prevent, treat, and/or monitor chronic medical conditions, many of which lack effective traditional therapies.

The new technologies that make these therapies possible also collect large sets of data very fast, fleshing out pictures of patients, their conditions, and the effectiveness of treatments. These new data sources are prompting discussions about data rights and data security. Panelists emphasized that they want to be careful to handle the data in the right way for the benefit of patients. Also, guarding against data breaches is critical—a data breach could ruin a small DTx company.

Keeping patients engaged and compliant can be done with a very light touch in DTx. One panelist said their users required less than 3 minutes of engagement a day to ensure compliance and collect all the needed data. The light touch is important; patients don’t want to be continually reminded about their illness.

A new set of challenges comes with implementing DTx. Payment is challenging because DTx usually lack insurance codes. There are no established pathways to prescriptability because DTx are not orderable via the EHR. Companies are working with health systems and employer–insurers to begin to forge those paths.

Also, creators must convince physicians of clinical effectiveness and, historically, providers and health systems have been reluctant to listen. But that’s changing. Companies that were turned away five months ago are now being invited in for discussion. Hospitals are more interested in the new research. Why? The markets change fast and so does patient autonomy.

Wave 3 – 28 – The Application of Clinical and Cultural Data: Delivering Safe, Optimal Care

Alan Frankel, MD
Managing Partner, Safe & Reliable Healthcare

Michael Leonard, MD
Managing Partner, Safe & Reliable Healthcare

About 30 percent of hospitalized patients have something happen to them that we wouldn’t want to happen to us—and 10 percent are harmed seriously enough to stay in the hospital longer and go home with a disability. So how do we change?

To be successful, organizations must foster a “culture of safety” that evidence shows is a prerequisite for better, safer patient care. In this session, Drs. Frankel and Leonard presented an effective framework and validated tools that provide actionable insights for the development, nurturing, and maintenance of a culture of safety.

The Safe & Reliable Culture Maturity Model is organized around three essential components: effective leadership, a culture of safety, and continuous learning and improvement. Building capacity in these areas enables organizations to move up the Safe & Reliable Culture Maturity Model from UNMINDFUL (“who cares as long as we’re not caught”) to GENERATIVE (“safety is how we do business”). And it starts at the unit level. Aligning leadership with frontline caregivers is essential for success.

Drs. Frankel and Leonard shared several practical tools to enable healthcare organizations to apply the framework:

• The SCORE Survey can be used to assess Safety, Communication, Organizational Risk, Resilience/Burnout, and ­Engagement. It provides detailed unit-level indicators to help organizations create actionable plans for improvement.
• Digital Learning Boards integrate data and communication tools that­ can be used to help create a higher degree of psychological safety and accountability—and make collaboration fun.
• Leadership rounding, team huddles, and personalized kudos can improve engagement and close the loop with front-line staff.

The presenters concluded that when organizations are self-reflective and improvement-capable, they can drive visible, measurable, and sustainable change.

Wave 4- 29: WASTED: What’s Holding Healthcare Back—and How You Can Move Ahead to Transformational Improvement

Presenter: Thomas D. Burton, MBA, Co-founder and President, Professional Services, Health Catalyst

In this lively interactive session, Tom Burton led participants through a collaborative game designed to the illustrate the categories and impact of waste in healthcare and explain how transformations (or mindset shifts) in key areas can help organizations increase efficiency, reduce within-case variation, and improve population health.

In three game rounds, players moved tokens across their gameboards, racing to “remove waste;” acquire “capabilities” by gaining new skills, knowledge, and attitudes; and “systematize jobs to be done” in different areas of a healthcare system. Each of the rounds focused on a different kind of waste—efficiency, within-case variation, or case-rate utilization—and had slightly different rules designed to mimic different incentive structures (fee-for-service, fee-for-value, or a mix of both).

Additional concepts and take-aways surfaced through the WASTED game analogy:

• Improvement via waste removal is your best business strategy—and this requires transformation in key areas.
• Data and analytics are critical to healthcare transformation; indeed, increasing removal of waste corresponds to higher levels on the Analytics Adoption Model. But analytics alone won’t transform your organization. You need to identify the “jobs to be done” in the ideal state, then work to build the knowledge, skills, and attitudes that drive new capabilities.
• It’s important to honor the complexity of your organization—and of healthcare transformation. Burton identified 11 discrete and important transformations in areas as diverse as data infrastructure, payment, patient safety, costing, reporting, and so on.
• Because financial pressure will vary depending on your organization’s current payment landscape, organizations need a strategic approach to transformation.

Wave 5- Session 30 – Rock Your Analytics World (Literally)

John Wadsworth, MS, Senior Vice President, Health Catalyst

Attendees were welcomed to this session with a cacophony of rock-and-roll hits. By the end of the session, the speaker, John Wadsworth, had the audience of 70 percent self-described introverts dancing to the music.

Throughout the session, John drew parallels with the music recording industry to highlight three essential principles:

1. Healthcare analytic ecosystems need tuning. Just as musical instruments need tuning, the people and processes that enable you to capture, model, analyze, and report data must be finely tuned to affect outcomes improvement. Analytics leaders should act as sound engineers—turning down some resources and amplifying others.
2. Not all analytics hold the same value. Analytics leaders must carefully quantify and adjust the investment in and allocation of resources. John used the Analytics Adoption Model as a framework and gave specific examples of what could be done to automate lower-level tasks to enable higher-level analytics.
3. Incentivize early engagers.Just as John tried to create a safe environment for the audience to dance, he challenged the audience to create a safe environment for users to engage with and learn from data.

John ended by suggesting three simple ideas to improve data literacy:

• Schedule a weekly one-on-one, 30-minute meeting with your neediest data users.
• Invite clinical, operational, and technical staff to “open labs” to explore data together—the equivalent of a “jam session” in the music world.
• Annotate graphics with conclusions from your analysis. This takes a little risk—but is a great way to engage users by giving them something to react to.

Wave 4- 31 – Physician Alignment and Data-Informed Decisions Increase Contribution Margins and Market Share
Holly Burke, Executive Director, Clinical Innovation and Quality, Pulse Heart Institute, MultiCare Health System
Needham Ward, MD, Chief Medical Officer, Pulse Heart Institute, MultiCare Health System

Breakout presenters Dr. Needham Ward and Holly Burke are a study in polarities. Dr. Ward, Chief Medical Officer and the “retroverted boomer,” and Ms. Burke, executive director of clinical innovation and quality and the “all-knowing ambivert millennial” play very different but complementary roles in the success of the Pulse Heart Institute, a subsidiary of the MultiCare Health System. The Pulse Heart Institute’s mission was to build a destination center for heart and vascular health in the Pacific Northwest region.

When Burke and Dr. Ward set out on this mission in 2015, they made a commitment to the MultiCare Health system board to move the needle on quality, engagement, recruiting, and economics. They used data and analytics and a structured process to drive improvements in each of these areas and deliver on their promises. The results of their improvement efforts speak for themselves:

• $48 million additional revenue.
• Surpassed the year three market share in year two.
• Overall market share improved in every submarket.
• 11 percent growth in Cath lab volumes between year one and two.
• 19 percent growth in operating room volume.

Lastly, Burke and Dr. Ward shared the following key lessons and recommendations:

• Organizations must be able to quickly obtain real-time accurate performance data.
• Start using data to drive improvements when data is at least 80 percent accurate.
• Aligned provider incentives are a powerful mechanism for increasing collaboration.
• Clinical and operational dyad drives better results than a physician or administrator working alone.
• Get comfortable with failing fast and learn from failure.

Wave 4- Session 32 – Machine Learning Marketplace: 10 Featured Stations

This unique session provided attendees with access to 10 innovative machine learning and artificial intelligence (AI) use cases in a walkabout format. Presenters provided abstracts of their projects including importance, objectives, methods, findings, and conclusions. Participants could visit stations of interest and meet directly with presenters to explore each project more deeply—asking questions about how to get started, where to look for opportunities, and key lessons learned.

Presenters offered the following use cases:

1. Machine Learning Models in Primary Care Decision Support (Acuitas Health)
2. ARUP Test Matching (ARUP Labs)
3. Opening Black-Box Models for Better Clinical Intervention (Humana)
4. Using Natural Language Processing (NLP) and Probabilistic Matching to Improve Efficiency, Reduce Redundancies, and
5. Simplify Data Warehouse Structure (Intermountain Healthcare)
6. Machine Learning Predicts Next-Day Patient Discharges Optimizing Hospital Capacity Management (Massachusetts
7. General Hospital / Partners Healthcare)
8. It’s Gonna Be a Rough Month: Calming Your Executives with Regression Forecasts and Bayesian Updates (Mission Health)
9. Temporal Variation and Anomaly Detection in High-Dimensional Spaces (TVAD-HDS) – (Mission Health)
10. Up, Up, and Away! Getting Grassroots ML to Stick: A Case Study on Nurse Flight Risk (UnityPoint Health)
11. Using SDoH Risk Analytics to Improve Population Health Among a Pediatric Population (UPMC & Socially Determined)
12. Using Natural Language Processing to Automate Quality Measurement (UPMC Enterprises)

Wave 4- 33 – The Doctor’s Orders for Engaging Physicians to Drive Improvements (Clinical, Financial, Course Level: Beginning)

Jack Beal, JD, Vice President, Performance Improvement and Deputy General Counsel, The University of Kansas Health System

David Wild, MD, MBA, Vice President, Performance Improvement, Assistant Professor, Department of Anesthesiology, The University of Kansas Hospital

The performance improvement team at the University of Kansas Health System has a scattered physician group that made it challenging to gain synergy to drive outcomes improvement initiatives. The team worked to solve this problem by first recognizing that the value of the data collected in healthcare is determined by how quickly it can be analyzed and shared. Their focus has been on compressing the timeline to disseminating results. However, the problem gets more complicated when physicians are involved. An “us versus them” mentality was a huge barrier, as clinical departments vary in their support structure for improvement work, and there was a level of distrust between performance improvement and physicians.

How did they do this? Three initiatives engaged physicians:

• Care connections: A physician-led, local improvement program; clinicians send in their ideas for funding approval. Results: 37 approved projects and representation from 17 of 21 clinical departments.
• Value-based performance: New department or decision-level improvements designed to become standard of care after one year. Results: measurable improvement and savings in 80 percent of the projects in the first month and 11 additional projects to go live in September 2019.
• Department finance and planning sessions: Engagement of physician leadership in strategy and planning for growth and improvement. Results: All 21 clinical departments have annual growth and improvement plans in place, and more than 100 additional projects are in the works or under review.

Engaging physicians is essential to success and worth the investment of time. Investing in aligned, not dictated, improvement efforts generates meaningful, successful physician engagement.

Wave 4- 34 – Using a Data-Led Action Framework to Combat Healthcare Burnout: Lessons from Minnesota’s Statewide Model (Strategy, Clinical, Course Level: Beginning)

Rahul Koranne, MD, MBA, FACP, Chief Medical Officer, Minnesota Hospital Association

Tim Sielaff, MBA, MD, PhD, Chief Medical Officer, Allina Health; Senior Vice President, Allina Health Group

With the goal to be the best place to receive and deliver the best care, Timothy, Chief Medical Officer at Allina Health, and Rahul Koranne, Chief Medical Officer at the Minnesota Hospital Association, know that a healthy, happy workforce is critical.

Healthcare burnout is on the rise, and widely considered an epidemic. Dr. Koranne shared the U.S. happiness survey and showed that America’s happiness has greatly decreased from 2006 through 2018. Physicians lead the pack of professions most prone to burnout.

Drs. Koranne and Sielaff and team created an action framework to address healthcare burnout. They also collaborated with a newly hired data analyst to create a custom analytics machine to collect accurate data about physician burnout, including a focus on burnout drivers and components.

The data the problems and the nuances of physician burnout. With a multilevel approach to tackle burnout, Drs. Koranne and Sielaff started at the organization level (targeting systems, culture, and leadership), then practice level (clinic, unit, and/or team), and then the individual. They implemented strategies to reduce physician burnout including investing in physician leaders, improving EMR functions, optimizing provider FTE provider, refining the provider specific employee assistance program, and improving team engagement.

Wave 4- 35 – Get Fit, Get Well: Medically Integrating Wellness Creates a Healthier Community (Innovation, Population Health, Course Level: Beginning)

Greg Stock, MHA, Chief Executive Officer, Thibodaux Regional Medical Center

Katie Richard, MA, BSN, RN, Education and Training Coordinator, Thibodaux Regional Medical Center

Only one in three U.S. adults receives the recommended amount of physical activity each week, while nearly one-third of high school students play video or computer games for 3 hours or more on an average school day and just one in three children are physically active each day. These realities, and many more, impact individual wellness, education, and earning potential and can increase healthcare costs for individuals and society.

Given the current state population health in the U.S., Thibodaux Regional Medical Center aimed to address the health and wellness needs of its community with a medically directed wellness center. Greg Stock and Katie Richard shared how their health system used improvement tools, analytics, and stakeholder engagement to establish a center with physician-guided, patient-centered clinical pathways and multidisciplinary teams focused on proper nutrition, appropriate exercise, and minimization of risks.

The Thibodaux team set strategic community wellness goals for its pathways:

• 30 percent of referred participants completing the eight-week program.
• 60 percent of participants converting to a full fitness center member.
• 80 percent increase in raw Patient-Reported Outcomes Measurement Information System (PROMIS®) score.
• 80 percent of participants with a BMI over 30 losing at least eight pounds.
• 80 percent increase in participant walking distance.

The Thibodaux team learned that an effective community wellness program relies on physician buy-in, preparation for unexpected events, recognition of unique pathway barriers, and preparation for rapid growth.

Wave 5 – 36 – Getting Patient Outcomes Predictions Right: Using Behavioral and Social Data (Innovation, AI, Population Health, Course Level: Advanced)

Imran Qureshi, Chief Data Science Officer, Clarify Health

A person’s health is incredibly nuanced, with social and behavioral components playing a major role. Yet, how do we measure social and behavioral factors? Even better, how can we use social and behavioral data to predict outcomes?

Imran Qureshi, Chief Data Science Officer at Clarify Health, provided 12 key takeaways for using social and behavioral data to predict patient outcomes:

1. Start using new, better AI tools for cleaning data.
2. Augment your death records with external sources.
3. Use weight loss as a feature in your models.
4. Use behavior in models.
5. Use member level social data, not zip code/county level.
6. Don’t rely on questionnaires or doctors for SDoH.
7. Use hypotheses to construct and test engineered features for better models.
8. Try to combine clinical risk with utilization risk into a composite score.
9. Automate data profiling.
10. (Almost) always use relative risk and identify seminal events on a patient timeline.
11. (Almost) always use case studies.
12. Separate people who we can help from those we can’t.

Because behavior drives 40 percent of healthcare outcomes, it is imperative that healthcare organizations include behavioral and social data to deliver comprehensive care and help patients achieve optimum health.

Wave 5- 37 – Getting to the Wrong Answer Faster: Shifting to Better Use of AI in Healthcare (AI, Innovation, Course Level: Intermediate)

Jason Jones, PhD, Chief Data Scientist, Health Catalyst

How can we take the same data and draw different conclusions? Through a series of examples, Jason Jones demonstrated how data misleads by quickly leading to the wrong answer—but through further analysis and by looking at data in different ways (e.g., using different kinds of visualizations), you can arrive at the right answer.

Jones offer the following examples of getting to the wrong answer faster:

• Cancer incidence. Data initially focused on areas of high and low cancer incidence revealed that some areas are more desirable than others to buy a house. By looking at the data in a different way, however, the data appeared on a bell curve, showing a pattern of random incidence. Same data, different results.
• Scared Straight versus education in Kenya. A Cochrane review of Scared Straight showed no impact or negative impact, yet it’s still on TV. On the flip side, ICS (Investing in Children and Their Societies) worked to improve academic performance for Kenyan students. A process of study, trial, and error (more books, flipcharts, more teachers) led ICS to work with the WHO to recognize that students were missing ¼ of the school year because of worms. Deworming decreased absenteeism and increased long-term earnings.

Tactics for resolving the “chief complaints” of augmented intelligence can help get to the right answer: low data literacy across the organization (don’t dumb the data down; provide the full analysis and help interpret it), lack of agreement of definitions (analyze the impact of changes to the definition using AI), low analytic throughput, lack of trust in results, and pilot-itis.

Wave 5- 38 – Diversity and Data for Improved Care (Panel) (Strategy; Course Level-Beginning)

Trudy Sullivan, MBA, Chief Communications Officer and Chief Diversity & Inclusion Officer, Health Catalyst

Maryellen Gleason, MBA, President & CEO, Solve ME/CFS Initiative

Andres Gonzalez, Vice President, Chief Diversity Officer, Froedtert & Medical College of Wisconsin

Eloiza Domingo-Snyder, PhD, Executive Director, Diversity and Inclusion, Astellas Americas

Noel Tenoso, DPT, Founder & CEO, Advance Sports & Spine Therapy; Partner, Solveglobal

Christine Neuhoff, JD, MBA, Vice President and Chief Legal Officer, St. Luke’s Health System

A person’s ethnicity, culture, gender identity, socioeconomic status, and other factors affect how they walk in the world—and their experience and outcomes as a patient.

Bias can begin with frontline staff who assume a patient’s ethnicity, rather than asking them. It’s important to train the people who collect the data and those who crosswalk the data.

For transgender people, there is implicit bias in EHRs. Many (70 percent) never have gender-affirming surgery, and their anatomy may affect the care they need. So, what are you asking when you ask them for their sex?

It’s important to aim for patient-centric healthcare, but who is that patient? American healthcare was a system built to serve one type of patient not a diverse community. There’s gender bias in some diagnoses. ME/CFS is considered to be a woman’s disease and often blamed on psychological factors, although it’s physical. It’s thought that 30 thousand men (unaccounted for) have the disease but are told to “man up” rather than being diagnosed.

When trying to measure health equity, one problem is that demographic data is bifurcated into male and female, white people and people of color. Regression analyses tend to create better, more focused solutions. When you intersect your data, things become more interesting.

The bias in the healthcare workforce can be hard to attack because leadership sees diversity and inclusion as a “nice to have” rather than a necessity. But it is a necessity. It’s the right thing to do. Also, there are plenty of studies out there for the business benefits of diversity. What has been the business case for homogeneity?

Wave 5 – 39 How Real-World Data Paves the Way for Personalized Healthcare
Okan Ekinci, MD, MBA
Chief Medical Officer, Diagnostics Information Solutions (DIS), F. Hoffmann La-Roche Ltd

Healthcare knowledge is growing exponentially, and more data is being captured and stored than ever before. In addition to healthcare data, vast amounts of clinical trial and real-world data is being generated across institutions and geographies that could help the healthcare industry gain actionable insights for decision support tools.

Okan Ekinci, MD, MBA, told breakout attendees that it’s more important than ever to implement a technology framework for healthcare. However, we can only capitalize on the value of healthcare data if we can bring the key players from all areas of healthcare together—including academia, medical associations, regulatory bodies, payors, providers, government, and non-profits. The Roche Group is in a unique position to bring diagnostic and pharma data together with both being generated under roof, providing access to meaningful data at scale and enabling the ability to leverage data towards truly individualized patient care.

He used examples from Roche to show how the life science industry is evolving to address personalized healthcare and offered breakout attendees the following three key takeaways from his presentation:

• Increasing availability of multi-domain data allows a “high-resolution view” on the patient along the care continuum–a foundation for personalized healthcare.
• Clinical decision support helps manage the increasing data-driven complexity, improves operational efficiency, reduces unwarranted variation, and improves outcomes.
• A decision support platform with workflow products and apps, enhanced with AI/machine learning is the foundation to establish learning systems that generate actionable insights through stream analytics and the creation of a comprehensive “digital twin.”

40 – The Future of Digital Health

John D. Halamka, MD – Chief Information Officer, Beth Israel Deaconess System; International Healthcare Innovation Professor, Harvard Medical School

Dr. John Halamka travels more than 400,000 a year to study emerging digital health trends throughout the world. He provided HAS attendees with several of the most interesting highlights from this year’s travels, including the following:

• One dermatology clinic Dr. Halamka visited sees 4,000 patient per day. Why? Because there is no primary care in China.
• Dr. Halamaka said India has a culture of “personal health records.” The reason? No doctor or hospital keeps medical records. Patients are truly stewards of their own data.
• Finland legally requires its citizens to share their healthcare data, with no ability to opt out.
• Scotland has a single data repository of every citizen’s medical records that all clinicians can access.

So, what new workflows are on the horizon to address these missed opportunities? Dr. Halamka suggests five strategies:

• The internet of things and wearable health devices will connect to the enterprise.
• AI and machine learning are mainstream tools available from multiple platform providers.
• The era of apps and cloud hosted services has arrived.
• APIs are increasing in number and sophistication.
• New incentives will shape strategy—value-based purchasing, the promoting interoperability program and telemedicine friendly reforms.

Dr. Halamka closed with encouragement that people across the world do incredible things with limited resources, and we will do the same as we move toward a future of digital healthcare.

42 – Closing Remarks

Paul Horstmeier – Chief Operating Officer, Health Catalyst

This year, attendees created 233 Braindate topics, went on 149 one-on-one Braindates, 92 group Braindates, and made a total of 754 connections. Stuart Gold and Mr. Horstmeier announced the winners of the HAS App game, Braindate, and the drawing winner.

Mr. Horstmeier also announced the dates for next year’s Healthcare Analytics Summit: Save the date for HAS 20 September 1–3, 2020!

Dan Burton – Chief Executive Officer, Health Catalyst

Dan Burton began his closing remarks by reflecting on some of the many highlights of HAS 19, including some personal highlights from the jam-packed past few days. He announced this year’s highest rated keynote and breakout speakers, brought attendees back to the improvement flywheel, and shared another clip from Apollo 13, reminding attendees that there can be a happy ending for all of us. He closed by bringing his space travel analogy full circle and told the crowd, “Houston, I can’t wait to see you again.”

‍