Virtually Kicking Off the 2020 Healthcare Analytics Summit

Frueh introduced the 2020 platform, which brings attendees to the summit’s customary home, Salt Lake City’s Grand America Hotel, via a navigable virtual environment. Visitors can “travel” around the venue, from keynotes in the Grand Ballroom to individual breakout rooms on the virtual grounds, and even participate in a scavenger hunt throughout the environment and take quizzes to earn points for the HAS Game.

HAS 20 Virtual will also provide some of the fun and good humor attendees have enjoyed at the in-person events. From the fun run to the friendly competition for the most notable socks, HAS 20 Virtual has moved these activities online, inviting participants to post photos on social media with the hashtags #HAS20virtualfunrun, #SocksofHAS, and #HAS20, as appropriate.

The summit’s popular Braindate—one-on-one or small group discussions around unique areas of interest—has also transitioned to the virtual format for 2020. Participants can propose topics and sign up for Braindates in the HAS 20 Virtual environment.

After the welcoming remarks, Dan Burton, chief executive officer of Health Catalyst, opened HAS 20 Virtual with this year’s theme, “Analytics in the New Normal.” A video followed that reflected on the role of analytics in COVID-19 response, data as the lifeblood of healthcare response and improvement, and united effort among the healthcare industry against the pandemic. Burton explained that HAS 20 Virtual would offer critical learnings from frontline leaders of the COVID-19 response with strategies to apply their experience to other health systems. Both keynote speakers and breakout presenters would address the value of a data analytics infrastructure in the pandemic response and how aggregated data can drive insights in testing, mortality, and treatment. COVID-19, Burton said, is a shared challenge, and organizations coming together is a reason for hope and optimism.

Keynote Session – How Analytics and AI Can Reboot Healthcare

In the opening keynote, Eric Topol, MD, a renowned AI expert and healthcare futurist, outlined the possibilities and realities for a data-driven COVID-19 response. Innovation, he said, that marries AI in the form of deep-learning algorithms and deep neural networks with sensor technology (e.g., wearable monitors and thermometers) and human insight promises to drive a more nuanced understanding of the virus. For example, the promise of a rapid home test that identifies infection and also shows viral load (a measure of infectiousness) can significantly impact control efforts.

The complexities of COVID-19 make the disease challenging to fight and, thus, build a strong use case for AI-enabled strategies. The virus contains a spike protein that attaches to a receptor widely expressed throughout the body, the ACE2 receptor. Symptoms are, therefore, not just respiratory but can affect any organ in which the ACE2 receptor is present. COVID-19 also has a broad spectrum of impact—from asymptomatic to severe and even lethal disease, as well as instances of “long COVID,” in which patients can experience durable and disabling symptoms.

Dr. Topol explained that AI relies on inputs, and its outputs will only be a good as the incoming data, building on the HAS 20 Virtual focus on analytics as an essential driver of COVID-19 response. With precise and accurate data inputs, deep-learning algorithms can identify a medical condition (e.g., a cancerous lesion) more accurately and in greater detail than human judgment alone. AI-aided imaging innovations may also improve the diagnostic process with greater speed and sharper images. Even smartphones will play a role—as Dr. Topol explained with the example of high-resolution ultrasound images he took using his phone.

Machines will not replace physicians, Dr. Topol said, but physicians using AI will replace those not using it. He explained a vision for the future in which a “virtual medical coach” used inputs from sensors and smartphone applications to understand the patient’s health and share that information with a clinician. Still, Dr. Topol emphasized that human touch and empathy are irreplaceable in healthcare encounters. The ultimate goal of AI-driven innovation in COVID-19 and beyond is to improve not just medical accuracy but also the human side. As machines get better, clinicians can focus more on the human experience.

Keynote Session – Data on the Frontlines of COVID-19

In his keynote address, Ari Robicsek, MD, chief medical analytics officer at Providence St. Joseph Health, shared how his health system quickly mobilized data to synthesize information about COVID-19 throughout the organization. Providence St. Joseph Health, a 51-hospital system, saw the first known case of COVID-19 in the U.S. in mid-January 2020.

Dr. Robicsek provided a timeline of events that showed the progression of the organization’s understanding of COVID-19 as well as the maturity of their analytic tools. Beginning in early March, they had limited testing abilities but were suddenly concerned about the presence of coronavirus in their communities. Dr. Robicsek drafted a rough outline of a dashboard the team could build to help identify coronavirus; the outline showed a percentage of patients presenting known symptoms by zip code. To build this tool, the team needed to extract relevant clinical information from each of the hundreds of thousands of clinical notes they produced each day. They started with simple text parsing, which led to falsely identifying many negations or irrelevant notes. Their natural language processing (NLP) became more sophisticated, using expert annotation to train the machine learning tool on how to detect negations and other nuances to identify relevant clinical data.

The team then used that data to create an epidemiological registry to track data—including patient demographics, relevant diagnoses, and clinical features—which allowed them to follow the patient journey and visualize activity by geographic location. While the team and their analytic tools have made incredible progress, there’s still much to learn about COVID-19. Using data, including how to safely open schools, which patients will benefit most from Remdesivir, how best to care for COVID-19 and non-COVID-19 patients simultaneously, and much more, Dr. Robicsek provided a glimpse into the future of sophisticated NLP and granular mapping tools.

Breakout Sessions: Wave 1

Session 4 – Machine Learning, Social Determinants, and Data Selection for Population Health

Jason Jones, PhD – Chief Data Scientist, Health Catalyst

Terri H. Steinberg, MD, MBA, FACP – Chief Health Information Officer & Vice President, Population Health Informatics, ChristianaCare Health System

Machine learning models are easy to build, relatively speaking. Eighty percent of the effort is getting data ready to find useful, actionable information that determines the success of population health programs. How can we select the right data—and is there more value in 800 candidate features or 8?

Terri Steinberg of ChristianaCare Health System identified three key components for population health management:

1. Technology—provide tools.
2. Analytics—identify populations.
3. Care managers—manage time between episodes.

In 2012 ChristianaCare started pursuing risk contracts with elements that weren’t billable in fee-for-service care but had a large impact on outcomes (e.g., social determinants of health). The organization found many benefits in identifying high-risk patients and high healthcare utilizers, leading to fewer inpatient days and lowered costs of care.

Co-presenter Jason Jones then outlined that risk identification required careful data selection upfront, before population health prediction machines work to accomplish the following:

• Identify the most vulnerable members.
• Segment by need.
• Optimize outreach and impact.

To find the few population members who can benefit the most and present this information to care managers in the most efficient, least overwhelming way, ChristianaCare has provided a machine learning tool for care teams. The tool shows care managers which patients to see, tracks which interventions work for which segments, allowing care managers to manage care between visits while monitoring all patients daily. The tool also populates worklists with individuals needing more care, and care teams evaluate those lists for accuracy.

The resulting combination of machine learning and human judgment supports successful care management.

Session 5 – Opening the Door for Patient Access: Growing Monthly Visit Volume by 27%

Carrie Rys, MBA – Assistant Vice President, Ambulatory Operations, Texas Children’s Hospital

Grace Karon – Assistant Director, Business Operations and Strategic Planning, Texas Children’s Hospital

Access to care is a crucial driver for patient satisfaction and patient site-of-care selection. Before the COVID-19 pandemic, Texas Children’s Hospital (TCH) created a structure and culture to enable sustainable improvements in patient access. Over 200 team members from across the organization, with the support of an executive steering team, accomplished the following over the past three years:

• Increased capacity and improved utilization by establishing standard four-hour scheduling blocks and converting from opt-in to all-in text messages reminders—resulting in a 27 percent monthly visit growth and reduction in the no-show rate from 11.5 percent to 7.7 percent.
• Enhanced patient referral optimization by standardizing 60 different referral processes to just one, thereby reducing the time to complete forms from 2 minutes to 30 seconds and improving the percentage of specialties able to offer new patient appointments within 14 days from 66 percent to 83 percent.
• Focused on patient portal activation by educating and incentivizing staff, creating unique marketing deliverables, and enhancing patient-focused portal features (e.g., pre-check-in)—increasing portal usage and improving the percentage of patients who rated their scheduling experience as “very good” from 57 percent to 78 percent.

With the above improvement groundwork, TCH quickly adapted critical healthcare services during the COVID-19 pandemic, including rapidly deploying telemedicine, creating a touch-free patient experience, and innovating care delivery with programs such as a drive-through “Patient Express” clinic. These efforts have helped sustain pre-COVID-19 visit volumes and create models of care that will continue to succeed in the new normal.

Session 6 – Data Quality: The Bane of Every Improvement Initiative

Dan Heinmiller – Vice President, Operations for the Application Suite Business Unit, Health Catalyst

Health systems’ COVID-19 response and recovery efforts provided urgent analytic use cases, each one highlighting data quality as a prerequisite. While data quality is essential, COVID-19 use cases brought the issue to the forefront. What was so different about COVID-19? The urgency with which the outbreak called for data, constrained resources at health systems, the changing scope and variety of data, and new report requests and value sets introduced are just some of the differences that highlighted the importance of data quality.

One example Heinmiller used to illustrate COVID-19’s unique aspects was the CDC’s isolation guidelines for persons with COVID-19 not in a healthcare setting. These guidelines changed four times in three-and-a-half months. For a care manager, who needs to convey the most up-to-date information or needs analytics to help identify how long patients were in home isolation, rapidly changing guidelines make it difficult to be agile and responsive to urgent requests.

Heinmiller shared five essential data quality lessons that COVID-19 highlighted and discussed how healthcare organizations can apply these lessons to prepare for the next analytics use case. One key takeaway was that just-in-time data quality, or just-good-enough data quality, is no longer sufficient for data quality requirements. Health systems need to be ready with trusted data they can quickly leverage for an analytic response.

Session 7 – Improving Revenue Cycle Performance with a Patient-Centered Care Access Center

Jennifer Livermore, MSIHM – Director of Access Operations, ProHealth

As COVID-19 continues to present new challenges, health systems are seeing the pandemic’s impact on access as it relates to revenue. ProHealth started with an opportunity assessment to identifying three key areas in which to improve access and revenue cycle performance:

1. Access advancement: With data-driven insight about access issues and capacity-constrained service areas, ProHealth implemented a consistent, centralized access approach that improved patient satisfaction and reduced abandonment rates.
2. Resource optimization: ProHealth’s analytics revealed an opportunity to implement proactive workflows, streamline cancellations, and automate appointment reminders, leading to a decrease in no-show appointment rates.
3. Stakeholder engagement: Inconsistent referral data highlighted the importance of systemwide support, resulting in comprehensive dashboards to collect referral data, streamline referral processes, and a decrease in referral leakage.

Better patient access led to increased revenue through standardized processes, improved patient experience, less referral leakage, and better measurement reporting and accountability. As a result, even during the COVID-19 pandemic, ProHealth leveraged technology with communication and care delivery to reach patients and generate revenue.

Session 8 – The Era of Disruption and Uncertainty Ignites New Data and Analytic Imperatives

Laura Craft – Vice President and Analyst, Gartner

Laura Craft explored the current turbulence in healthcare and described how, despite collective discomfort, analytics leadership has an opportunity to advance positive change. Among Craft’s key points were the following:

• Three influences—digital transformation, AI and automation, and COVID-19—have formed a “perfect storm” of opportunity. Each relies foundationally on data and analytics, underscoring the need to modernize capabilities across analytics skills, tools, and governance.
• Digital transformation (e.g., telemedicine), which a vision for an integrated care ecosystem drives, has accelerated during COVID-19. Leaders have an opportunity to align their data strategy with their digital strategy to ensure the data fluidity, agility, and liquidity that digital healthcare demands.
• Providers are adopting AI and other automation capabilities to the point where these capabilities touch every area of the healthcare enterprise. With continued investment in AI, leaders can continue to realize efficiencies that improve clinical and operational outcomes.
• COVID-19 has highlighted the indispensability of data and analytics to help guide responses—but it has also revealed gaps. Leaders need to ask themselves: are my data and analytics positioned to be accessible, accurate, and reliable in a time of crisis? Where the answer is no, leaders must bolster this critical infrastructure.

Craft named data and analytics as the primary offensive tools in uncertain times—and invited participants to use the COVID-19 era to re-ignite their data and analytics mission.

Session 9 – Addressing Health Equity in the New Normal

Vivian Anugwom, MS, CHES – Health Equity Program Manager, Allina Health

Vivian Anugwom, who manages health equity at Allina Health, defined equity within healthcare as healthcare delivery tailored to each patient’s unique needs. Health disparities have always existed, but COVID-19 has exacerbated these gaps, leaving communities of color and lower-income with even worse access to care than before the pandemic. Anugwom explained the framework Allina Health followed to identify and address health disparities amid the pandemic:

1. Leverage data to identify disparities.
2. Understand the root causes and identify solutions within the sphere of influence.
3. Implement resource tactics to close the gap measurably.

Following this data-driven framework, Allina Health’s analytics discovered that providers referred African Americans patients to the hospice program at lower rates compared to other groups. Based on these findings, Allina Health engaged leadership and offered implicit bias training to providers to address the hospice disparities. The implicit bias training highlighted African Americans’ perceptions of hospice care, explored the effects of biases on hospice communication and referrals, and provided resources for effective communication around hospice services. The training helped providers better understand their African American patients, individuate these patients, and recognize their individual biases and situations that might magnify these biases. Following this data-backed framework, Allina Health increased hospice referral rates for its African American populations.

Session 10 – Effectively Restarting Elective Surgery After COVID-19

Nirav Patel, MD – Medical Director for Surgical and Procedural Services, Banner Health

Martina Brooks, MHI, CSSBB – Surgical and Procedural Standardization Program Director, Banner Health

Working in five phases, Banner Health built on lessons learned from pre-pandemic surgical planning while growing their abilities to think ahead as an organization.

Martina Brooks looks back to a 2017 review that found inconsistencies between charge capture and utilization data in the top 20 procedures performed at Banner. The organization knew they needed to focus on understanding why.

By 2018 they established a surgical procedural and value alignment program with team members from supply chain, perioperative services, and care management, who developed a dashboard for seeing variations across the top 20 procedures. In just 15 months they achieved a $3.2 million reduction in surgical supply costs.

As Dr. Nirav Patel moved ahead in the Banner story to talk about COVID-19 impacts, he shows how the earlier work of building transparency into surgical planning helped with rapid cycle deployment and surge planning for 2020 and beyond.

Banner went from having no playbook for addressing unexpected losses and surges to having a playbook with some key elements:

• Provide transparency in real-time, in one source, so everyone in the organization sees information presented the same way each day.
• Communicate information succinctly and welcome feedback from all individuals in the organization—they may have insights no one considered.
• Build predictive models and then integrate the models across areas of utilization
• Provide transparency to reassure patients that you’re planning in a methodical way to ensure safety.

Patel and Brooks provided details on how they optimized data to achieve a $27.9 million revenue recapture.

Session 11 – Beginning a Predictive Analytics Journey: Choosing the Right Use Cases for Success 

Jaclyn Bernard – Innovations Lead, Application Development and Predictive Analytics, Texas Children’s Hospital

This two-part session tackled the early challenges healthcare organizations typically face as they aim to leverage predictive analytics to drive improvement: securing resources (expertise, technology) for the program and choosing use cases that best position projects for success.

To secure approval and funds for an initial investment in predictive analytics, Bernard’s team demonstrated the program’s value via multiple proof of concept (POC) projects and strategically selected use cases. POC models proved broad applicability across clinical and operational areas, and the use cases ensured that projects were aligned with organizational priorities. With executive support obtained, TCH then operationalized the most promising of their models and trained employees to use predictive analytics.

Operational partnerships emerged as a key success factor in developing successful projects. This and other early lessons informed subsequent work establishing a framework for evaluating use cases. The framework helps Bernard’s team reliably and efficiently investigate, select, prioritize, and execute projects that deliver significant value for TCH.

Session 12 – Population Health, More Important Now Than Ever? An Equity Approach to Supporting Communities

Amy Flaster, MD, MBA – Senior Vice President of Population Health Management, Health Catalyst

As one of the population health leaders at Mass General Brigham (MGB) in Boston, Massachusetts, Dr. Flaster and her team at MGB created a plan to ensure health equity for residents living in underserved communities during COVID-19. An equity-focused approach was critical because national and local data show that disasters disproportionately impact vulnerable and minority populations. Data revealed that cities in Massachusetts with the highest rates of COVID-19 were primarily communities of color and translated to worse outcomes related to the virus.

With the rapid onset of COVID-19, MGB had to act quickly. To reach these underserved, disproportionately affected communities, MGB focused on addressing the social determinant needs of these patients by scaling existing human and technological infrastructure to scale a COVID-19 equity solution.

The community-based equity COVID-19 strategy used four pillars:

1. Identification: Focused on expanding testing capacity and increasing testing.
2. Mitigation: Aimed at limiting the spread of COVID-19, particularly in high-density areas where people couldn’t socially distance.
3. Isolation: Creating safe sites for COVID-19-positive patients who could not access a safe place to isolate.
4. Phone communication: Outreach to patients via telephone or text message to assess overall health state and needs for any additional services.

With a data- and equity-focused approach, MGB increased health access and support services to vulnerable communities during the pandemic.

Session 13 – The Ethics of AI in Health: Creating Value that Benefits Everyone

Tom Lawry – National Director for AI, Health & Life Sciences, Microsoft

AI in health will be a force for good for all or for some; we get to decide. This was one of the powerful takeaways from Tom Lawry’s breakout session on the ethics of AI in healthcare. As AI solutions become more prevalent in health systems, a host of ethical issues are coming to the forefront. How we address them will determine whether AI will be a force for good for some–or all–healthcare consumers.

Mr. Lawry shared several proof-of-concept vignettes illustrating the decisions that healthcare leaders must make. For instance, one AI goal to reduce adverse events outside the ICU using machine learning resulted in a 60 percent reduction in those adverse events. However, drilling into the details revealed that there was a 95 percent reduction of adverse events outside the ICU for white males and only a 25 percent reduction in Hispanic females. The health system improved quality and efficiency, but is this discrepancy acceptable when considering the goal of equal application of machine learning and AI? He also examined several COVID-19 examples, including exploring the question of why African Americans are twice as likely to die of COVID-19.

In addition to these illustrative vignettes, Mr. Lawry shared corresponding tenets of the ethics of AI, such as, “AI-driven systems may create unfairness or harm if healthcare professionals do not understand the limitations and accuracy of a system.” In this session, Mr. Lawry defined key ethical issues for using AI in health–and during a pandemic–and provided recommendations for creating a framework to guide its usage.

Session 14 – Success in the New Normal: Integrated Data to the Masses

Sy Johnson, MBA – Chief of Staff, Renown Health

A few years ago, only a few people at Renown Health had data, believed in it, and could use it, and now hundreds of people—and soon thousands—can leverage what they have for the communities they serve.

With a mission to “Exceed people’s expectations for access and affordability,” the Renown not-for-profit health network has engaged in population health for 10 years, with a recent strategy for:

• Infrastructure optimization.
• Network development and integration.
• Reduced clinical variation.
• Innovation investments.

Sy Johnson discussed changing organizational culture from having data available–but only if you do a lot of work and spend a lot of money getting it–to making it available for everyone to use in real time for frontline care and to keep the community healthy before they need care.

Johnson has held firm in focusing on out-of-the-box solutions that get to most people’s problems most of the time, creating a menu of options for problem-solving. This menu creates a common language for data requests, avoiding the technical lift of customization until later in their journey of data democratization.

Having a solid infrastructure in place, Renown deployed COVID-19 dashboards in several weeks, building on data from a community already engaged in their health partnerships. They use predictive analytics and risk stratification to identify the most vulnerable for outreach and help them stay healthy. Having data deeply embedded in their strategy also shows in their small patient populations, where inpatient encounters for COPD dropped from 822 to 629 from FY2017 to FY19.

Day in Review and Closing Remarks

Paul Horstmeier – Chief Operating Officer, Health Catalyst
Paul Horstmeier closed the first day of HAS 20 Virtual by thanking attendees for joining and recapping an exciting day of two keynote speakers and two waves of breakout presentations, all in the virtual Grand America. He then introduced the agenda for day two, including keynote speakers, the Analytics Walkabout and Machine Learning Marketplace, Digital Innovation Showcase, and Braindate opportunities. With the HAS competition well underway and attendees actively earning points, Horstmeier also invited participants to stay on the platform after the day’s programming to earn points in the scavenger hunt. He closed his remarks by reminding attendees of the 9 am start time (doors open at 8:45 am) for day two of HAS 20 Virtual.

Additional Reading

Would you like to learn more about this topic? Here are some articles we suggest:

1. HAS 20 Virtual: Featuring a World-Class Lineup of Keynote Speakers
2. HAS 20 Virtual: Reimagining the Healthcare Conference
3. The Healthcare Analytics Summit™: 2019’s Top Data Discoveries and Insights
4. The 2019 Healthcare Analytics Summit: Thursday Recap
5. The 2019 Healthcare Analytics Summit: Wednesday Recap