Lessons from Yubin Park, Chief Data and Analytics Officer of ApolloMed, empowering value-based care providers with daily, actionable insights

Yubin Park, Chief Data and Analytics Officer of ApolloMed, a risk-bearing, technology-enabled value-based-care provider

and

Nov 27, 2023

Subscribe to our substack for updates and listen on Apple Podcasts or Spotify. Connect with Andrew or Lipsa if you find this post insightful and want to learn more.

Welcome back to the Pear Healthcare Playbook! Every week, we’ll be getting to know trailblazing healthcare leaders and dive into building a digital health business from 0 to 1.

Today, we're excited to get to know Yubin Park, PhD. Yubin is the Chief Data and Analytics Officer at ApolloMed, a leading physician-centric, technology-powered, risk-bearing healthcare company. Yubin leads Apollo’s data team in building data analytics tools to improve the efficacy of value-based care delivery.

In this episode, we cover Yubin’s serial founder journey, how he pivoted into value-based care, approach to leadership, how ApolloMed differentiates itself in providing value-based care, and advice for technical founders.

If you prefer listening, here’s the link to the podcast!

Yubin’s serial entrepreneurial journey

At the time Yubin was finishing his PhD in Machine Learning in 2014, everyone who majored in ML wanted to work for Google, Facebook and Amazon. However, Yubin was noticing that healthcare costs were rising at an alarming rate. Recent news coverage in the NYTimes shows that consumers were paying astronomical amounts by just going to the ER.

“I grew up in Korea, and I had no knowledge about US healthcare – I’m still learning, to be honest. It’s not like I had this idea about building a data-driven platform for value-based care. I tried a bunch of different things and learned.”

For his first company, Accordion Health, Yubin initially worked on a price transparency product to tackle exorbitant costs in the US healthcare system. The goal was to create a market-friendly healthcare ecosystem by providing consumers with comprehensive price information, believing that transparency could enhance consumer decision-making.
He funded his project with a Small Business Innovation Research (SBIR) Grant from the National Science Foundation. These can be great resources for companies to bootstrap their first ideas.
Yubin struggled with that product because, at first, he didn’t grasp the dynamics of the market and the underlying principal agent problem (where the purchasers of healthcare are not the users of healthcare). He didn’t realize that people might not care much about the actual price, since many costs were covered by insurance.

At first, Yubin didn’t grasp the dynamics of the market and the underlying principal agent problem (where the purchasers of healthcare are not the users).

After some iteration, he realized there was a huge gap between the negotiated price and the actual cost of care, so Yubin's focus shifted to value-based care like bundled payments.

Bundled payments are a risk-sharing arrangement between a provider and insurer in which a provider is paid a fee (adjusted for the health risk of the patient) for an episode of care (like a knee replacement) including the initial event and subsequent follow-up care. If the patient incurs additional complications, the provider covers the cost of those complications. The program is designed to improve quality of care while reducing unnecessary costs.

Bundled payments are only one type of alternative payment mechanism and they were in vogue in 2014 as CMS had just released their Bundled Payment for Care Improvement program.

Accordion eventually became a data platform for value-based care which essentially means it helped providers understand how to be successful in that type of program with dashboards and reports.
Almost immediately, the complexity of healthcare pricing became evident. Quantifying prices for medical procedures, like knee replacements, proved challenging (how would one decide which services were and were not included?). Further, providers who were being held accountable often offered constructive criticism for the methodology.
Despite encountering criticism from doctors and recognizing the complexity of healthcare pricing, Yubin remained motivated to speak to customers and be open to feedback. Meetings with doctors and critiques of initial solutions provided valuable insights, fueling his determination to delve deeper into the intricacies of healthcare.

Following Yubin’s exits, he expanded his horizons and took the opportunity to scale his company to have greater impact (‘cut the bullshit and do what’s meaningful’)

In 2017, Yubin sold his first company, Accordion, to broaden his exposure to the healthcare industry at Evolent Health. Accordion had secured a client in the risk adjustment space so he could have become a deep expert in risk adjustment, but that would have limited his exposure to the broader complexities of the industry, like running a complex Accountable Care Organization(ACO) program.

Some short years after Yubin left Evolent in 2018, he started his second startup, Orma Health, in 2020. Orma Health, another value-based care company, had two main products:

A value-based care data platform that provided clinicians with real-time population health data to help them quantify and manage their financial risk in a value-based care program.
A remote-patient monitoring program to help clinicians track their patients after they left the clinic or the hospital, with a comprehensive, real-time view of patient history.

Eventually, Orma was acquired by ApolloMed and Orma’s acquisition, on the other hand, was about scale. At the time, Orma was working with one direct contracting entity and a fairly sizable remote patient monitoring program, but he needed resources to scale the company. ApolloMed seemed like a natural fit and he was drawn to the young, energetic leadership

(Source: Orma Health’s approach to real-time clinical AI)

Tips for building data products and platforms in healthcare

Yubin recommends leveraging the modern data stack heavily when building an MVP to help you build quickly and experiment with different ideas.

Since Yubin started his companies, the data tooling landscape changed quite a bit with the advent of cloud computing (Azure, GCP, Amazon), cloud data warehouse (Snowflake, Motherduck, Redshift), data engineering (dbt), data science (Databricks), and business intelligence (Tableau, Metabase, Looker).
Healthcare is also becoming more open to these modern tools, modularized solutions and APIs. More healthcare organizations are preferring to use fully-managed cloud-based tools over on-premise, self-hosted solutions. A good example of this is Snowflake, a modern data warehouse designed to dramatically improve the speed of analytics.
Since these technologies improve the speed and quality with which you can deliver a reliable data product, when building a data platform MVP, you should consider stringing these tools together to build concepts quickly in a cost-efficient way.

Demonstrating ROI in Value Based Care

Yubin initially focused on long-term outcomes, not realizing many customers are more interested in immediate savings with a financial time horizon of one year.

Yubin initially focused on long-term outcomes, not realizing that many customers are interested in more immediate cost-savings value with a financial time horizon of a year. Clinical quality metrics are usually mixed with some being focused on the long-term health of the patient (e.g., HbA1c, 5-year mortality) and others focused on improving short-term health (e.g., ER visits and hospital admissions).

Consider adding in real-time information that can reduce your time horizon for short-term ROI. For example, to measure social determinants of health, we can start to pull in real-time data sources like weather forecast data feed, e.g., AccuWeather, as extreme climate events tend to damage underserved communities more than others.
Alternatively, you could add in real-time information about Google reviews and Social Media feeds that can identify local events and accidents, such as water main fixes and power outages, to see immediate savings in operations. When building your MVP, consider focusing on these real-time, short-term outcomes oriented gains and then expand that strategy to bring in the long-term outcomes.

ApolloMed’s Value Based Care differentiator

Switching gears — at ApolloMed, Yubin runs the entire data and analytics team of 50 people. ApolloMed is a consolidated set of independent physician associations (IPAs which are like provider groups) of over 10k providers who care for more than 1.3 million members through next-gen analytics.

ApolloMed also participates in value-based care arrangements and takes risk on their member’s lives through designation as an Accountable Care Organization (ACO) (both commercially and in ACO Reach) and a Commercial Exclusive Provider Organization (EPO).

Finally, ApolloMed supports other healthcare companies and IPAs as a Management Services Organization (MSO). Other provider groups can contract with ApolloMed to obtain access to their high-quality administrative tooling and increase their practice efficiency.

When asked about how ApolloMed differentiates itself, Yubin notes that value-based care companies sometimes focus too much on data science, ML and AI at the expense of improving the underlying data foundation.

“If you don’t put effort into your foundations, your higher level data science layers will fall like a house of cards.”

In order to derive excellent insights through predictive and prescriptive analytics, value-based care companies will need very clean administrative data (often called claims) and electronic data interchange transactional information.
However, many of the data systems that produce data are very outdated which results in messier data. This makes it difficult to drive innovation in those higher level analytics. If you don’t put effort into your foundations, your higher level data science layers will fall like a house of cards.

ApolloMed’s engine of clinical efficiency

Certain types of clinical metrics do not require AI or other state of the art technologies, but just require clear, high quality rules-based algorithms.
The HEDIS clinical measures, developed by the NCQA, are often used to measure value in value-based care but are fairly complex to implement. “More than 227 million people are enrolled in plans that report HEDIS results.”1
The data we collect is not always a great proxy for what care has been provided, but HEDIS measures can help bridge that gap. For instance, when evaluating the success of your diabetes program, HEDIS can provide guidance for what to measure (e.g., HbA1c), cohort selection, and choosing leading or lagging metrics.

“Imagine trying to reduce your body weight and only being allowed to use a scale once a quarter.”

Imagine trying to reduce your body weight and only being allowed to use a scale once a quarter. Most other industry players calculate these metrics on a monthly or quarterly basis. ApolloMed heavily invested in a NCQA-certified python and databricks engine to automate calculating daily HEDIS measures. As they increased the efficiency of the engine, they saw their physicians logging in to check their results every day.
ApolloMed is also ready and excited to start using LLMs and is currently focused on evaluating which point solutions work with experiments so that they can decide which applications provide the most ROI based on the investment.

Source: A publicly available HEDIS dashboard shows how different acute-care hospitals (ACHs) score on a single HEDIS quality measure. The wide variation in rate of anti-depression medication management and the difference from target show opportunities for improvement. These are the types of problems ApolloMed is trying to solve. ApolloMed wants to make it easier to measure and take action to improve HEDIS scores.

Technical founders should avoid trying to solve every problem

Yubin’s advice for technical founders is to be wary of trying to solve every single problem in healthcare. The healthcare system is too complex to be an expert in the whole thing. Instead, find a specific, concrete problem and focus on that instead of trying to boil the whole ocean.
This is especially useful advice for technical founders coming in from other industries (like big-tech and finance). Technical founders will easily be problem-overloaded with all of the problems in healthcare and might be quick to think of shiny solutions for all of them.

Leadership Lessons

Yubin’s approach to leadership has evolved over time, growing in patience and sympathy. His PhD advisor had the largest influence on his leadership style.
The most important value Yubin holds is a growth-mindset which manifests in the way he leads his teams. He encourages his team to grow their knowledge by encouraging them to be independent as quickly as possible.
To do this, he gives his team a large degree of flexibility and tries to delegate as much as possible. He knows he may not always be successful, but it’s important for him that they feel motivated by growth.

Yubin has a growth-mindset and encourages a similar attitude amongst his team.

Thanks for reading Pear Healthcare Playbook! Subscribe for free to receive new posts and support our work. If you found this post insightful or want to learn more, please feel free to connect with Andrew or Lipsa.

Interested in Apollo Med? Learn more on their website and LinkedIn.

A note from our sponsor: PacWest

Looking for guidance, connections, resources, opportunity? Pacific Western Bank’s banking products and services are built to support your evolving needs as you navigate the challenges of growing a successful business. As you continue to scale, our team will be with you every step of the way. Ready to take your business to the next level? Learn more: pacwest.com

https://www.ncqa.org/hedis/measures/