Q&A with the Predictive Analytics Experts from the 2016 SOA Health Meeting

During this week’s 2016 SOA Health Meeting, the Society of Actuaries will host a panel discussion entitled “Health Care’s Sea Change: How Predictive Analytics is Improving the Way we Deliver Quality and Value-Based Health Care.” Virgil Dickson, Washington, D.C. bureau chief for Modern Healthcare will moderate the panel with Syed Mehmud, principal and senior consulting actuary at Wakely Consulting Group; Dr. Nicholas Marko, chief data officer at Geisinger Health System; and Brian Doty, principal at Deloitte Consulting LLP as the panelists.

The three panelists recently sat down with the SOA to share their insights on predictive analytics in healthcare and how actuaries can prepare themselves for a non-traditional career in the field. Here’s what they said:

Syed Mehmud, principal and senior consulting actuary at Wakely Consulting Group

Q: Regarding predictive analytics, where is the most opportunity in the healthcare industry?

A: The short answer is “virtually everywhere.” More and more of the decisions in the healthcare industry are being driven by an analysis of data. The opportunities begin before the patient is sick, where large amounts of data are being analyzed in innovative ways to improve interventions or develop new ones. The opportunity continues at the point of care, where vast amounts of data are analyzed to improve the patient experience and even the care delivery (e.g., better diagnostics, evidence-based care). Predictive data is utilized to aid the performance of facilities and providers, and then of course, it is used to better the performance of payers as well, which is more familiar ground for actuaries.

Q: As a trained actuary, how did you choose to follow a non-traditional career path, and what advice can you offer other actuaries who may be considering a non-traditional career?

A: In my case, I gravitated to where I thought the opportunity was to add the greatest value to my chosen profession and to those that use my work. It may be surprising now, but in the early 2000s, predictive analytics was not a focus area for actuaries. I thought the writing was on the wall, though. The Data Revolution had well and truly begun, and it was only a matter of time before the ranks swelled. There was, and continues to be, a great opportunity to improve the practice of our profession through the application of creative mathematics and programming. My advice to other actuaries considering a non-traditional career would be to reflect upon where the juncture is for them in terms of their passion, skill and societal need…and then go for it.


Dr. Nick Marko, chief data officer at Geisinger Health System

Q. What kind of skills do actuaries need to have to do predictive analytics?

A:  I think of predictive analytics as part of a larger, organizational effort.  Doing the actual predictive analysis part generally requires solid programming skills and a mathematics background coupled with at least a functional understanding of the problem the algorithms are attempting to solve.  But, more importantly, the actual computational work is only one link in a chain that must be entirely sold, or the effort will fail.  So, while the mechanics of predictive analytics can be thought of as exercises in programming and applied mathematics, an organizational effort to use predictive analytics is more about building a team, coordinating the many moving parts of the solution and keeping the focus on delivering results with organizational value.

Q: How can large enterprises, like large hospital networks, use predictive analytics to improve operations and add value?

A:  Predictive analytics in large enterprises is a multi-step process that starts with treating predictive analytics—one type of advanced analytics—differently than business intelligence and reporting activities and constructing questions carefully before attempting predictive analytics to find an answer. This means that the questions must be answerable with the available data and expertise. There must be a plan to use that answer, and the result must somehow be tied to value.

After a question is constructed, a great deal of effort goes into data management and preparation before the actual predictive analytics work begins.  At this stage, it is critical to have someone who understands the data (a domain expert) and someone who knows how to organize and structure data (a data engineer) to create an appropriate data set.

With good data, the actual predictive analytics work is best done by a data scientist who understands the question, data and downstream mathematics associated with the various predictive strategies to answer the question.  The data scientist may partner with a programmer to write clean code because few people are high-level experts at both mathematics and programming.  For this reason, I tend to think of data science and predictive analytics as a focused team effort.

Finally, once the computational work is done, there must be someone on the implementation side who is ready to put the answer into some operational implementation.  Each step in the process needs focused attention and careful thought if an exercise that includes predictive analytics is to be successful.


Brian Doty, principal at Deloitte Consulting LLP

Q: How is the industry changing to address emerging healthcare needs, and how can actuaries help?

A: The healthcare industry is changing for many reasons—both voluntary and involuntary— to focus on the cost and outcomes of care delivery versus simply the volume of services provided. This is causing healthcare organizations (both providers and health plans) to reexamine their core business model.

The unique capabilities that actuaries bring will enable healthcare organizations to do many things. For health plans, actuaries can help redesign more efficient state Medicaid programs, create payment models to align incentives within care systems and between plans and care systems, design new consumer-oriented products to engage consumers more in healthcare decision making, and develop high value networks considering physician and hospital performance.

For healthcare providers, actuaries can identify and quantify cost reduction opportunities through benchmarking, analysis of appropriate site of service and identification of unreasonable variation in episodes of care. They can develop efficiency measures to educate physicians about better ways to practice care, support care systems strategies to make a smooth transition from volume to value-based, analyze population health to direct attention to gaps in care and prioritize patients.

Q: What does the future look like for predictive analytics in healthcare?

The applications for predictive analytics are increasing to better manage the quality of care and the financial performance of the health system.  As providers assume more risk (MACRA as a driver for this), the need to quantify risks of poor health outcomes in smaller and more finite subpopulations will become even more important. That means we need the ability to predict the likelihood certain patients within a population are at higher risk of a given condition or complication requiring intervention or management using indicators, like clinical, demographic, phenotypic, socioeconomic data.  Health systems want to identify individuals at risk for adverse outcomes to develop more targeted preventative management strategies and to simulate or model outcomes based on specific treatments and protocols to develop individualized treatment plans.

Predictive analytics and big data will converge as new types of patient data are assessed as substrates to power new predictive analyses, and a “knowledge economy” will arise from more prevalent use of predictive analytics. Providers will start to create analytic content that will be valuable to other providers and other sectors.   The true promise of predictive analytics lies in the core promise of identifying patterns or insights that the human brain wouldn’t even notice, which means there are potential applications that we literally can’t imagine yet.

Sharing is caring.
  • Subscribe to our feed
  • Tweet about this post
  • Share this post on Facebook
  • Share this post on Google
  • Share this post on LinkedIn


No responses to "Q&A with the Predictive Analytics Experts from the 2016 SOA Health Meeting"

There are no comments yet, add one below.

Leave a Comment