Medical Data Science
Medical Data Science
By: Ray Mierwald
Visual Connections Senior Data Scientist
Data science draws together data analysis, statistics, and advanced analytics like machine learning to extract knowledge from structured and unstructured data. At Visual Connections, I work on projects utilizing health care data, and particularly, data from the Centers for Medicare and Medicaid Services (CMS). My background is health services research, claims data, programming, CMS data, and CMS payment and risk adjustment reconciliation.
Medical Data
Currently Industry Trends. Medicare costs are expected to grow from 15% of Federal spending to 18% in the next 10 years https://www.kff.org/medicare/issue-brief/the-facts-on-medicare-spending-and-financing/. To combat this growth, CMS has looked to data scientists to help control costs through developing cost and quality-based incentives as well as new methods to identify fraud, waste, and abuse. The CMS Center for Program Integrity (CPI) has looked to data science to adapt machine learning and advanced analytics to identify fraud, waste, and abuse. The Center for Medicare and Medicaid Innovation regularly develops new pilot projects to assess new methods of quality-based care. The CMS administrator has also made great efforts to release additional publicly available files to spark innovation in the private sector.
One trend that CMS has embraced is bringing new players into the healthcare space including other research organizations, artificial intelligence companies, and stakeholders who have developed fraud, waste, and abuse techniques in other fields. Visual Connections has been a willing partner with these projects through our ad hoc work with CMS as a performance measurement contractor as well as our PADOS contract that specifically has tasks set aside for Program Integrity pilots. As companies come into the healthcare space, they need people who can help them understand the data so they know what to measure, how to calculate cost savings, and how to use CMS data.
Through the Visual Connections (VC) Lens. Visual Connections is uniquely positioned in the field of data science. First and foremost, we have some of the most powerful tools at our disposal for model building and advanced analytics. The Visual Connections Analytic Services Environment, or VCASE, is a complete analytic services solution powered by the SAS Viya platform. This environment gives all the advantages of programming in SAS plus the additional features offered in the Viya platform, including distributed computing, machine learning procedures, and an ability to connect to open source software such as R and Python. In addition to our computing, we also have a direct data connection with CMS that allows us to quickly and securely transfer data to and from our system.
Medical Data Analysis
VC Medical Data Solutions within CMS. Visual Connections has multiple contracts with CMS including projects that identify fraud, waste, and abuse in the Medicaid and Marketplace programs, develop models and measurement techniques to evaluate the performance of Medicare administrative contractors and implement pilot programs applying artificial intelligence and machine learning techniques to a variety of emerging CMS issues.
As a data scientist, I regularly pull from all areas of my background. For example, a simplified project might require that we estimate CMS’ liability for a certain type of fraud. To complete a project like this, we need to gather data from raw data files in the CMS mainframe, stage the data in a processing environment where we can take advantage of our analytic tools, process the data which may involve identifying other instances of that type of fraud either through direct identification or through a likelihood match, reprocess the data without the instance of fraud, recalculate the payment, and provide our clients with reports and visualizations that explain the issue. Data science is very important to this type of work because it touches each part of the analytic lifecycle and provides a framework for approaching each research question.
My work at Visual Connections gives us the chance to not only develop our own models of healthcare cost and quality, but also to work with CMS researcher partners including the Oak Ridge National Laboratory, SAS Institute, as well as AI and Machine Learning start-ups. In these studies we investigate trends in Medicare Advantage error, proactively identify bad actor prescribers in Medicare Part D, and assist the Medicare Fee for Service program in identifying metrics that best represent high quality work for Medicare Administrative Contractors.
New customers looking to incorporate advanced analytics into their work or looking for a partner with experience using CMS data would find a valuable partner in Visual Connections, our analytic professionals, and the VCASE system. Some of the areas we can help with include:
Identifying the best CMS data source to complete an analysis of health care cost, quality, or effectiveness
Developing Data Models
Supplying Advanced Analytics Tools through use of the VCASE system
In the Medical Data arena the nature of the work is inherently collaborative and creative. At VC our commitment to continued growth and innovation breads a wealth of opportunity for our clients, and our interdisciplinary team environments find innovative ways to complete projects.
Ray Mierwald
Health services researcher with extensive experience working with the Centers for Medicare and Medicaid Services (CMS) on analytic projects related to Medicare Risk Adjustment, Medicare Part D, Medicare Advantage, and Fee For Service (FFS) Medicare.
Specialties: Medicare Part D, Medicare Advantage, Medicare Part C, Fee for Service Medicare, Improper Payments, CMS Medicare Audits, RADV, risk adjustment, policy analysis, health care data analysis using SAS and SQL, z/OS environment programming, ETL processing, survey research, extensive knowledge of health care claims data.