Skip links

On-Demand Webinar: To Learn How Payers Can Automate Prior Auth at Scale with AI. View Now.

Technology blog

Engineering a Solution to Healthcare’s Data Problem: My Summer Internship at Apixio

The Search

Last summer, I interned in the technology department of a large financial institution in Manhattan. Though I found the work that I contributed to the Daily Market Risk Team to be rewarding, working 11-hour days in a non-stop, bustling city was exhausting.

My hard work was rewarded, and I was extended an offer to return full-time after the internship ended, but instead decided to pursue a Master’s degree in Data Analytics & Statistics at the University of Washington – St. Louis in the coming fall. It was an opportunity to continue my education, and also offered me something that not many of my peers could afford: another summer to intern. After spending fours years of college in the Midwest and a summer working in New York, I knew I wanted to head to the Bay Area to get a taste of “West Coast culture” and figure out where I’d prefer to live after completing my graduate degree.

When searching for this final internship, there were two things I was looking for: (1) It is a paid position; and (2) I find the company’s product to be exciting. While the first reason is self-explanatory (grad school isn’t free), the second demands a little more explanation. Different things are exciting to different people, and I personally love to think about optimizations and how to improve anything and everything. When I was in middle school, I saved time by optimizing the procedure of pouring a bowl of cereal (there are a lot more ways to get a bowl, spoon, milk, and cereal than you’d think). In high school, I optimized my driving speed to balance gas consumption, time spent on the road, and probability of getting a ticket, all while taking into account traffic patterns. In college, I dreamed of ways to eliminate traffic and other inefficiencies throughout transportation and society at large. I craved an internship that would allow me to bring this natural desire to think critically and improve efficiencies into my professional work.

Amidst my search, I was connected to Jeff Gill, Apixio’s Vice President of Engineering. Jeff took the time to explain Apixio’s Computing Platform to me and detail the intricacies of the applications it powers. I was instantly intrigued. With Apixio, I could apply what I’d learned in the classroom to health care’s data problem –– a real-world challenge, impacting hundreds of millions of Americans, that doesn’t currently have many available solutions.  I accepted my internship offer, and proceeded to trade in my suit and lunch box for jeans and catered lunches.

The Culture

On my first day, I was given a tour of the office— three different times! My Apixio colleagues saw that I was new and were eager to welcome me to the team. I accepted each offer and was able to meet and interact with different colleagues. I could instantly tell that this would be an accepting, collaborative workspace, and I was excited to develop my relationships with my coworkers throughout the months of my internship.

When it was time for my first lunch, I was happy to see that because we got catered meals in a shared lunch space, everyone ate together. Lunch wasn’t an event that you penciled into your calendar to network like I was used to; it was a chance to just relax with the team, where almost everyone takes the time from their busy days to catch up with one another and talk everything under the sun from politics to investments to basketball. The Golden State Warriors were in the midst of their championship campaign and my coworkers were getting very excited about their prospects of a rematch with the Cavaliers. The conference finals hadn’t even started, yet they were already acting like the Cavs already swept my Boston Celtics. By the end of lunch, I was at least able to convince them that the Celtics would lose in a respectable five games.

The Work

After orientation, I met with my boss to kick off my first project. As a member of the Applications team, I was tasked with managing our internal modelling of International Classification of Diseases (ICD) Codes, Hierarchical Condition Categories (HCC) Codes, and their relationship. For those unfamiliar with these codes, ICDs are used by physicians and health care providers to classify all patient diagnoses and each ICD corresponds to specific HCCs. The Centers for Medicare and Medicaid Services (CMS) uses HCCs for risk adjustment ––wherein Apixio’s clients (health plans and provider groups) submit HCC codes from their member population, to generate a risk score. This score is used to determine reimbursements to the plans and providers for covering these people. Think of “risk” as a snapshot of how sick the population is, and CMS needs to look at this snapshot to figure out how sick a covered population is, and therefore how much it costs to care for them.

Every year, the World Health Organization (WHO) releases a new list of ICD codes with additions, deletions, and revisions from the previous year. In response, CMS will release a new set of mappings from ICD codes to HCC codes. It’s important that our internal model is using the correct mapping version and codes at all times so we can provide the correct information to our application users (both Apixio’s own internal coders and external client’s coding teams).  

This may sound straightforward, but upon closer examination it gets a little tricky. WHO releases their ICD codes on October 1st. CMS won’t usually ratify their new code mappings until April of the following year; however, we need to use those code mappings for all patient visits dating back to October 1st. Apixio has projects that span a calendar year, meaning that some of the findings from patient data correspond to a different set of codes and mappings than other findings in the same patient data. It is crucial that Apixio stays on top of these mappings or risk extended project timelines, which isn’t good.

Maintaining the codes and mappings has been a manual process so far at Apixio. My specific task was to make it easier for the company to maintain and use this information as correctly and efficiently as possible. Working with Python and SQL, I have created a database for these values and automated the process of uploading new information to the database, as well as extracting it for use in HCC Profiler. It will serve as a single source of truth for all projects going forward. Additionally, I’ve been able to work with AngularJS and Scala on the front end and back end of HCC Profiler to ensure we utilize these mappings properly. This was a great chance to step through the code and familiarize myself with HCC Profiler and our business logic.

I am now over halfway through my internship and have definitely enjoyed my time here so far. I’ve been able to work on a project that eliminates technical debt for the company and directly impacts our users. Coders will do a better job identifying each patient’s conditions. Using this information, insurance companies can work to provide better personalized care for their members. While this was only a small project, my work feels significant and I am extremely proud to be a part of the Apixio team this summer.

Noah Lieberman

Technology Intern


Interested in being part of a great engineering and science team using data and AI to change healthcare? Check out our Careers Page.

🍪 This website uses cookies to improve your web experience.