Quick Q&A: Life at QPID Health, an EviCore Company

Prior authorization for medical benefits is a proven strategy to ensure evidence-based care, but historically has imposed high administrative costs on healthcare systems. However, EviCore healthcare and QPID Health joined forces in 2016, creating a total game changer in the healthcare industry. We are taking a process that used to take hours or even days—the prior authorization process—and through the power of clinical reasoning, natural language processing (NLP), and machine learning, reducing it to seconds or minutes in most cases.

So what is QPID Health exactly and what is it like to work there? We interviewed Virginia Partridge, Natural Language Processing Engineer at QPID to find out.

From your perspective, what does QPID Health do as a company?

QPID’s overall purpose is to reduce waste and improve outcomes in healthcare. Our team tackles that in many different ways, whether it’s reducing time and turnover for prior authorization or using evidence-based guidelines for clinical decision support. I probably can’t do it justice here, except to say that we have our fingers in many pies.

QPID’s work is driven by real problems facing clinicians and patients. I knew I wanted to join a company that makes a positive difference for its customers and community, so I was drawn to QPID because of their practical approach in using machine learning and AI technology to solve all kinds of challenges in healthcare.

What responsibilities are associated with your position as a Natural Language Processing Engineer?

I am a Natural Language Processing Engineer, which means helping EviCore and QPID more smartly handle the data that comes in to us from EHRs (electronic health records), especially free text, which is hard for computers to make sense of. Basically, I teach computers how to answer questions from a document that are obvious to people, such as ‘Does this patient have a temperature?’ or ‘What date did they see the doctor?’. In the day to day, I write code, run experiments, and analyze linguistic data.

How long have you worked at QPID and how does your role here stack up to your past experience?

I’ve been with QPID for a year and a half. Prior to joining, I was at a large tech company working on dialogue and speech-recognition applications. Obviously healthcare has challenges that differ drastically from those in the consumer market, so my role at QPID is very different for many reasons. The team size and unique obstacles that we face here at QPID mean we have more autonomy to try out new ideas in our problem solving.

One challenge I have working on NLP in the healthcare domain is that I didn’t go to medical school! Say, for example, I want to find all the key phrases related to some procedure or illness, so I’ll train up a model and when I’m analyzing results, I’ll see a list of drug names I’ve never heard of before. Then I have to go ask a doctor or do research before I can decide if my model is performing well. I’ve learned so many things I never thought I would about MRIs and chemotherapy since I started working at QPID.

What do you like most about your role and working at QPID?

I most enjoy uncovering new patterns in health records and using that to improve our algorithms and ultimately make users’ experiences better. Since health informatics and natural language processing are both quickly growing fields, there’s always something new and interesting to learn, which I love.

It sounds a little cliché, but I find QPID to be a work-hard, play-hard office, which makes it fun to come to work every day. We have a ton of fun team-building activities. The most recent one was an outing to Urban Axes to throw axes. Yoga at the office every Wednesday and snacks do wonders to get me through a hectic work week.

On a more serious note, everyone supports each other in both career and personal goals. There’s flexibility to work from home or take care of personal issues as necessary. Delivering the best product we can is very important to everyone, so sometimes that means staying late to get it done, although that doesn’t happen very often, in my experience.

What advice would you give to those seeking a job with QPID or in natural language processing?

NLP is a unique field in that humans use language for almost everything, so you can start almost anywhere. The widespread use of the internet means that data not as difficult to come by as it used to be, so I would encourage tackling a problem you’re passionate about. If you speak another language, try building a machine translation tool. If you like video games, try analyzing in-game chat logs. My experience is that if you start with a creative, fun challenge, you’ll want to learn more. Once you get to that point, in addition to classes on Coursera and edX, there are several good podcasts about linguistics, like Lexicon Valley, and Kaggle often provides data science challenges that use NLP. All of these are good places to start.