UCSF integrating oral and overall health records is a huge step that should not be understated! Given all the focus…
Amihai Neiderman is co-founder and CEO of Nym of New York, NY.
Tell me about yourself and the company.
I’m an engineer by training, in computer science. I did my bachelor’s in math and computer science when I was 14. Later on, I joined the Israeli army and did mostly cyberintelligence. After I left the army, I was influenced by my wife — she’s an MD-PhD ENT surgeon – and I started Nym. We do autonomous medical coding.
How do you distinguish autonomous medical coding from computer-assisted coding?
Computer-assisted coding is a tool that helps coders, giving them suggestions on the most appropriate code to use. It doesn’t make the final decision. Our fully automated solution requires zero human intervention. We code a chart in around two to two and a half seconds on average, then submit it directly to the billing systems in our clients’ facilities without anybody having to review those charts to make sure that the coding is correct. From there, it’s usually being sent directly to the payers. We completely remove the human from the loop.
Do payers trust the system’s consistent coding more than that of humans, where human coders may not all code the same way?
That was one of the reasons that it was hard to do automated coding until now. There was this lack of trust between the payers and providers, for obvious reasons, probably. The unique approach that we took when we started Nym is language understanding into clinical intelligence, what we call today clinical language understanding.
We can generate an audit trail that gives providers an end-to-end explanation of our entire thought process. If one of our clients ever gets a denial, an audit request, or even just during the onboarding process, they can assess and understand how we’re doing our coding. They can review those audit trails. We’re not hiding anything. We’re proud to show our internal thought processes and how we ended up deciding the right codes to use.
Payers who are receiving our audit trails are starting to become more interested in learning more about our solution, because we are not hiding anything and we have full transparency into our logic. They are usually afraid of fraud, waste, and abuse. We can show on the fraud part that we have nothing fraudulent in our process. It addresses that trust issue in a very interesting way.
Do you test in parallel with the customer’s human coders during implementation to measure the impact?
Yes. We know that our clients are moving something that is business critical for them. If we make a mistake, they could be exposed to compliance risk. They could lose revenue. We do a shadowing period, where when we do side-by-side coding for 30 or 45 days, depending on the client and the complexity of the project that we’re doing with them. During this shadowing period, we will have weekly or biweekly meetings with them and let the client choose choose any charts that they wish to review. They have access to our audit trails through a dashboard.
Sometimes they understand for themselves why we code something in a different way than their coders. Sometimes they’ll want to surface it up to us for an explanation of why we coded it this way when they might have done it differently. If we need to, we reconfigure our system based on their standard operating procedures.
How often do payers ask to review the audit trails, both initially and after they become comfortable with the system’s coding?
The payers don’t have direct access to the audit trails. It’s only if our clients decide to submit the audit trails as part of their appeal process if something was denied, or for an audit process that they have with the payers. Our relationships are directly with the providers who are our clients. But we do see from some of our clients in our periodic review that there is a significant drop in denials rates. Mostly because one of the things that we’re striving for is high coding accuracy, following the guidelines to the letter. You won’t see a lot of deviations.
That’s one of the beauties of using software to do coding. It is deterministic. You’re going to get the exact same results every single time. Once you become confident that you know that the results are up to your standard, you’re going to have reproducible results every single time.
Are issues with fraud, waste, and abuse usually caused by improper coding or are the provider’s notes themselves inaccurate?
We only code charts where we are fully comfortable with our understanding of those charts. If we see ambiguity, contradictions, or missing documentation, we drop those charts and let them go through the client’s ordinary process. Coders can reach out to the physicians if they need to. We code charts only when we are 100% certain that we fully understand everything in them. If there is any missing information, or if the chart might have any issues that will lead us to have wrong coding, we will just drop them. We are not taking any risk there.
Do clients implement your system primarily for efficiency or for accuracy?
Every organization has their own reasons for using our system. Sometimes it is speed. Provider groups that take five days to a week and a half to get their charts coded now see it happen in two and a half seconds. That reduces several days from their days in AR, giving them more cash on hand to operate their business.
Sometimes organizations, especially those who have tried outsourcing, usually offshore coding, encounter compliance issues and quality issues. Running an entire operation to try to reduce the compliance risk is expensive and not usually as fruitful as they believe. They are looking for a better solution to help them from a compliance perspective. This is where we come in.
In other cases, it’s reduced cost and improving their overall revenue cycle operations. We show the client that we are not just impacting directly their coding operations, but we will do it much faster than any one coder will do and we’re going to have a reduced number of denials, meaning fewer people that have to do scrubbing and running the appealing process. We affect the entire revenue cycle process downstream from us on the coding side.
Companies, including Google, are using natural language processing to support searching electronic charts and extracting data from free text notes. Will Nym use its experience in extracting chart information to extend beyond billing functions?
Definitely. When we started the company, we took a whole different approach into language understanding. What you see most of the bigger companies doing, almost all of them actually across the board, is using language models or AI solutions that weren’t tailored for healthcare, which has its own unique needs and problems. You gave Google as an example. They have an AI solution that solves their issues for search for understanding websites or news articles, but it’s not necessarily tailored for healthcare.
We built our own. We take a lot of pride in this. We built our own AI stack for language understanding, for clinical language understanding, and for coding that is specific for the healthcare domain. We’re a great coding company, but what we are really good at is capturing clinical narrative. Capturing the true clinical picture of what happened to a patient. Once you understand this, coding is a relatively easy thing to do. Because we built this strong technology and invested a lot of time in doing this, we can expand to other product lines or areas, using this technology to power new products.
Why does Israel produce so many successful entrepreneurs?
The vast majority come from a couple of well-known intelligence units in the army. One is the unit I served in, called 8200. It’s not the army that people envision or what they’ve seen in a movie. You get a lot of responsibility at the age of 18. You can become an officer when you’re 20, commanding soldiers and being tasked with some of the most complex things that could change the course of our history. They encourage you to do whatever it takes.
There’s a mandatory army service, so people get replaced all the time. You come in, do your three years of service or four or five if you extend it. New ideas are being surfaced all the time. People come in and challenge what people have done before them. You have a huge number of new ideas coming all the time and people are energized to try them. I was lucky enough to serve in a place where I was constantly asked to innovate and was given the backing of my commanders to do it.
What parts of your background have been most relevant to being an entrepreneur? What do you find most challenging?
I was doing cryptoanalysis in the army. When we started the company, we knew that we were going to tackle a challenging problem that some of the largest healthcare key companies have tried and failed to solve for the better part of the last 20 years. We needed some of the best problem solvers in the world to work with us, so a lot of our R&D folks are people who we knew back from the army. My co-founder Adam Rimon and I both served in the cryptoanalysis department, which was a good place to find great problem solvers. That has helped us with the early work of trying to prove that the unique approaches that we took to solve the problem could actually work. We had the right people to do this.
The challenge is that the army doesn’t teach you how to manage a company. The army has a very different management style than what you’ll see anywhere else. I felt pretty relaxed because of the nature of the business that we were doing back there, but it’s still not very similar to what you’ll see in a company. We had to learn a few things the hard way, but we try to fail fast, learn from it, and not repeat the same mistakes again. As long as we have a smart team that can follow the same kind of principle, then it’s OK make mistakes.
We just run, run quickly. We try to learn as fast as we can. One thing that we want to bring into the company and to the healthcare space is rapid prototyping. See if something works. If it doesn’t, you throw it away. If it does, great, you iterate over it and it creates value almost from Day 1 of the company.
Where do you see the company going in the next few years?
We are building great business in the coding space. The quality of our product and our results speak for themselves. We get our clients just from word of mouth, and our clients are highly referenceable. It sometimes amazes me that we are coding several million charts per year. We have three coders right now on the team who are serving as subject matter experts, but are not doing the coding itself. We are building a great coding company, bringing in work, adding to the client base, and expanding our footprint. We are going to be the top coding company in that area.
While we are doing this, we are also maturing our core technology. Our CLU engine gets better all the time. The more clients that we’re seeing, the more edge cases that we’re seeing, the better it gets and the smarter it gets. This allows us to take this unique core technology that we built and apply it in other areas that we’re still exploring. Coding is interesting because it sits between the clinical side and the revenue cycle side. We have the opportunity to influence the clinical side, to assist physicians both in the documentation side of the house and the revenue cycle process downstream from us.