Reducing Medical Coding Complexity with Nym Health

By Michael Canino

GV's Ben Robbins interviews Nym's CEO, Amihai Neiderman.



In 2016, there were 883.7 million patient visits to healthcare providers in the United States. Every single visit required the creation of a medical bill with codes that health systems submit to insurance companies for payment. Translating medical notes written by caregivers into billing codes that correspond to various diseases and treatments is mostly a cumbersome, manual process. It's prone to errors in part because it's done by hand, even in our increasingly digital world. Virtually every health system our life sciences team has met over the last ten years shares this common pain point.

During my time as a resident in psychiatric training at Massachusetts General Hospital and Mclean Hospital, I became intimately familiar with this tedious process. Several times I found myself sorting through months-old notes to update clinically meaningless sections to ensure compliance with billing requirements. For example, rather than write "all systems reviewed and were normal," I needed to specifically state which normal systems were reviewed, writing that there was no nausea, fever, runny nose, joint pain, etc. And this was for patients who came in to review their psychiatric medications! Increasingly on drives through Boston, I have been struck by how many buildings are being erected, primarily for armies of human labelers to process medical notes into billing codes and submit those codes in billing forms to insurance companies. I often thought that there has to be a better way.

Enter Nym Health. It's a startup created in 2018 by Amihai Neiderman and Adam Rimon, two graduates of the elite Israeli military unit Unit 8200. Amihai and Adam have recruited an impressive and eclectic team of physicians, computational linguists, and software engineers who are hard at work in Tel Aviv, developing the company's medical coding technology. I was introduced to the company by Zach Weinberg, a Nym angel investor and co-founder of GV portfolio company Flatiron Health (acquired by Roche in 2018). Zach excitedly told me how the company had finally figured out how health systems can efficiently and accurately convert handwritten medical provider notes into digital records, and I was intrigued. GV's life sciences team had been interested in this space for quite some time, but we had yet to come across a company with Nym's unique approach. I met the team via video calls and immediately jumped at the chance to invest — and today, Nym Health announces its $16.5M Series A funding led by GV.

Nym speaks the language of healthcare in a way that medical providers and computer scientists both understand. At GV, we talk about finding companies that bridge the divide between breakthrough technology and healthcare. Nym actually does it, with a new approach they call Clinical Language Understanding, because they recognize the massive opportunity to automate medical coding. Some machine learning methods have limited usefulness since health systems are often not receptive to black-box solutions they don't understand. And it isn't always easy to use AI-based approaches to provide feedback to providers about how to document their work better.

With Nym's novel technology, they have created a rules-based approach that converts notes into billing codes while simultaneously generating real-time audit reports, which are critical to health systems. Nym uses these reports to replace burdensome auditing processes as well as to give clinicians feedback about missed opportunities to document their work in order to optimize insurance reimbursement. For example, if a physician orders an echocardiogram, but does not mention that she reviewed the results in her note, the practice loses out on revenue. That physician is likely to get a phone call from a coder, much like the ones I received during residency. A human coder might identify this omission and email the clinician to ask for clarification and to amend their note. But a human coder could also miss the omission, and then the health system would lose revenue. As opposed to Nym's approach, a machine learning-based system may correctly identify that some payment was missed, but not determine which of the encounter's several characteristics led to the missed revenue. Nym's system can accurately identify the specific omission and do so at scale without errors — a groundbreaking development in this category.

Nym is already working with 40 hospitals in the U.S., and they are just getting started. As Nym continues to address one of the most error-prone areas of our health system, they have an opportunity to reduce a measurable part of healthcare costs across the country. No matter which hat I'm wearing — as a physician or an investor — that's something to get excited about.


See it in action