Decimal Code (Techstars 2025) is an Ann Arbor-based startup that is fundamentally changing how health systems handle the administrative burden of medical billing. Born out of six years of research at Michigan Medicine, the company uses a patented process for custom AI model development to translate complex clinical notes into accurate medical codes, ensuring that healthcare providers are properly reimbursed for the care they provide.

Emilie Vallauri, Global Program Manager at Techstars, sat down with Joey Panella, CEO and co-founder, to discuss how a "Swiss Army knife" operator teamed up with academic researchers to turn an internal innovation into a high-growth commercial solution.

Emilie Vallauri: To kick things off, can you introduce yourself and tell us a bit about your background?

Joey Panella: My background is in research commercialization. I did my BBA at Michigan and an MBA at Booth. I spent time at the Venture Center at Michigan, where I acted as a "Swiss Army knife" for a hundred different research spinouts, helping them move from the lab to the market. While I loved the investing side, I really wanted to try my hand at being an operator within one company instead of a small piece of a hundred different things.

Emilie: And who are the other members of your founding team? How did you meet?

Joey: My co-founders are Mike Burns (Chief Medical Officer) and John Vandervest (Chief Technology Officer). Mike is an MD PhD, a practicing anesthesiologist, and also the Associate Chief Medical Informatics Officer (ACMIO) for AI at Michigan Medicine. John is the technical muscle — a data scientist with a background in computer science and a master's in machine learning.

I actually met them in 2019 when they pitched for grant funding while I was at the University. In 2021, as I was preparing for my MBA, my team at the Venture Center suggested I get more involved with them because their internal results were massive. They were looking for a business person to handle the commercialization side so they could focus on building.

Emilie: How did the idea for Decimal Code start? What problem were you solving?

Joey: It started with anesthesia billing codes. Mike and John were leading a research consortium comprising 84 health systems and 15 million collective procedural cases. They needed anesthesia billing codes to provide real-time quality feedback to clinicians, but those codes often take weeks or months to come back from the traditional billing process. They realized they could use the operative notes to predict the codes immediately.

When the CFO of the health system saw the impact, they asked how to get it everywhere, not just in anesthesia. That led to a process patent and the creation of 52 live models across various specialties at Michigan Medicine.

Emilie: For clarity, what does Decimal Code do, and why is it so important?

Joey: Decimal Code reads medical notes and predicts final billing and diagnostic codes. This is critical because today, that process is high-cost and often inaccurate, leading to lost revenue for providers. It basically helps hospitals "turn up the couch cushions" and capture the revenue they are rightfully owed from insurance payers.

Emilie: Who loses out when the coding is wrong? Is it just the hospital?

Joey: It’s mainly the healthcare provider because they aren't reimbursed properly by insurance, which is tough on low-margin organizations. But it also affects patients — incorrect diagnostic codes can actually follow you and impact your future care and treatment plans because clinicians see you in a "wrong bucket".

Emilie: What are the largest challenges you’ve faced building the product?

Joey: The biggest problem is finding a "source of truth" to validate our models. National medical coding accuracy is only about 83%. If our model predicts a code that differs from what a human coded in the past, we have to audit it and often find that the model was actually correct and the human made an error.

Emilie: That’s incredible! Apart from the precision of your models, what sets you apart from competitors?

Joey: It’s our procedural expertise and our patented methodology for building custom AI solutions from a provider's data. Many competitors don't have coverage for procedures. Because we came out of an academic medical center with 15 million cases to train on, we can build effective models more rapidly across different specialties.

Emilie: What’s your proudest recent milestone?

Joey: We now have three health systems outside of Michigan Medicine. One is officially deployed and two are actively deploying. Transitioning from a cool research project to a legitimate commercial entity with paying customers outside our home base is what I feel proudest about.

Emilie: And are you fundraising right now?

Joey: Yes, we are actively closing a $750K pre-seed round. We have $525K committed from current investors. This capital is to help us build out the Decimal Code engineering team as we scale to these new sites.

Emilie: Turning to the news, ChatGPT recently launched "ChatGPT Health." How do you see that impacting your business?

Joey: I think it's great — it’s the next wave of condensing information for clinicians and patients. However, coding is a categorization problem. We’ve found that traditional machine learning is actually better than LLMs for bucketing a case into a list of 10,000 choices. Healthcare is a boat with a million holes; we’re just tackling a different hole than OpenAI.

Emilie: What advice would you give aspiring founders starting out?

Joey: Two things. One: offer to help for free on projects you find interesting. It's the best way to get your foot in the door without a science background. Two: don't let perfect be the enemy of good. You can get paralyzed trying to find the "perfect" door to go through. Just make a good decision and move forward.

Emilie: Finally, why did you participate in the Techstars accelerator, and how did the Northwestern partnership specifically help?

Joey: We highly prioritized the relationship with Northwestern Medicine. Having two of the largest academic medical centers (Northwestern Medicine and Michigan Medicine) in the country as early pilot partners and equity holders provides huge validation. Northwestern gave us the "answers to the test" for their procurement process, which helped us hone a sales strategy that we've already used effectively at other sites.

Read More