DSOPro: Tell us about your career and what brought you into dental.
Before I got involved in the dental world, I ran data science for a company called Publicis Groupe, the world’s second-largest advertising holding company. I was monetizing and building data structures for Fortune 10 and Fortune 15 companies for many years.
Ultimately, I decided to leave and try my hand at some other stuff and started exploring opportunities. My brother-in-law has long been involved in dental IT. His company, Fireworks, helps dentists set up internal systems, particularly with imaging, practice management, and digital practice integration. He’s seen all the different evolutions dental offices and DSOs have gone through over the last 15 years.
When we started talking, I realized that dental is underserved, that as an industry there are not that many people really pushing the envelope compared to other industries. I saw an incredible opportunity, not only for a new business, but also to help people live better lives and to help dentists in their mission to help people. So, that is what brought me into dental.
DSOPro: What are other industries doing that dental is not?
I’m coming at this from a data perspective. When I compare dental to other industries, we do not have as sophisticated a knowledge of our patients as other businesses do. For example, total patient lifetime value, total patient churn, total patient practice composition or procedure composition, lookalike models to understand tranches of patients, ways to start to think about modeling patients—know the ones you have, the ones you’d like to get, and the ones you’d like to expand to in the future. Many other businesses are very sophisticated, particularly in healthcare and pharmaceuticals. There is a lot we could do within dental to help catch up.
One of the ways we at GoDental.ai are focusing on helping is by bringing more sophisticated data use to dentists by applying the new technology of AI and large language models (LLMs). ChatGPT is arguably the most famous LLM, but there are many models out there. A model is just a mathematical formula. And these LLMs are also just mathematical constructions on data.
We are using LLMs to create the perfect dental insurance claim the first time and every time. We’re using LLMs like ChatGPT to help write insurance claims. Why start there? When I looked at the dental industry, I saw a lot of amazing businesses offering things like practice management software (PMS), helping them understand appointments, understand patients and move them through the practice from the appointment out to payment. But I did not see anyone addressing the core issue of the insurance claim, which was the claim itself.
Dental insurance, as we all know, is absolutely labyrinthian. There is very little standardization across plans, businesses, states, and insurers. There are thousands of subsets of plans, and they change all the time. They’re constantly being reworked and renegotiated. So, that means we have a very diverse landscape of dental insurance. Because of that, until now, this has been an incredibly manual process. A lot of humans are sitting on hold with insurance companies trying to solve problems for their patients. This seems like a bad allocation of humans because what are humans better for? They are great for empathy, communication, and problem solving.
We are much better served having these humans treating our patients, listening to them, and helping them navigate their dental challenges. That is superhuman work compared to punching in prompts and waiting on hold to reach an insurance company. That’s what a machine is good for. So, we are very motivated to empower dental team members to spend more time with patients.
SPONSORED|
The fastscan.io™ Scanning Solution combines Medit® i700® technology with the digital workflow of glidewell.io™, plus personalized ongoing support from a Glidewell technician. Includes a portable cart, laptop, monitor, and no monthly or subscription fees. |
DSOPro: How does the system work?
When we think of humans and how AI is becoming integrated into our world, fundamentally what we’re really talking about is trust. Can I trust what is coming out of this computer? Can I trust where my data is going? Can I trust it has not created something false?
We also don’t want to give dentists another tool to deal with. They do not need yet another program to learn, another thing added to their tech stack. So GoDental.ai sits on top of the existing PMS, whether it’s Dentrix Enterprise, Eaglesoft, Open Dental, etc. GoDental reads what’s on your computer screen as you are entering a claim in the PMS that you already know and trust. Then a popup, like the old Microsoft Clippy, the little paper clip character, offers suggestions for the human to input into that claim based upon our AI algorithm. Staff members have control over what they choose to copy and paste from our recommendations or what they add themselves. Our LLM focuses on helping individuals populate those claims from past procedures. We are focused on making the process simpler. This is the dentist’s or a DSO’s business, and it needs to be a simple, controllable, visible, totally transparent, secure experience.
DSOPro: Could you summarize the typical RCM process now and what GoDental is doing to simplify it?
First and foremost, other people have created excellent products. The current RCM and PMS softwares are very sophisticated systems that are helping patients get appointments, reminding them of their appointments, and other subsequent and very important tasks. We are focused downstream on that single claim. We want to ensure claims are absolutely perfect, so they are not rejected.
However, if it is rejected, our systems are self-educating learning systems that loop back and learn from that rejection. With multiple offices, every rejection adds to the knowledge base. This means individual office knowledge is no longer siloed. We are at this incredible moment where we can take a learning from one place that got rejected for reason A, adapt that into our AI system, and apply it to another office so they don’t experience the same type of rejection.
One thing about dental is that everything is so unique. We want our system to be able to learn and adapt as insurance, patient care, and imaging technology adapts. We want to help create a point of control around the submission of the claim itself, so it flows in the RCM 110%. We’re verifying patient eligibility, we’re making sure the deductible is present and that the insurance will reimburse it, because it all feeds directly into the claim.
Our little helper, if you will, pops up and says, “There was a filling on this tooth. Don’t forget to mention the facet.” Or you’re doing a bridge claim, and it says, “This type of procedure needs a narrative.” We are training GoDental to generate narratives the dentist can read and refine.
This system checks for all those little things that need to be filled out for a claim and communicates them to the staff. We’re submitting claims faster and easier that are returned more quickly from the insurance company. And we’re seeing far fewer denials.
SPONSORED
|
DSOPro: How are you ensuring data security?
We are incredibly serious about security. Not only are we HIPAA-compliant, but we are also SOC 2 compliant, which is another layer of data security validation even more rigorous than HIPPA. Also, a lot of AI companies pass the data into a cloud AI server like OpenAI, a ChatGPT, and that server does the computations and sends the result back to the user. We have written our own proprietary LLMs that live safe and sound on our own server. All the AI computing is done locally on our own LLMs, which allows us to be more data privacy focused.
Furthermore, we intentionally trained our LLMs using no personal identifiable information (PII) whatsoever. It was our goal to teach the AI to solve these problems without using PII because the major point of a data breach is to obtain PII. Frankly, it took a lot longer to build these models because we added that layer of complexity. We are very proud that these LLMs do not use PII to create perfect insurance claims. We learned from the mistakes other people have made.
DSOPro: Tell us more about the company.
The company is very new. We have been around for only about seven months. We are currently working with five partner dentists who are willing to help us learn and train our model using their data. We are also talking to several DSOs about installing our system and are looking for more DSO partners interested in deploying this in their offices. We think that it could be an incredible game changer from a DSO point of view because of the power of this model to help with claims rejections, and subsequently, to ease accounts receivable.
Our current data sets are being trained on 4 to 5 million records. Each one of these is an appointment, if you will. So, we’re running about 5 million appointments to train our model. As our partner dentists bring in new patients, we’re learning from those experiences and banking that knowledge back into our model as well. We’ve had amazing success with our model building and have experienced open arms when talking to the industry. Everybody has been so supportive, and I’m thankful to be working in this space.
DSOPro: What happens when a claim is rejected?
Whether a claim is successful or denied you get an explanation of benefits (EOB). The EOB is how we are actually learning this model. It reads the EOB and understands the code used, the reason it was declined or accepted, and the amount that was reimbursed.
DSOPro: How does this impact revenue?
In multiple ways. The first is by reallocating staff towards more patient-focused time. That allows patients to feel better. There is less churn and higher retention. Patients see more value in your office.
The second way is that if an insurance claim is rejected it needs to be resubmitted and that process is quite laborious. We wanted to nip that in the bud by doing it right the first time so all the downstream problems disappear.
SPONSORED
|
DSOPro: Would you like to add anything about the company?
One of the things that sets GoDental apart is that while we have team members from within dental, most of us are from outside of dental. Our head of AI was the head of AI for Shutterstock and worked with Amper Music. Dr. Cole Ingraham has been in AI for over 15 years, which is about 100 years in “AI time.” So, we have deep-seated talent in the AI space, a massive differentiator because this technology is so new that it is hard to find individuals with such sophisticated skill sets.
We are approaching this challenge from a non-dental, straight-up software, “consumer first” point of view. This allows us to apply perhaps new ideas that having a different point of view brings.
DSOPro: What do you think about the scrutiny and the potential limitations that are being put on AI? Do you think it’s a good thing to monitor it to make sure it’s used appropriately?
First, AI is using data that is generated by humans. All data is computers capturing human behavior, like clicking on a website or inputting a search term. So fundamentally, this is a people puzzle. And because of that, humans deserve respect. We deserve privacy, and we deserve to not be manipulated. So, these systems need to be observable and transparent.
We can, at any time, know what the algorithm is doing, why it’s doing it, and trace these things back to better understand why it’s making those decisions. Because we believe at the core that this is about respect for humanity.
Now, if you look at AI more broadly, from an industry or a technology point of view, technology seems to grow in leaps and bounds and in directions that no one really knows where it’s going. That’s the nature of humans building things.
I am reticent to paint with too big a brush regarding controlling AI. I believe we absolutely need oversight. We need systemized ways of making sure that things are being treated responsibly. But I have a hard time with a legislative approach to this type of control. What gives me pause is that there’s an equal chance of life-changing positive things coming out of it as there are totally negative apocalyptic things. Ultimately, we don’t know. The future is literally a 50/50 because it’s all statistical coin tosses.
I agree we need to be very thoughtful about how we deploy this, particularly in healthcare. There should be checks, which we are 100% committed to. Perhaps I believe too much in the goodness of humanity, but I’m just reticent to close down potentially amazing opportunities out of fear and misunderstanding.
I don’t want to shut down the possibility of incredible advancements because of fear of a bad actor. Respect your fellow humans. Treat them honestly and fairly and that’s what you get back. And don’t let the bad actors prevent us from building something that could be radically transformative for us in a positive way for society.
More from the Newsletter
Tim Rich is the CEO of GoDental.ai. Previously he served as the CDO of MuseWorks Labs, a web3 company bringing live theatre into digital collectibles. Before that he was the EVP of Data Science at Publicis Groupe. When he is not building with data, he is fly fishing.
GoDental.ai
GoDental.ai was founded to use AI to help dentists create the perfect insurance claim the first time, every time.