How AI Can Help Speed Up Physician Credentialing ChoresHarman Dhawan, founder of Bikham Healthcare, on Streamlining Frustrating Processes
Physician credentialing and healthcare billing are two areas that can be dramatically improved by using AI technologies, said Harman Dhawan, CEO and founder of Bikham Healthcare, a revenue cycle management services firm that is applying AI to its services offerings.
When a new physician joins a medical practice, the process of credentialing is time-consuming and arduous, Dhawan said. "Basically, it's a very extensive background check. You check against the qualifications, degrees, licenses and all of that," he said.
"And then you grant certain privileges. If it's an orthopedic surgeon, you grant them the privileges to perform surgeries, access to certain areas. You have to enroll them with payers so that when they see those patients, the hospital or the health system can actually bill the payers and get reimbursed for those claims," he said.
In many cases, credentialing a new physician can take two to four months, delaying the ability for clinicians to treat patients and bill payers for their services.
"But through the use of Bikham Healthcare's upcoming AI-driven software credentialing service, Provider Passport, which is currently in beta and slated to launch in the first quarter of 2024, the company is aiming to reduce the process of credentialing and onboarding physicians to about seven days.
"This is a big pain point. You've hired a new physician, but they're not seeing patients for months. Not only is it a drawback for patients, but it has a huge economic impact on the health systems," he said. "So this can help change all of that."
In this interview with Information Security Media Group (see audio link below photo), Dhawan also discussed:
- How AI can help improve clinical decision support;
- Billing and administrative challenges facing healthcare sector entities and how AI can help;
- Fraud and deepfake threats involving AI.
Dhawan, founder of Bikham, has previously launched and grown other businesses, both organically and through acquisitions. With over 20 years of entrepreneurial experience, he initiated Bikham in 2005 with the initial offering of revenue cycle management and later diversified into physician and payer credentialing, clinical diagnostics with acquisitions in the laboratory space and other services.
This transcript has been edited and refined for clarity.
Marianne McGee: Hi, I'm Marianne Kolbasuk McGee, executive editor at Information Security Media Group. Today, I'm speaking with Harman Dhawan, founder and CEO of Bikham Healthcare, a company that provides revenue cycle management, billing and other services to healthcare sector entities. We're going to be discussing how his company is utilizing AI. To begin this discussion, Harman, for people who are not familiar with your company, please describe briefly what your company does, the types of healthcare sector entities that you service, and the mix of companies that you work with.
Harman Dhawan: So I'll keep it pretty simple so that everybody can understand. So we help providers of all sizes - individuals, solo practitioners, hospitals, health systems, clinics and facilities - to get paid faster. Now that is the RCM cycle, which involves your credentialing, billing, follow-ups and the whole gamut of services. So in short, we help providers get paid faster. We build and deploy technologies to help with that.
McGee: So how is Bikham utilizing AI right now in the services that you provide to clients? And what type of AI are you using?
Dhawan: I'll rewind a little bit, right here, and will address the AI aspect of it. Companies have been touting the AI word for years now. And it was not exactly AI, but it was more RPA, which stands for robotic process automation. So it's basically coding and making sure certain actions take place. And only until, in November last year, when OpenAI launched their AI model, ChatGPT, that AI got available to the common person. That's when the cat, exactly like they say, got out of the bag. Everyone had access to it. So now is when we will be seeing a lot of impact - processes, services and efficiencies. Let me just explain how AI will impact. For example, let's say if a doctor right now builds a claim, they have a billing company or someone else doing it for them. And it's a whole process that they have to follow. They have to see the consult, the notes, and then bill accordingly, what CPT codes, etc. Now in the too distant future, you could have a bot listening in. For example, if it's a chiropractor or a dermatologist, someone goes in and says I've got thinning hair, I need something. And the bot in the background is listening to all of it. And based on that consult, it runs it, queries it and with all that vast database comes up and builds the claim. And once the provider approves that, it gets billed. And then of course, there's automated follow-ups with the insurance companies based on all the rejections or the questions that get asked. So all of this process can be automated and will be automated intelligently. So in a nutshell AI is that. And in terms of administrative services, the other side of it, for example, right now, for a provider, if he/she enters a health system, it takes anywhere from two to four months for that provider to start seeing patients. What does that involve? First, there is a word called credentialing. It's a process wherein you credential the provider; basically, it's a very extensive background check. So you check against the qualifications, degrees, licenses, and all relevant criteria. And then of course, you grant certain privileges. If it's an orthopedic surgeon, you grant them the privileges to go perform surgeries, certain access to areas, among other things. And then of course, you got to enroll them with payers so that when they see those patients, the hospital or the health system can bill the payers and get reimbursed for those claims. So all of this process is like a maze, interdepartmental communication delays and many other issues. And it takes two to four months. And in that time, those providers are not available to patients, they can't see them, you got scheduling delays, etc. And of course, it's a monetary loss as well for hospitals, facilities and all systems. So how do you fast track it? I believe, even with technology of 10 years old, it could have been done, but it's not been done. So what we've done is credentialing has been automated till a certain extent, which is your background checks. People are doing it; there are a couple of companies that are doing it. But in our case, we've come up with a product called Provider Passport. So it's going to be a massive provider directory on the front end. And on the back end, if I'm a health system and I've got Provider Passport deployed in my internal systems, I just need an NPI number. And within seconds, your credentialing will be done. Your privileges as per your bylaws or whatever it says, it will be fed into the system during onboarding but it will run it up. It will go to the board for approval and automatically other bots in the background will file the applications with the insurance payers and follow it up intelligently. So it's a large language model in the back end, which basically then analyzes the responses that are coming from the payers and does appropriate follow-ups, accordingly, till resolution. So from two months, what we're aiming to do is have a provider onboarded within seven days - that's a big claim that we're making. And we're almost there, and it's going to be amazing. In addition to other things, doctors will be readily available to take appointments. So that's the kind of impact AI is driving right now. So just imagine into the future when decision support comes in, and for example, take a chiropractor, you're going in with a lumbar problem, or L1-L2, whatever problem you have, and the doctor is conversing. But the doctor would have access to the AI that can go and query huge sums of data and say, with this case, what you're prescribing is this, but you got a 70% efficacy system with a combination of these drugs or these procedures. And that's decision support. So of course, as opposed to one mind treating you that mind now has support of all the minds that have ever specialized in that area. And that's been documented online. So it's exciting times.
McGee: Is your company right now working with the credentialing technology you were just talking about? Is that something that you're already doing?
Dhawan: Yeah, we are doing credentialing, Marianne. So credentialing, privileging and enrollments, we are doing it currently. And we're set to launch in the first quarter of 2024, the Provider Passport - that's the name of the AI-driven software. So we're currently onboarding beta clients. See, this is a big pain point. You got providers, they are coming into the system, but they're not able to see patients two to three months. Not only is it a drawback for patients, but it has a huge economic impact on the health systems' bottom line. And so it's going to change all of that. So from like days, months to minutes - that's what we're striving for. And it's here.
McGee: So with that said, for instance, with the credentialing, it'll dramatically cut the time it takes for the doctor to be able to start seeing patients. What other examples of AI are you encouraged by that either your company might be interested in pursuing or that you see as potentially very beneficial for the healthcare sector overall, moving forward?
Dhawan: I think one critical area is the amount of time physicians and healthcare professionals spend in administrative tasks. One we've already addressed, which is credentialing, just making sure that the providers are able to onboard faster. The other is billing. It's a huge pain point. I believe it's something like I think more than an hour or around two hours that a physician or a healthcare professional typically spends making sure that everything is in order, so they get paid. So that's something that we're looking at. So you got this software sitting in a laptop, while you're consulting with the patient. And it's picking up everything, what you're talking, what you're not talking, and then it will give you a bill out everything. And the beauty of it is you got machine learning, where if you correct something it's going to learn, it's going to get better at it. And of course, instead of having people follow up, you got intelligent engines that understand you got the large language model coming, you can train it with vast amounts of data. And it's following up. So which means and all of this can be done pretty efficiently and cost-effectively. Right now, I believe on an average, a health system or hospital pays anywhere from 3% to 8% of their revenue in making sure the billing services are performed and they get paid. So all with AI coming in all of these efficiencies, it will be able to drive in an amazing way. So that is one area we're looking at, just to basically streamline billing. Billing is a puzzle, a maze, CPT codes, etc. So with AI, it's going to get simpler and just trying to make sure that the payments come in faster.
McGee: So in terms of potential risks of using AI in healthcare, including those related to data privacy and security, what are you most concerned about? And how are you planning to approach those types of risks in your use of AI?
Dhawan: Whenever a new technology, in this case, I won't even call it technology. It's a huge paradigm shift with AI coming in. I think it has to be assisted with humans first, for sure, wherein they are validating each and every step along the way. For example, in our case, SMEs, which are subject matter experts, let's say in credentialing, they were busy answering clients' questions, trying to resolve their processes, their requests, and trying to fulfill their orders efficiently. Right now, what they're busy doing is basically trying to train the model, so that it's better than them in a way. So those SMEs are now involved in making sure that we are able to effectively and efficiently train those models. And of course, it's going to make less mistakes than a human, for sure, there is no comparison.
McGee: Looking ahead into the upcoming year, what sort of AI developments are you keeping your eye on maybe things that you're not doing yet at your company, but are watching carefully for potential application at your own company or for your clients?
Dhawan: One of them is, which is fascinating, you can train a model to talk and write like you. So if you got vast amount of data with you about your recordings, phone calls, etc., you train it with that. You got a lot of emails, let's say going back 10 to 20 years, you put it into the system, and then it will talk like you, write like you - just like you. If you keep using it for a couple of months, or even not even that, you will just say that it's as good as you, it sounds like you and that is something that can be applied in a big way in healthcare as well. One of the areas I believe other than this is your decision support, which is massive. In the sense, right now we go to a doctor, and we are just dependent on whatever they prescribe us. It could be well on point, or it could not be on point. So they're also under pressure in seeing more patients, they got appointments, etc. But to have that kind of decision support wherein whatever you've been prescribed or whatever procedure you've been recommended, it's been vetted against the database of all human population. This is the best treatment there is that's documented. So I think that would be huge.
McGee: You gave the example about the emails, that perhaps you could teach an AI tool to write a batch of emails out to people that you need to send emails out, and it sounds like you, but you don't also want that sort of being spoofed by fraudsters who are coming out sending emails and sound like you, they have your signature on it, but it's not you. How much vetting do you think there needs to be in terms of humans looking at whatever it is that's produced? And where does that fit in do you think?
Dhawan: I think it's going to be spooky, it's going to be so good. So you going to have probably laws that are going to come into place. Those deepfakes and whatnot that's coming out, I'm sure Facebook's going to have algorithms that's going to detect all of these fakes, etc. So I would say, pretty soon, you're going to have laws around it. So if you are using an engine that is emulating you, you probably have to declare it just like if you're recording something you got to say, this conversation is being recorded. So this would be probably driven by government, law and others, because AI is progressing at a scary speed. And it's fun. But along with fun, you got to keep it balanced. So I'm sure the government is busy at it. And they're going to come up with laws pretty soon that are going to define what you got to disclose.
McGee: And as you talk to healthcare clients, what is it that they're looking to do with AI that perhaps you haven't done yet? Or, something that maybe isn't in the works, but something that you think will be an interesting application that would help them make the most of their work or productivity and streamline other administrative things that they perhaps still struggle with, like you mentioned billing, other areas?
Dhawan: So one would be that we've addressed already, and we are set to launch the Provider Passport in the first quarter of 2024. That takes care of your credentialing, privileging, insurance enrollments and licensure renewal. So now, if I'm a physician, our system is going to go in proactively renew your licenses for you. So I'll give you an example. We were showcasing our platform to this hospital, and they were asking us questions. What about this? Who will enroll? Who will preview? I said no, it's just a system, you turn it on and you forget about it. If there's an exception, if there's a sanction against or something that you've defined that all right, a human needs to look at, it's only then that you will be notified. Otherwise, the system works for you. You don't have to work it. So that's one side of it, which is your credentialing and privileging. And you're getting these kind of reactions, they're surprised. They're like, wow, is it? Yes. And it's not like Star Trek or next-age technology; it's just simple AI and RPA that you use, and you're able to do it. And other than that, of course, its billing and decision support. These areas, I think, have to be addressed first. From there on, it's more about, I believe, government, level playing field, in terms of the legacy systems where you have access to more data freely, of course, within the confines of privacy, so that more entrepreneurs can enter this space and disrupt. It is ripe for disruption. The only problem is HIPAA - all of these regulations come in and they are kind of hurdles. Before you even begin, you got to have lawyers, teams, etc. I'm talking huge expenses here before you even start to play around with it or tinker with it and innovate. But it's coming in the next year or two. It's going to be massive efficiencies driven across the whole sector.
McGee: Well, thank you. Harman. I've been speaking to Harman Dhawan. I'm Marianne Kolbasuk McGee of Information Security Media Group. Thanks for joining us.