Before becoming chief product and technology officer at EHR vendor PointClickCare, David Pessis did stints at Amazon Web Services and Google. He spoke with Healthcare Innovation recently about some of the ways AI and automation are being applied to billing, compliance and data analysis in skilled nursing and long-term care. 

Toronto-based PointClickCare serves thousands of senior care facilities, skilled nursing providers and long-term care organizations across North America. 

Healthcare Innovation: David, could we start by talking a little about your career and how it led you to PointClickCare?

Pessis: I actually entered the University of Illinois pre-med, and my father is a physician. I always wanted to be a doctor. My sophomore year I took a computer science class, and I fell in love with it. That was the beginning of my software engineering career. But I always wanted to get back into healthcare.

When I look at why Amazon was so successful, in the most basic sense it's that they had this incredible data set and they built an intelligence layer on top of that data set to give us this great shopping experience. And it’s kind of the same thing with Google. They had this incredible data set, and they built this intelligence layer on top, which enabled us to get the best search results in the world. Well, PointClickCare has the most comprehensive senior care data set in the world. So I thought: How can I take that experience from Google and Amazon and build this intelligence layer on top and build AI capabilities that are incredibly useful to our customers?

HCI: From your resume, it looks like you worked on agentic AI solutions at AWS. Are there some agentic things happening at PointClickCare?

Pessis: We have an Advisor Suite. We have a Referral Advisor, which is an agentic-powered capability that helps our customers get patients into the facility faster, make sure it's a financial fit, a clinical fit, etc., so they can make really good decisions. We have a Chart Advisor, which is powered by an agent that detects regulatory risks, compliance risks, and litigation risks. We have a Billing Advisor that looks at the care delivered to the patient and makes sure that the facility is actually billing for the care they delivered. 

We also have other agents that help fill out forms. As you know, there's a lot of paperwork, so we want to help the caregivers fill out these complicated forms like MDS [Minimum Data Set], as an example. But your question is very pertinent. All these agents currently operate independently, but what we're marching toward quickly is figuring out how to bring these things together. 

Also, just yesterday we went to beta with our new ambient capability. Ambient records the entire session with the patient. The doctor can be there, the nurse can be there. It records the whole session, and fills out the assessment and the progress note. Now when you think about this agentic framework that we're putting together, as soon as that ambient session is complete, it will automatically kick off Billing Advisor. It will automatically kick off Chart Advisor. It'll automatically kick off the MDS Advisor without the clinician having to lift a finger, so that's where we're going really quickly. The goal is to get a lot of these end-to-end orchestrated agentic capabilities to our customers by the end of this year.

HCI: When we talk to health system leaders and the ambient technology companies they work with, they say that they started working on primary care or a specialist in an office talking to a patient across a desk. They said capturing hospitalists and and nurses in the hospital setting is more complicated and and more challenging. Is a skilled nursing setting somewhere in between those two as far as the challenge of capturing the conversation and getting it into a chart?

Pessis: I can't comment on the complexity or simplicity of hospital vs. skilled nursing. What I can say is that we have been working on this for a very long time to get it right, because accuracy is so important to us. We've built some custom models, and the empowered AI models are so good at understanding what's going on. Often during these care sessions, you have a doctor, a nurse, the patient, and they're all talking. It  recognizes everybody's voice, who's talking, and then it actually makes recommendations like, “you missed this part of the assessment.” Were only in beta, but our customers are super excited about it. It’s not just a Siri -like recording of my voice, but it is actually helping me fill gaps in the care session and helping me fill out forms and automate some of these workflows. I think because we worked backwards from the customer and invested really heavily in this custom model, it is allowing us to deliver a really good experience for these caregivers, but I don't think it's easy.

HCI: Now you will have to scale it to a large number of institutions and see what their experience is…

Pessis: That’s exactly right. Going back to the Advisor Suite, now that ambient's capturing that entire session, Billing Advisor has better data to act upon and make sure you're getting paid for the care; Chart Advisor has a more robust record, so it can do a better job detecting regulatory issues; MDS Advisor is going to fill out the form. The ambient is actually this catalyst to make our Advisor Suite way more powerful. It's like putting our Advisor Suite on steroids in many ways.

HCI: Could you give an example of ways it might improve outcomes through better handling of care transitions?

Pessis: The most prevalent feedback I hear from customers is about pain points around transitions — discharge from the hospital to the SNF, prior authorization, and concurrent review. This is when all that manual work, faxes, and phone calls happen. It's a huge burden on these facilities. In our Advisor Suite, we have a product called Referral Advisor. In working with hospitals, Referral Advisor takes all the documentation, surfaces the most important data to the administrator or the admissions coordinator, and helps them rapidly make a decision on a good clinical and financial fit for that facility. It's something that previously would take hours to do, and it now takes minutes. For SNFs, it’s not just about the volume of patients; it's also quality of patients — quality meaning their facility is equipped to serve that advance capability. Referral Advisor has really helped our customers get the right patients in at the right time, and we're dramatically improving their census.

HCI: We have always heard from the hospital and ACO side that getting access to data about what's happening with their patients in the post-acute care setting is difficult. Can AI help with that part, too — with building those connections to the ACOs and the hospital systems?

Pessis: I love that question. We're working on something right now around that. As you think about all this AI stuff that I've mentioned to you today, the nurse has a viewer, the clinician sees all this great data. But how do we expose that to the ACO or the hospital or the payer? Let's give them that same view into that data to help streamline these transitions of care and help all these stakeholders make better decisions. We're building that really quickly. We have a bunch of development partners that we're working with now to get that up and running.

HCI: Going back to the beginning of our conversation, you said what drew you to the company was having this huge amount of data on senior care, and putting an intelligence layer on top of that. But what about all the healthcare that's happening to those people outside of the SNF setting? Can you import that kind of data from an HIE or even broaden the amount of data about the patients that you're making decisions about?

Pessis: We use the HIE data to help make better decisions. We also have insights that are unique to us. We're able to detect patterns in how these patients are being taken care of and we can make recommendations that the clinicians would otherwise never have had because of the data that we have access to. It helps our AI make much better recommendations. I always tell people that the AI is only as good as the data that it has access to. The better the data, the better the AI, the better the insights, the better the outcomes for our customers. When you think about our 25 years of experience plus ambient, we have this flywheel going that enables us to empower our customers to deliver exceptional care.

HCI: Could this help detect potential problems with patients' health before they become apparent and might lead to a readmission?

Pessis: That’s exactly right. We have a predictive model around return to hospital. We have a machine learning model that we built in house that looks at every single patient and basically gives a predictive return to hospital score based on all the things going on while they're in the SNF. 
Our customers use that score to decide where they are going to spend most of their time during the day. Who are the patients at the highest risk?

HCI: Do you ever see examples on the customer side of organizations that have bought point solutions to solve particular problems? They may be AI-driven or not, but does it cause a problem if the solutions are siloed off from each other?

Pessis: That does introduce a problem. As I said earlier, the AI is only as good as the data that it has access to. What you'll see with a lot of these bolt-on point solutions is that their insights are not as good. They can't be, because they don't have access to that comprehensive data. From the customer perspective, it's another security review for them; it’s another place where their data resides; and it’s another log-in. It is this huge administrative burden for them to use a tool that has such a small slice of the workflow. Our customers love this end-to-end orchestration that we can offer, because it leads to better results, better capabilities, better insights because of that data that we have access to.