“Interoperability is essentially front page news now, and I think that’s exciting,” Don Woodlock, Vice-President of HealthShare Platforms at Intersystems, said.
“We spent all those years adopting EHRs, and now we’re wanting to get the most out of them. Now we have the digital data, so it should be more liquid and in control of patients and put to use in the care process, even if I go to multiple sites for my care.”
As ONC and CMS prepare to digest the voluminous public comment on their proposed interoperability rules, especially the emphasis on exchange specs such as FHIR and open APIs, he sees the future only getting brighter for these types of advances as data flows more freely.
“We’re in the interoperability business, and we like having data being more available and more liquid, and systems being more open to getting data out of them,” Woodlock said.
Want to publish your own articles on DistilINFO Publications?
Send us an email, we will get in touch with you.
“A lot of customers are starting to embark on their journey with with FHIR, and they’re really bullish on this as well: having a standards-based API way to interact with medical record medical record data,” he added. “We’re sort of at the beginning of that as being another giant breakthrough, I think, in terms of liquidity of healthcare information.”
InterSystems has touted some client success stories recently that are indicative of current interoperability trends.
In one, NY Care Information Gateway, a New York City and Long Island-based regional health information organisation, is using HealthShare technology to run its master patient index, with its matching algorithms to closing gaps in care.
The tool has helped NYCIG sync and de-duplicate patient records, enabling the sharing of real-time updates to the state’s master patient index and boosting care coordination efforts for the network’s provider customers. That’s helped them boost quality improvement efforts, lower readmission rates and increase reimbursement rates under value-based care.
In the other, New York-based Healthix has built a new initiative atop the InterSystems HealthShare platform that enables providers to gain much more visibility into where their patients are receiving care. Billed as the largest public HIE in the country, Healthix wanted to tackle the problem that millions of the alerts it generates could not be sent to providers and care managers without patient consent – something that could slow or impeder the care management and intervention process.
With its new Essential Alerts project, clinicians can get notified when their patients are admitted been admitted to the EDs at most area hospitals with chest pain complaints, for instance, even if they haven’t given express permission for it. The alert is limited to the patient’s name, initial symptoms, hospital, date, and time. That’s enabled a big increase in the number of alerts fired, giving physicians more valuable insights into their patients’ care while still complying with New York privacy laws.
AI poised to bring enhanced value to interoperability
Both HIE projects depend on finely-tuned algorithms, and as artificial intelligence continues to mature, Woodlock said the potential for them and others like them to further improve care coordination and population health management is big.
“We’re really bullish on the potential of AI and in healthcare I’m always interested in seeing different use cases coming to fruition,” he said.
“I personally have been less focused on the ‘out-doctoring-the-doctor use cases, because I think that AI obviously can be targeted toward knowledge work, where people are taking in data and making decisions about that data, and it seems like the physician role is one of the most complicated knowledge workers we have.”
Patient matching is an obvious and compelling place to start, and is “ripe for artificial intelligence to help out with,” according to Woodlock.
“You can watch a person go through that, look at essentially the features of the two potential patients to see whether they match or don’t match, and eventually learn the relationship between the features of the patients and whether they should be matched or not. There’s a lot of there’s a lot of very good artificial intelligence use cases that aren’t necessarily getting into care decisions.”
The NYCIG initiative does use InterSystems matching algorithms, but doesn’t use AI per se yet,” he explained. But the company is working on the HealthShare Patient Index technology to “use artificial intelligence so it’ll match in better ways,” he said. “The new version of the product we’re working on will use AI to improve the matching algorithm.”
But even using the existing version of the technology, NYCIG, whose network serves millions of patients at hundreds of provider organisations in the New York City area, has been able to realise some notable gains as they navigate value-based reimbursement, Woodlock said.
Both when it comes to reducing readmissions and improving management of chronic conditions, the HIE – and others like it – have helped customers with accountable care.
“When you’re leaving the hospital – if there’s an automatic alert to your primary care physician or your care team that you are even in the hospital; you’ve left the hospital and here’s the discharge summary – they can be a more active part of a successful transition into the ambulatory environment,” Woodlock said.
“Having that more robust awareness in the community of the fact that you were even in the hospital has a positive impact on this readmission rate.”
NYCIG is using the technology to help do just that, and another Healthshare Customer, Rhode Island Quality Institute, has reduced its own readmit rate by 15 per cent, he said.
That’s a “significant drop,” Woodlock said. “And since you’re not paid for those readmissions, that’s real revenue for you.”
For another example, he pointed to the fact that practices linked to NYCIG improved data on the care delivered to their diabetic patients. Many of them, for example, were able to discover that their patients weren’t just getting A1c tests with them, but with other providers too.
“When they had that more comprehensive view of their diabetic population, they actually had a larger percentage of patients that were getting A1c tests on time, and they actually moved up in their quality rating, and so got more revenue,” Woodlock explained.
“Their patients were actually getting the right care, they were just getting it at other sites that they weren’t aware of. So now that they had that unified picture, they were able to demonstrate that they had a better quality quality outcome.”
Machine learning helping power pop health alerts
Healthix is one use case where AI and machine learning are at work, via InterSystems partner HBI solutions – which is helping the HIE’s customer realise the benefits of advanced predictive analytics.
“We do the work of aggregating and normalizing and deduplicating the health information across multiple sources, and what they do is leverage that data and build machine learning models that predict the likelihood of occurrence of many things happening in the next 12 months,” Woodlock said.
“They can predict for each patient the chance that that patient will be readmitted, will go into the emergency department, will get diabetes, will have a suicide attempt, will have an opioid problem,” he said.
“They can basically use regular machine learning techniques, using the clinical data correlating that with events that actually happened to actual patients to generate a model that equates those two and then what they do is they run that model on all the other patients. When any Healthix members pull up a patient then on the top right part of the screen they’ll see these probabilities.”
For example, a hospital might be able to see from the AI-powered analytics that a patient who presents at the ED with a broken bone also happens to have a 70 per cent chance of getting diabetes, he explained. “You may then have a different conversation with the patient than you would otherwise, now that you know that,” he said.
“It doesn’t mean they have diabetes, doesn’t mean they’re going to get diabetes, but their risk factors are much higher based on this sophisticated model. So you can have a different conversation about their diet or exercise or follow-up visits or those those sorts of things.”
The predictive alerts also offer a weapon against the opioid crisis, since “part of the behavior of people that have this addiction is they visit a lot of different EDs at a lot of different health systems, and it’s very hard to put the puzzle together,” Woodlock said.
Moreover with the HBI machine learning technology, “you would see that they would have a higher risk of opioid abuse let’s say, and it would look at ED visits as an example factor. So this is a kind of a probability scoring on an individual patient level,” he added.
“The AI is kind of put to use on an individual care basis to identify risks or to help with directing population health health activities.”
Date: April 17, 2019
Source: HealthcareIT