New York-based Mount Sinai Hospital has begun a project which puts it in the vanguard of predictive analytics, working with a partner focused on artificial intelligence. Mount Sinai plans to use the Cloud Medx Clinical AI Platform to predict which patients might develop congestive heart failure and care better for those who’ve already done so.
As many readers will know, CHF is a dangerous chronic condition, but it can be managed with drugs, proper diet and exercise, plus measurement of blood pressure and respiratory function by remote monitoring devices. And of course, hospitals can mine their EMR for other clinical clues, as well as rifling through data from implantable medical devices or health tracking bands or smartwatches, to see if a patient’s condition is going south.
But using AI can give a hospital a more in-depth look at patterns that might not be visible to the unaided clinician. In fact, CloudMedx is already helping Sacramento-based Sutter Physician Services improve its patient care by digging out unseen patterns in patient data.
To perform its calculations, CloudMedx runs massive databases on public clouds such as Amazon Web Services and Microsoft Azure, then layers its specialized analytics and algorithms on top of the data, allowing physicians or researchers to query the database. The analytics tools use natural language processing and machine learning to track patients over time and derive real-time clinical insights.
In this case, the query tools let clinicians determine which patients are at risk of developing CHF or seeing their CHF status deteriorate. Factors the system evaluates include medical notes, a patient’s family history, demographics and past medical procedures, which are rolled up into a patient risk score.
In moving ahead with this strategy, Mount Sinai is rolling out what is likely to be a common strategy in the future. Going forward, expect to see other providers engage the growing number of AI-based healthcare analytics vendors, many of whom seem to have significant momentum.
For example, there’s Lumiata, a developer of AI-based productive health analytics whose Risk Matrix tool draws on more than 175 million patient-record years. Risk Matrix offers real-time predictions for 20 chronic conditions, including CHF, chronic kidney disease and diabetes.
Risk Matrix bases its predictions on its customers’ datasets, including labs, EHR data claims information and other types of data organized using FHIR. Once data is mapped out into FHIR, Risk Matrix generates output for more than 1 million records in less than three hours, the company reports. Users access Risk Matrix analyses using a FHIR-compatible API, which in turn allows for the results to be integrated into the output of the existing workflows.
While many startups have flocked into the imaging and diagnostics space, expect to see AI-related activity in drug discovery, remote monitoring and oncology. Also, market watchers say companies founded to do AI work outside of healthcare see many opportunities there as well.
Now, at least at this stage, high-end AI tools are likely to be beyond the budget of mid-sized to small community hospitals. Nonetheless, they’re likely to be deployed far more often as value-based reimbursement hits the scene, so they might end up in use at your hospital after all.