To meet the growing demand for greater transparency, simplicity, and value, Xerox has chosen SyTrue’s Clinical Natural Language Processing (NLP) and medical terminology platform in the development of Midas+ Advanced Analytics capabilities at Xerox. The combined solution will leverage NLP and interoperability solutions to transform clinical documentation and other types of data in the electronic medical record (EMR) into standard diagnosis and procedure codes in real-time; which may then be used to calculate risk and expected outcomes of care for patients while they are in the hospital.
In the past, risk adjusted outcomes using medical claims data were calculated for patients, days and often weeks following discharge from the hospital. Expected values for mortality, 30-day readmission, hospital acquired complications or length of stay were generated in hind-site. While useful to hospitals for benchmarking overall performance, this retrospective information provided limited value to proactively direct care management decisions for individual patients prior to discharge.
Next Generation Case Management
The new capabilities of concurrent coding open up the door to “next generation” case management and care transitions planning. “Not only can we understand which patients are at greatest risk for hospital complications, extended lengths of stay or unplanned hospital readmissions, we can prescriptively match the right discharge resources to the right patients to create the most optimal outcome following discharge”, says Jim Kirkendall, Vice President of Advanced Analytics at Midas+ Xerox.
Additionally, concurrent coding along with other information in the EMR such as medications, lab results or vital signs can be leveraged to create other advanced analytics, such as life-saving alerts to clinicians when a patient’s condition is changing. For example, the Midas+ ICU Transfer Predictive Analytic sends alerts to clinicians up to four hours before a medical or surgical patient’s condition changes that could require unplanned transfer into a critical care unit. In this way clinicians have time to intervene and may prevent a costly admission into a critical care unit. “These types of innovations are not only essential to hospitals who are facing alternative reimbursement models such as bundled care payments, they are essential to the patients they serve”, says Kirkendall.
Midas+ Machine Learning Capabilities
The SyTrue medical terminology engine supports these capabilities by creating standardized codes for medications (RxNorm codes) and laboratory tests (LOINC codes) from virtually any EMR system so that they can be consumed by machine learning algorithms developed by Midas+. Machine learning capabilities result in clinical alerts that are highly reliable and accurate as compared to older generation algorithms and rules based on clinical logic alone. In this way, clinical surveillance alerts for emerging complications such as renal failure or sepsis can become part of the clinician’s trusted clinical decision support system.
Kirkendall sees major advances in using the SyTrue platform. “Not only is the ability to access clinical conditions and procedures without having to rely on historical claims data a giant leap forward for us, but it solves another challenge. We also needed to know the patient story across multiple facilities – what care was provided to the patient in one hospital versus another hospital. We now have that answer and far greater insights into care coordination across multiple points of service and over time; which is where population health management is heading,” said Kirkendall.
“Our vision is to bring actual clinical intelligence for patient care, which is relevant and usable for the care provider, in that moment of care. Care providers in turn will see this as improved case management solution allowing for better patient care and therefore improved outcomes – this is our end goal,” said Kyle Silvestro, Founder and CEO of SyTrue, Inc.
