By Dr. Michael Weiner
It’s been 53 years since IBM created the first electronic health record (EHR) for Akron Children’s Hospital, built on IBM’s Ramac 305. In those days, we could never have dreamed of the advances that would soon unfold for the modern EHR. From the amount of data they collect and store to the ability to access them remotely on mobile devices, EHRs have truly transformed medicine over the past few decades.
EHRs have also begun to transform our healthcare ecosystem. As a physician, I can attest to the value of an EHR to help improve the quality of care we deliver to our patients.
EHR’s can also facilitate care coordination between clinicians and help achieve greater administrative efficiencies.But as we look to the future of EHRs and to the requirements of Stage 3 meaningful use in the U.S., we continue to ask ourselves how to integrate structured and unstructured clinical data. Many of us have often wondered: When will the technology be able to read our notes?
It turns out, they can now. IBM’s natural language processing technology – the same software that’s at the foundation IBM Watson – can turn doctors’ notes into EHR insights that will ultimately help improve patient care. With the technology, doctors can upload their text notes to be translated into discreet data fields that can be integrated into today’s EHRs.
To continue to evolve the meaningful insights that are gained from the EHR, it is important that patient records include a complete and accurate snapshot of each patient. As we have often suspected, there is a great deal of rich and meaningful information within doctors’ notes, which contains some of the most valuable information on a patient’s condition.
From symptoms and medications to lifestyle and family issues that can affect health, the “unstructured data” in doctors’ notes contains some of the most critical information that can enable clinicians, hospitals and health systems to better serve their patients.
Take for example Carilion Clinic, a hospital system in Virginia. Carilion Clinic wanted to find out which of their patients might be at risk for congestive heart failure – a disease that is notoriously difficult to predict prior to its onset. Heart failure currently afflicts more than 5 million U.S. adults, half of whom will not survive five years after diagnosis, according to the CDC.
By applying IBM’s natural language processing technology and predictive analytics to their electronic medical records, Carilion detected 8,500 patients at risk for congestive heart failure. These 8,500 patients could benefit from preventive care to slow or even stop the onset of congestive heart failure.
The project underscores the value of the data from doctors’ notes. By adding natural language processing to their robust analytic capabilities, Carilion’s data set included important “unstructured” data. As a result, Carilion was able to identify more patients at risk for heart failure – and with an impressive 85 percent accuracy rate – than they could have without the inclusion of doctors’ notes.
Tools like this can help us recognize and understand the unstructured data in health care and enable a truly holistic and complete view of the patient. We need the ability to draw predictive insights from patient records so we can see who may be at risk for a medical crisis and could benefit from preventive care.
The future of electronic health records and the continued promise they hold will continue to evolve as we learn how to gain greater insights from the data. In today’s digital world, the best care must be based on the best information.