Geisinger injects machine learning into clinical workflow to find health problems faster

Geisinger doctors and researchers have trained computers to read CT scans of patients’ heads to detect a life-threatening form of internal bleeding called intracranial hemorrhage.

Because early and accurate diagnosis is critical, leading hospitals are moving forward with precision medicine tactics built on artificial intelligence and machine learning technologies. Last week, for instance, Intermountain-owned Navican partnered with Philips to offer precision cancer care services. 

Last month, Vanderbilt University Medical Center said is researchers are analyzing genetic data to identify diseases within large populations then tailoring treatments for individuals and Mayo Clinic revealed that it is applying IBM Watson supercomputing capabilities to more effectively match patients with appropriate clinical trials. 

In Geisinger’s case, doctors noted that they incorporated machine learning, using computers to detect patterns in data, into clinical workflows — and the approach has enabled specialists to reduce the time to intracranial hemorrhage diagnosis by 96 percent.

“This is not about replacing doctors with machines,” Aalpen Patel, MD, chair of Geisinger System Radiology, said in a statement. “This is about the smart use of machine learning technology to aid medical providers in delivering better and faster care, especially in areas where time is critical.”

Geisinger has been able to combine radiographic and other medical imaging data to train computers to pinpoint the worst cases. By flagging the most urgent images for priority review by radiologists, they can provide earlier diagnosis and life-saving interventions.

Intracranial hemorrhage affects approximately 50,000 patients per year in the United States and, what’s more, 47 percent die within 30 days.

Officials said Geisinger is also applying machine learning in other areas, such as congenital heart disease.

Twitter: @Bernie_HITN
Email the writer: [email protected]

Source: Read Full Article