It seems that more and more the medical community is turning to less than obvious sources in order to increase the accuracy and reliability of diagnosis. From cancer sniffing dogs to Pigeons deciphering MRI images, caregivers are dedicated to finding even the most bizarre ways to increase early diagnosis rates. Well much like almost every other thing in our lives, it seems the machines are taking over. Zebra Medical Vision, an outfit out of Tel Aviv dedicated to creating radiology based diagnosis algorithms, has some new mechanisms to predict dangerous medical conditions to help caregivers understand which of their patients are at risk.
The company, headed by CEO Elad Benjamin, claims to have created two algorithms that can work separately but can also be applied together to medical scans and images. According to the company, the first algorithm is dedicated to detecting plaque build up in the cardiovascular system. This technology could help identify a wide variety of cardiovascular conditions, and understand those at risk for a host of diseases. The second algorithm is supposed to help detect various metabolic conditions, specifically in the liver. It’s mostly dedicated to catching Fatty Liver and NASH, both major liver diseases.
Zebra claims that while each of these algorithms does it’s job well on its own, together they can be used to pre-detect major cardiovascular events and help caregivers prescribe preventative courses of treatment and even change patient’s lifestyles in order to avoid the conditions they are at risk to contract.
While all this sounds great on paper, outside of the obvious “we’re making the machines too smart, and they will rise against us” risk, there are other hazards to using machine learning based diagnosis tools. Accountability is one. If an algorithm produces a false negative, or a false positive for that matter, who’s to blame? The doctor that ran the algorithm? Maybe it’s the programmer who wrote it? Other risks include how easy it is to hack since these images are diagnosed by a machine learning algorithm, it’s safe to say they will have to be connected to the cloud, this could lead to confidential medical information leaking to the internet.
But, if these algorithms can truly do what the team at Zebra claims they can do, these obstacles could be dwarfed by the years of life that will be gained by patients who will be diagnosed with these tools. “This is yet another step in our mission to help provide faster, more accurate radiology services at lower cost, by teaching software to read and identify key clinical conditions in imaging,” said Elad Benjamin, Zebra’s CEO. “We believe that these tools, as well as new algorithms which we continuously release, will help Radiologists deal with the continuous pressure they face to increase output and maintain a high quality of care.”
Zebra isn’t done yet, as they claim to be on the way to “create one hundred insights over the next three years.” Exciting times we’re living in.