By: Mila Orlovsky, Clinical data scientist, Zebra Medical Vision
“Big Data is like teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it…”
Dan Ariely published this phrase back in 2013, when Big Data was the main buzzword around. New technologies and applications had the ability to deal with massive amounts of data in a fast, scalable way, mainly implemented in retail and finance where huge amounts of data were accumulating, and immediate business impact was achievable. However, for various reasons, almost no attempts were made to address problems in the healthcare industry.
In fact, in 2007 when I started my career in a public hospital, it was not a popular decision among my peers. I was frequently asked “what innovation can you possibly do there”? But we sure did, we used statistical and analytical tools to correlate clinical processes with outcomes or to near-real-time automatically identify incidences of acute situations; however, the medical community was not yet ready to receive and wasn’t mature enough to accept these new developments. Initial projects were a particular challenge in helping physicians understand the benefits of automatic tools working alongside them.
Fast forward, April 2019 BigDataTLV in Healthcare convention, I’m presenting two projects completed here at Zebra Medical Vision. In the projects I presented (and were a mutual team effort!) we developed a Big Data tool – a pipeline that was able to run in parallel hundreds of instances to produce large scale inference for algorithms that identify Pneumothorax on chest X-ray and calculate emulated T-score for bone density estimation on spine CT. Thus, we were able to estimate algorithms yield and expected behavior in a real-life setting in a matter of several hours. We used Big Data methodology to support AI application in medical imaging.
Observing the hall around me, I was overwhelmed by how many companies eager to turn endless complex clinical data, accumulated by years of usage of EMR (Electronic Medical Record), into meaningful and helpful online applications and algorithms. The potential is indisputable, and the healthcare community is coming forward with developing and accepting new technologies. Almost every healthcare provider or digital health startup representative I interacted with had a potential solution to a clinical problem, going from pathology to oncology, from triage management to hospital acquired infections. Some companies presented innovative infrastructures that strive to deal with the complexities of the mostly unstructured and sparse nature of medical records.
I was happy to see that many familiar ideas turned into products, but above all, I was excited to hear top Israeli and foreign HMO leaders talk about investment and policy establishment towards daily implementation of smart tools to interact with clinical staff. Yes, it is a winding road, yes there are many challenges ahead, but AI solutions and Big Data tools are a huge leap forward in healthcare, not to replace the domain experts but to maximize the benefit of composite effort.
Today, everyone is just doing it, Dan 🙂