Earlier this week, Cerner announced a multi-faceted, strategic collaboration with Amazon Web Services, Inc. (AWS), an Amazon.com company, to accelerate health care innovation across the globe. As part of this agreement, Cerner is naming AWS its preferred cloud provider.
It was only a matter of time before someone decided to merge electronic health records with a cloud-based service, since those services allow for immediate and substantial upscaling of the EHR platform. It’s an added bonus that Amazon includes machine learning applications in all AWS-based services. Machine learning algorithms applied to EHRs are supposed to be the “next big thing”, reducing medical misdiagnosis and identifying the likelihood of obscure, underlying conditions. It could extrapolate data taken from an entire platform to identify population health patterns. One success story that ML engineers cite is in the application of mental health: in academic studies, algorithms identified someone as battling depression better than the care providers could (they say you can lie on a form or to a person, but you can’t hide from your data).
We’ve been following how Epic has been developing their own in-house machine learning algorithms as well, but they have been hesitant to use a third-party system like AWS because they don’t want the data in their system being given to Amazon (or anyone else, really; they didn’t even initially participate in an AMA-led collaborative pilot that aimed to keep the patient data within the EHR companies). This isn’t an unreasonable concern, since patient health data IS their business, and it’s quickly becoming Amazon’s as well. Amazon already knows everything we browse and buy (and after buying Whole Foods: what we eat). Now they may have access to our health data as well (well, in any hospital that uses Cerner, anyways). They haven’t abused the potential synchronicity of all their various ventures yet, but there is that opportunity.
Returning to the Epic example, the only thing really stopping Amazon from using our medical records to suggest what groceries we should buy has been regulatory. Epic hasn’t been loudly touting their artificial intelligence capabilities because it’s still a grey area to use (ostensibly anonymized) patient data to feed a ML data analysis algorithm for anything more than proof-of-concept studies. While the potential benefits of using our health data to feed a machine-learning application are enormous, we still need to make sure this most personal of data is sufficiently protected. Amazon, while not as bad as Facebook et. al, hasn’t exactly been a paragon of privacy, so it’ll be interesting if/how they overcome this barrier that other health record systems have been running into.
Cerner is marketing a future of “enhanced clinical experiences, increased efficiencies… and other cutting-edge innovations thereby advancing better patient health outcomes”. If this partnership succeeds in bringing about the innovation it promises, collaboration between tech companies and healthcare organizations will become the norm. When historical data can accurately predict (or even drive) future needs, hospitals will need to be able to quickly adapt, both in space and in staffing. Better data can mean less unproductive floor area, less time spent waiting (and less space allocated for such), and fewer duplicate uses. Most importantly, if healthcare organizations want to maintain the integrity and trust they have built within their communities (in order to realize the potential these partnerships offer) they will need to make sure that their patients’ personally-identifiable health data is secure.
To join our discussion on healthcare design innovation, contact me here.