AI is no silver bullet for e-healthcare interoperability woes

Published in

Iryo Network

5 min readMay 14, 2018

The latest digital craze seems to be artificial intelligence (again). With Google leading with it at its annual I/O event, artificial intelligence again promises to offload work to smart algorithms, speed up the production processes and generally improve our lives.

Digital health is all about algorithms

ZDNet writes that Google “hypothesized that these techniques would translate well to healthcare; specifically, deep learning approaches could incorporate the entire EHR, including free-text notes, to produce predictions for a wide range of clinical problems and outcomes that outperform state-of-the-art traditional predictive models. Our central insight was that rather than explicitly harmonizing EHR data, mapping it into a highly curated set of structured predictors variables and then feeding those variables into a statistical model, we could instead learn to simultaneously harmonize inputs and predict medical events through direct feature learning.”

Algorithms are the logical step from data economy where the amount of big data simply becomes too great for a person to analyse it and shape actionable results from it. Algorithms offer a machine approach to the data piles and promise to finally make sense of it.

Digital health is one of the fields that hope to benefit the most from algorithmization. Huge amounts of health data gathered in individual medical records can benefit researchers discovering and testing new drugs, healthcare professionals who can optimize treatments and patients getting better care.

Of course there is a but.

AI still needs clean data

According to Health IT Analytics report, “For artificial intelligence to flourish in the wild, however, developers must establish a firm foundation of trust in their algorithms’ accuracy, objectivity, and reliability. The industry is close to making the leap from pilot to practice, ensuring that AI in healthcare is transparent, appropriately regulated, and implemented in a meaningful manner will be one of the industry’s most pressing concerns.”

It is interesting to see artificial intelligence being used as a cure for a very old problem in digital healthcare — interoperability.

“The recommendations in the new report underline the importance of ONC’s efforts toward interoperable and standardized health data and AHRQ’s efforts to effectively use those data sets to improve the quality and safety of patient care. These efforts will improve capabilities to exchange and appropriately use high-quality health data — critical elements in powering AI efforts in health and healthcare,” is one of the conclusions of the Hype to Reality: How Artificial Intelligence (AI) Can Transform Health and Healthcare consultation facilitated by the US government.

Artificial intelligence still needs useful data to analyse it and make meaningful deductions from it. Rob Schneier writes: “…because the cost of saving all this data is so cheap, there’s no reason not to save as much as possible, and save it all forever. Figuring out what isn’t worth saving is hard. And because someday the companies might figure out how to turn the data into money, until recently there was absolutely no downside to saving everything.”

Data collection and analysis will be one of the cornerstones of artificial intelligence development says Financial Times as well. And clean data is very hard to come by. “In a perfect world, organizations would gather a vast amount of data, analyze it, and generate solutions to the problems they’re facing. The truth is, as most know, we do not live in a perfect world. Insights from big data often have to be derived in a short amount of time. The technology a business has on hand might not be advanced enough to process so much information quickly,” writes Data Economy.

A further challenge stems from ensuring that the data used to train a system is representative and leads to reliable answers. There is now widespread acknowledgment of the risks that come from biased data. Often, the problem stems from applying information collected for one purpose to a different problem, without making allowances for gaps in the dataset.

Iryo network and AI

Iryo network is enabling AI to be used by medical researchers to reach specific patients in the Iryo system. Because of the use of zero-knowledge data storage, the use of AI for researchers is essential. It enables them to reach specific patients and ask for their permission to analyze patient data in place (on patient’s device, without initial patient data being copied from the medical record).

The patient’s device will receive a silent notification which will wake up a background process to query the requested criteria i.e. female, 30–35 years old with diabetes. If a patient does not fall within the defined parameters, the silent notification disappears. It will do so without providing a report to the requester thereby keeping patient-users anonymous. If the patient meets the criteria, a notification would be shown on the patient’s device.

The notification would include the name of the research institution, the justification for the query requested i.e. the aim(s) of the research, and the number of tokens available as an incentive to allow query results to be sent back.

Iryo envisions three types of opt-out, anonymous requests that present various potential implications for privacy which would require distinct user consent. These types are identified as a pseudo anonymous query, an anonymous query used for AI validation across a dataset and an anonymous query to deliver patient value.

We have already written on the challenges of interoperability and the ways Iryo is solving that problem. Just to recap:

We are putting the patient in the center of our strategy to decentralize e-health medical records and letting the user have total control over them.
We are developing Iryo network with openEHR standards in mind, and by doing that we are bringing down data silos and making the data flow more persistent and sustainable.
From avoiding vendor lock-in to preventing data toxicity, the Iryo network is enabling medical data research while at the same time guarding your privacy at all times.
Privacy by default, the Iryo is implementing decentralized records paired with zero knowledge storage, using blockchain as a permission control tool. Not only is the data encrypted, the only control key resides in the hands of the user.
And finally, since Iryo network is using open-source standards that are recognized by the global e-health industry today, we are providing sustainable model for medical researchers, doctors and patients.

AI is no silver bullet for e-healthcare interoperability woes

Digital health is all about algorithms

AI still needs clean data

Iryo network and AI

Written by Domen Savič