For more than a decade, big data has been taking the business world by storm, with the growth of healthcare data as the most astounding. It is estimated that each patient generates close to 80 megabytes of data every year in imaging and EMR data. Adding the amount of data that is now being generated by sensors and wearables, Google estimates that we could see 2,300 exabytes of global healthcare data by the end of 2021. As a comparison, one exabyte of data could hold 300 times all the contents of the Library of Congress.
Harnessed effectively, the explosive growth in healthcare data holds the promise of spurring groundbreaking innovations in areas such as drug discovery, precision medicine, population health, administrative automation and cost/revenue optimization. We have already seen a glimpse of that promise and are poised for many more exciting breakthroughs. But unless we find better ways to access more complete data without jeopardizing individual privacy, the boundaries of innovation will be limited.
In the face of privacy issues, here are four of the big data challenges corporations are wrestling with when building healthcare data strategies:
1 – Access to diverse data both internally and externally- A broad data ecosystem offers tremendous value, but legal and operational challenges significantly hamper organizations’ progress. For example, companies involved in RWE (Real World Evidence) have to ingest vast amounts of external data which means much of the incoming data are anonymized, detracting a great deal of its core value. This risks the loss of critical discoveries and creates suboptimal results relative to the investments made to capture this data.
2 – Data accuracy and adequacy- To measure data adequacy is to determine if the data can meet its intended purposes. Here again privacy regulations require masking important information—such as full five-digit zip codes or exact age—compromising both the accuracy and adequacy of the data. Imagine having to run sophisticated Social Determinant of Health (SDoH) analytics without access to a full zip code – the resulting models would be less specific, and therefore less valuable to the intended use.
3 – Drawing insight from the data- Many techniques have been developed to draw insight from big data, such as visualization, predictive analytics and real time trend analysis. Context, relevance, specificity and clarity are the key attributes of actionable insight. Due to privacy regulations, such analytics tools today deal with limited data and as a result can generate information but not enough insight.
4 – Workflow integration- This is the process that connects disparate business systems, allowing the smooth transfer of data or analytics between various platforms. Workflows bridge the gaps between applications, and allow for processes to be automated. Imagine a world where data can be shared in the workflow in a “trustless” environment, all while preserving individual privacy. This may sound like an ambitious goal, but we now have the technologies to make it possible.
To date, the burden of privacy laws has materially hampered the full promise of data analytics. But with new privacy preserving technologies, we are entering a golden era in data sharing and data exchange. One such technology ready for prime time is Differential Privacy (DP), which previously roamed the world of academia for 18 years. DP is one of the most mathematically rigorous cryptographic technologies for privacy preserving data use- able to expose the entirety of data content to data analysts for statistical inquiry whilst shielding the identity of individuals. And most importantly, this technique meets or exceeds the stringent requirements of HIPAA, GDPR and other privacy regulations.
If digital transformation through big data use is a high imperative for your corporation, it’s time to look at your data strategy through a whole new lens. We can move from a “we can’t” mindset to a “what’s possible” outlook. Let’s map the future of analytics with these exciting new privacy preserving technologies in mind.