This tip is part of a series exploring big data in healthcare. Each story in the series will break down an aspect of analytics and where it fits into healthcare needs. This one explains how insight is lost in the process of converting information into healthcare data sets.
Many people today would agree that the healthcare industry has lots of data but not much information. I tend to disagree. Well over 70% of care-related information is not data but information. I would say that the health industry has lots of information but not much data.
The challenge we are faced with is how to synthesize such vast information sets to draw meaningful knowledge. To make this happen, we first extract interesting characteristics (attributes) as analyzable data sets. These healthcare
For example, images and genomics sequencers generate extremely large volumes of streaming content at terabyte scale. Tons of information is hidden deep in these monstrous content clusters.
The information is first converted to data structures that algorithms can digest to develop new drug therapies and personalize care. Present information synthesis techniques are still very data-oriented.
Similarly, social media is a rich source of healthcare information. Here too, we still have to first convert such rich content to structured data before running mining algorithms.
An interesting fact is that we first end up converting information to data sets, then use data to regenerate information -- and during such a transformation, we lose a lot of insight. Perhaps in the future, we will have smart algorithms that derive insights directly from such vast information sets without need to convert them to data.
Go on to the next tip in the series.
Naeem Hashmi is chief research officer at Information Frameworks, as well as an expert in healthcare data analytics, information management and data exchange. He is an active member of HIMSS, CHIME and AMIA. Let us know what you think about the story; email firstname.lastname@example.org or contact @SearchHealthITon Twitter.
This was first published in April 2013