by Brian Witte, Ph.D.
The Big Data revolution is coming to the blood banking community. Other areas of health care have gradually embraced the transformative power of large-scale data collection and analysis. The impact of this change is difficult to predict, but will almost certainly include reductions in wasted units and improvements in distribution.
The Data Revolution, like the Computer Revolution that enabled it, will transform industries, alter society, and change the world. Also like the Computer Revolution, the Data Revolution will not be evenly distributed. Endeavors that can leverage data-driven insights will be able to realize dramatic improvements in efficiency and productivity. Better still, huge data sets encompassing even seemingly innocuous information can lead to remarkable new insights into how the world works.
Data and personal genomics
Health technology is one of the sectors where the benefits of Big Data have yet to be fully realized, or even extensively explored. On the one hand, personal genomics is taking full advantage of modern data collection and analysis tools. In 1990, sequencing a single human genome required 10 years and $3 billion. The cost today is under $1000 and requires a few days to complete.
The exponential decrease in the cost of sequencing human genomes outstrips even Moore’s Law. (see photo) The cost is so low now that researchers can compare genomes of cells taken from healthy and malignant tissue in individual cancer patients. With dozens of possible mutations that can contribute to cancer, each with its own unique effect on cell physiology, traditional treatments were limited to the chemical equivalent of carpet bombing: destroy all rapidly dividing cells. The impact on the patient is often horrific. Epithelial cells lining the digestive tract that naturally have a high turnover rate were affected along with malignant cells. Personal genomics is still in the exploratory phase, but the promise is immense.
Data and health policy
Big Data is coming to other areas of health care, too. Exactly how the revolution plays out remains to be seen, but the huge scale of the health industry and the vast amounts of resources devoted to health care ensure such power tools will be employed.
Health policy is poised to be the next major beneficiary. As David Lazer and Ryan Kennedy, academic researchers in public policy and computer science, wrote in WIRED Magazine:
[R]esearch has demonstrated the value of big data in modeling disease spread, real time identification of emergencies, and identifying macro economic changes ahead of traditional methods.
The full article is devoted to a post-mortem of Google Flu Trends (GFT) and a look ahead at the future of public-private collaborations in merging big data and health policy. While they laud the achievements of GFT for displaying some of the immense promise of Big Data in epidemiology, they ultimately lay the blame for its demise on the restricted data available to researchers outside Google. In short, a model can only be as good as the data used to build that model.
Data and blood products
In a recent editorial for Transfusion, Dr. Harvey Klein brings these concerns home to the field of transfusion medicine. His essay accompanied the release of two complimentary reports on the statistics of blood product use in the U.S. The reports, while insightful, rely on data collected from voluntary participation. As Dr. Klein relates, the costs of such reporting can be significant:
At the NIH Clinical Center, which collects, manufactures, and transfuses blood, a skilled, experienced supervisor spent the better part of three days completing the AABB Blood Survey only to find that HHS’s NBCUS asked for additional details, requiring further database queries and manual compilation. The two reports combined took an estimated week to complete.
The effect of such a high burden on reporting institutions is reflected in the response rate of 55% of AABB member hospitals and 33% of hospitals surveyed by the Department of Health and Human Services for the National Blood Collection and Utilization Survey (NBCUS).
The result of such fragmentary reporting is a data set that makes prediction difficult or impossible. And yet, as Dr. Klein notes, “…blood, like clean water and essential energy reserves, is a national strategic resource [and] data about blood supplies and uses are vital for disaster- and emergency-preparedness planning…”. The data, in other words, is invaluable.
With accurate data, we could make meaningful inroads into the seasonal shortages in blood products. With real-time data, we could assess the impact of advertising campaigns targeted to combat downturns in donations – or even to forestall downturns forecast with a model primed with high-quality data.
The obstacle preventing better data collection is simple logistics. It’s an additional level of record keeping on top of the already complex and detailed requirements. One solution, as Dr. Klein suggests, is to increase the cost of blood product transactions, perhaps with federal oversight, and to fund data collection with the proceeds.
Another approach is to capture data on blood products use as an automatic and invisible consequence of normal hospital and blood center activity. Bloodbuy is building just such a system, and we would welcome participation in the national discussion around blood product data collection. The Big Data revolution is growing and it will take all of us working together to build a secure future.
Doyle, A. C. (n.d.). The Adventure of the Copper Beeches. Retrieved November 07, 2016, from https://en.wikisource.org/wiki/The_Adventure_of_the_Copper_Beeches
Ellsworth RE, Decewicz DJ, Shriver CD, Ellsworth DL. Breast Cancer in the Personal Genomics Era. Current Genomics. 2010;11(3):146-161. doi:10.2174/138920210791110951.
Klein, H. G. (2016), Blood collection and use in the United States: you can't manage what you can't measure. EDITORIAL, 56: 2157–2159. doi:10.1111/trf.13724
Lazer, D., & Kennedy, R. (2015, October 15). What We Can Learn From the Epic Failure of Google Flu Trends. Retrieved November 07, 2016, from https://www.wired.com/2015/10/can-learn-epic-failure-google-flu-trends/
Personal genomics. (n.d.). Retrieved November 07, 2016, from https://en.wikipedia.org/wiki/Personal_genomics