Big data in a healthcare setting can help with records and accounting, but applied to genomic studies of patients, it can offer real medical benefits.
This combination of real health and bottom-line benefits is why the healthcare industry is now looking into technology solutions to make the most effective use of their data. One example is the University of Pittsburgh Medical Center.
A medical combination of 3.2 petabytes
Though its headquarters is in Pennsylvania, University of Pittsburgh Medical Center (UPMC) is a global health enterprise with assets of $10 billion. It encompasses a health insurance services division, over 20 hospitals, and 400 doctors’ offices. All the associated records on patients, administrative actions, and financial transactions involved generate over 3.2 petabytes of data.
To manage and make use of it, UPMC announced that it's investing $100 million dollars over the next five years in building an advanced data warehouse that would bring all aspects of data together for improved care and greater efficiency.
In UPMC’s press release, Dr. Steven Shapiro, chief medical and scientific officer, is quoted as saying that this technology upgrade is needed because in its current state, the exponential growth of data “is moving faster than our ability to transform that information into intelligence and improved decision-making at the point of care.”
Tailoring to each patient's unique story
Though the entire project will take five years, the two-year goal is to assimilate "data from more than 200 sources of information" within the organization itself, as well as external labs and pharmacies. The access to this data, and the advanced analytics tools, would allow both doctors and managers to use predictive modeling for better understanding and decisions.
Dr. Shapiro points out that in medicine, one size does not fit all: "Every patient is different; every patient has a unique story." But how does one access that unique story of the patient? One way is through comprehensive electronic medical records, which provide more information than the standard patient files found at hospitals.
That is why genomics are one of the four facets to be enhanced by the new system, as reported by Healthcare IT News. Lisa Khorey, vice president of enterprise systems and data management at UPMC, explains that a genetic scan can indicate that the standard treatment may not have the desired effect on a particular patient. Consequently, access to the full record allows the doctor to make a more informed and better decision.
Data is also central to the other three aspects: outcomes, data mining, and accounting. To assess effectiveness, one has to be able to measure outcomes for particular segments in comparison to "a national baseline." Data mining offers answers to the question "What patterns exist?" That is the basic foundation for analytics. For accounting, the goal is to get beyond where the money goes to identify which expenditures add value and which are wasteful.
Referring to the initiative in an address to the Detroit Economic Club on November 14, 2012, Oracle’s chairman, Jeff Henley, declared: "UPMC’s big data project is ambitious, but one with the potential to revolutionize our approach to healthcare by reducing over-diagnosis and unnecessary treatments, improving patient outcomes, and creating new and highly personalized-care pathways."
Will it truly revolutionize healthcare? Time (most specifically, that two-year goal) will tell.
User Rank: Exabyte Executive 11/30/2012 | 7:40:04 PM
Re: Customized Security Too? Applying the necessary security to instill a patients trust is the reponsibility of the healthcare facility. Security must ease the insecurity that exists today which patients expect healthcare to meet that trust barrier whether it is personalized security measures, firewalls, encription, etc al. Healthcare today is still challenged with lacking resources to provide data security patients expect.
User Rank: Blogger 11/27/2012 | 10:30:04 PM
Re: Customized Security Too? @smkinoshita True, but the fact is that there already is a great deal of personal data about patients floating about their doctor's and hospital's files (often in unsecured paper format, as well as digital formats). So this does not necessarily create a new problem but merely another possible outlet for an existing one.
User Rank: Blogger 11/27/2012 | 10:28:20 PM
Re: Generalization problem @Anna I'm certain that the data on patient outcomes will prove very beneficial to others. The problem in the status quo of the medical industry is that it tends the other way -- going with the most common treatment and diagnosis. From my own limited experience, I can tell you that I was ignored when in labor because the nurses judged my contraction to be too "mild" to be significant just a couple of ours before my baby was born -- in the elevator of the hospital. (Now that's a whole story.) But another time, I complained that what the doctor prescribed for me was causing severe stomach pains, something that does occure in about 5% of the population. Doctor tend to go by the majority, but that can be forcing all patients into a Procrustean bed that fails to address the individual's needs.
User Rank: Exabyte Executive 11/27/2012 | 9:48:04 PM
Customized Security Too? Something that occurs to me that custom-fit healthcare data must identify the indivduals that the data is about. That means that there better be some serious security incorporated into their plans. I know it's pretty obvious, especially for medical data, but in my experience security tends to be an after-thought when in some circumstances it needs to be incorporated into the plan so it neither interferes with the work nor leaves the sensitive data vulnerable.
User Rank: Exabyte Executive 11/27/2012 | 9:41:46 PM
Generalization problem The key to effective Big Data generation and utilization is the ability to make generalizations but what happens when generalizations open up the possibility of creating errors that can be fatal as in the health industry? What's the point of having data that cannot be harnessed for a wider public than just the individual?
User Rank: Blogger 11/27/2012 | 11:31:28 AM
Re: Some obstacles are on the way @legalcio Good point, as there is always new data, it is a moving target so to speak. I'd imagine that the new systems will take in the new data to integrate it automatically, in contrast to the legacy systems that will have to be transferred over to the new center.
User Rank: Blogger 11/27/2012 | 11:30:12 AM
Re: Some obstacles are on the way @Saul Yes, the initiative is significant because it can become a model for other health care systems. If it does, there may be a paradigm shift in the way such data is managed in the future.
User Rank: Exabyte Executive 11/27/2012 | 11:17:05 AM
Re: Some obstacles are on the way Five years might be a bit optimistic because the data will just keep growing. To that extent, big data is always a moving target, isn't it? Ultimately the payoff here will be more comprehensive patient data, mitigating the chances of medical mistakes. Non-profit hospitals without a wealthy parent company will have to struggle to keep up.
User Rank: Blogger 11/27/2012 | 10:28:01 AM
Re: Some obstacles are on the way It's great to show such ambition... All the other healthcare facilities will have such a good basis of success (and failures they can learn from) because of places like UPMC... I just wonder how much it will hurt their bank balance in the process.