A lire sur: http://www.cio.com/article/742142/5_World_Problems_Too_Big_for_Big_Data
Despite advances in computing power, storage and analytics technology, some challenges are still too big for big data. Here's a look at five such problems -- and what it will take to solve them.
By John Brandon
, Thu, October 31, 2013
CIO
—
What if big data could do even more to help with world
problems? So far, companies such as IBM, Google and HP have taken on
immense challenges, from analyzing the number of cars that use a bridge
on a busy highway or calculating how many people see one small ad in a
Web browser. Google has even announced an ambitious project that will address human aging.
But several major world problems remain elusive. In some cases, the data is not available to analyze at all. In others, computers fast enough to process the data haven't been invented yet. Here are five problems worth tackling. Will any big data companies step up to the plate? We'll have to wait and see.
Analysis: Why Healthcare Providers Aren't Happy With EHR Systems
More: Is Healthcare IT Interoperability (Almost) Here?
What's missing? Access to the global data. "Health records are kept in a whole bunch of disparate systems, and providers don't have an incentive to share them," says Mike Miller, co-founder at chief scientist at Cloudant, a distributed database provider. "Even if we had all the data in one place, we'd still need to optimize it all with machine learning algorithms and real-time analytics. That's the piece we're still working on today to get it right."
The problem? The supercomputers required for that kind of processing will have to be 1,000 times faster than those in use today. There are millions and millions of neurotransmitters, all interconnected and processing "data" in the brain.
Related: How Genomic Research Could Improve Healthcare
Also: How Big Data Will Save Your Life
"This will require substantial developments away from conventional silicon chips onward to biological chips for molecular computing," says Oliver G. McGee, a former U.S. deputy assistant secretary of transportation for technology policy in the Clinton administration and a professor at Howard University. "Molecular computing has the vast potential of 750 times faster extensive data management than conventional silicon chip computing for rational-intuitive cognition mapping of our cranial-abdominal brains."
But several major world problems remain elusive. In some cases, the data is not available to analyze at all. In others, computers fast enough to process the data haven't been invented yet. Here are five problems worth tackling. Will any big data companies step up to the plate? We'll have to wait and see.
Similar to this Article
Health Records for the World: Medicine Where It's Needed Most
Most people have some semblance of an electronic health record (EHR), even if it's simply a notation about a recent health check-up. The tools and technology are in place to maintain a world health record repository, too. With a global database, pharmaceutical companies could develop the most-needed vaccines and medications — that is, the supply chain would be optimized for actual needs. More: Is Healthcare IT Interoperability (Almost) Here?
What's missing? Access to the global data. "Health records are kept in a whole bunch of disparate systems, and providers don't have an incentive to share them," says Mike Miller, co-founder at chief scientist at Cloudant, a distributed database provider. "Even if we had all the data in one place, we'd still need to optimize it all with machine learning algorithms and real-time analytics. That's the piece we're still working on today to get it right."
Human Brain Map: See How the Rest of the Body Works
A model of the human brain could help science immensely. Doctors could see how a tumor grows or which functions in the brain controls other organs in the body. New science initiatives such as Europe's Human Brain Project seek to build a brain simulator in the next 10 years.The problem? The supercomputers required for that kind of processing will have to be 1,000 times faster than those in use today. There are millions and millions of neurotransmitters, all interconnected and processing "data" in the brain.
Related: How Genomic Research Could Improve Healthcare
Also: How Big Data Will Save Your Life
"This will require substantial developments away from conventional silicon chips onward to biological chips for molecular computing," says Oliver G. McGee, a former U.S. deputy assistant secretary of transportation for technology policy in the Clinton administration and a professor at Howard University. "Molecular computing has the vast potential of 750 times faster extensive data management than conventional silicon chip computing for rational-intuitive cognition mapping of our cranial-abdominal brains."
Aucun commentaire:
Enregistrer un commentaire