Big data dominates as deep learning, AI seal the deal

Tools

Deep learning. Artificial intelligence. Machine learning. Cognitive computing. There are so many ways to say that your computer is smarter than you. But it's just taking a little time for it to figure that out. As wearables, apps, devices, genomic testing and imaging tech proliferate, all that data doesn't just sort through itself. A variety of computer-enhanced analytics are starting to emerge to make sense of all that Big Data. And while mostly humans are sorting through and making the analytical rules for now, soon that will no longer be the case.

Radiologists are likely to see some of the first applications of deep learning--in an effort to better standardize what now tends to mix science with a bit of art drawn from experience. But if the industry is working to get rid of that human touch, machines and their analytics need to get a lot smarter. Initial attempts to inject software into analyses of medical images have had mixed results so far.

But the trick might be to get the machines to start thinking for themselves, using analyses and algorithms drawn from the data. As artificial intelligence makes rapid leaps outside the field, healthcare stands to benefit. And the hope is that all this automatic analysis will translate into better, more efficient and cheaper patient care.

IBM Watson Health's Kendall Square, Cambridge location

IBM's Watson, which is used to review massive stores of unstructured data to determine a pattern and establish a path for analysis, is a prominent practitioner all over healthcare. In 2015, the artificial intelligence of Watson was variously featured in a diabetes partnership with Medtronic ($MDT), Johnson & Johnson ($JNJ) and Apple ($AAPL), cancer genomics deal with dozens of hospitals and a rare pediatric diagnostics partnership with Boston Children's Hospital.

Most prominently, IBM ($IBM) bought medical image management player Merge for $1 billion--indicating a major commitment to using the lessons derived from machine learning to medical images. Expect to see some big moves next year in image analysis as this application starts to achieve some initial results. About 90% of medical data is comprised of images, according to IBM. That offers a lot of fodder for potential improvements to the diagnostics based upon them.

But IBM is hardly alone. Imaging giant GE Healthcare ($GE) grabbed a lot of attention with a recent partnership with startup Arterys just as it nabbed a $7 million Series A; it promises to apply machine and deep learning to imaging.

As part of the deal, GE Healthcare has developed a new product, ViosWorks. It's a cardiac solution that reduces magnetic resonance imaging (MRI) assessment to a fraction of the time of conventional cardiac scans. It's expected to offer 3-D cardiac anatomy, function and flow in one free-breathing, 10-minute scan. Prior to this system, cloud-based, real-time image processing at these resolutions wasn't feasible, the partners said.

"Medical images are some of the most complicated data sets imaginable, and there is perhaps no more important area in which researchers can apply machine learning and cognitive computing. That's the real promise of cognitive computing and its artificial intelligence components--helping to make us healthier and to improve the quality of our lives," summed up IBM SVP John Kelly when the computer giant said it would acquire Merge. -- Stacy Lawrence

For more: 
Dozens of hospitals to deploy IBM Watson computer in fight against cancer
IBM brings Watson to Boston Children's Hospital for rare pediatric disease diagnosis
IBM's Watson deepens engagement with Apple, adds new pharma and hospital customers
IBM to add image management to Watson via $1B buy of Merge
IBM, Novo Nordisk team to seek new diabetes treatments through the cloud
IBM partners with Medtronic, J&J, Apple to use Big Data to optimize healthcare coaching and diabetes care
Deep learning radiology startup Enlitic wins Aussie partner, $10M Series B
Startup gets $7M Series A, GE partnership to apply machine learning to medical imaging