zentilia - Fotolia
It's become commonplace for large software vendors to make sweeping new product announcements during their yearly events. So, it was expected that Microsoft might drop one at its annual Research Faculty Summit. Despite the inevitability of the reveal of its new artificial intelligence project called Project Adam, the technology could still have a few surprises in store in the form of healthcare applications.
During a keynote speech at the event, Microsoft introduced its flavor of AI, displaying some of its visual recognition capabilities. The company has developed its deep-learning platform by using massive amounts of data and computing power to perform visual recognition tasks.
In a demonstration during the keynote, the tool was able to identify a dog's breed based on a photo. Microsoft is looking to expand Adam's capabilities beyond dogs and move into identifying and measuring the content of our meals.
Project Adam holds the potential to cross over into healthcare. There are two potential specialties in which this this image processing tool can be used. The first is dermatology, where patients may be able to photograph and submit images of skin conditions from their mobile phones. They could then get a response from Microsoft or a third-party provider that's hooked up to Project Adam and receive a diagnosis on their skin condition.
Another area where this platform can provide assistance to healthcare professionals is in the radiology space. High-quality images, experience and an eye for detecting abnormalities can mean the difference between life and death in this field. Medical imaging may one day be able to use a platform like Microsoft's for radiological assistance in pre-processing or pre-diagnostics. It could also be used during post-diagnostic review to ensure nothing was missed.
In the above use cases as well as some others, Microsoft or the platform users must be able to provide the system with a huge amount of data to get a meaningful analysis in return. The good news for radiology is that there are imaging standards that will make it easy for a system such as Adam to consume large quantities of similar data. By using DICOM, Health Level International 7 or other commonly used formats, health organizations or research groups can enable their systems to learn from existing cases -- knowledge that can be used to detect similar abnormalities in future images.
Dermatology may present some additional challenges because dermatologists have just recently begun capturing photos of patients' skin and storing them in an EHR system. Unfortunately, based on some of my interactions with dermatologists in Charlotte, N.C., it's clear that there will be interoperability and sharing issues for dermatologists using different systems. For example, a dermatologist using an eClinicalWorks system to scan images into their notes will operate differently than another dermatologist that is on McKesson's Practice Partners attaching images from digital cameras because there does not seem to be a common practice on how to store images and data.
Microsoft is taking a chance on this technology, and it might just be what the healthcare market needs. While it is not likely to replace radiologists and other imaging-focused specialties, it may play a role in assisting and supporting diagnoses. In a market where Google and IBM have already introduced AI and other learning systems, Microsoft is now penetrating a space that could be a game-changer for them. Whether it's aiding the diagnosis process through medical imaging recognition, or simply helping consumers identify insects during a walk in central park, Microsoft is likely going to have suggestions for these questions on its cloud.
About the author:
Reda Chouffani is vice president of development at Biz Technology Solutions Inc., which provides software design, development and deployment services for the healthcare industry. Let us know what you think about the story; email firstname.lastname@example.org or contact @SearchHealthIT on Twitter.
Radiology could benefit from peer review process
Less marketing, more data mining will drive radiology improvement
Business intelligence a wise strategy for radiologists