Harvard Medical School - Flickr

Where are artificial intelligence and deep learning going to be applied next? Beth Israel Deaconess Medical Center and Harvard Medical School believe it is breast cancer pathology. According to FierceBiotech, researchers at Beth Israel Deaconess Medical Center and Harvard Medical School have worked together to create an analysis of breast cancer pathology that incorporates artificial intelligence. They found that their system–and the evaluation by pathologists themselves–worked better when used in conjunction than either did alone.

In an evaluation of slides of lymph node cells, the automated diagnostic method was accurate about 92 percent of the time. This was almost as accurate as human pathologists–who are about 96 percent correct. When combined, the results were even better.

“The truly exciting thing was when we combined the pathologist’s analysis with our automated computational diagnostic method, the result improved to 99.5 percent accuracy,” said pathologist Dr. Andrew Beck, director of bioinformatics at the Cancer Research Institute at Beth Israel Deaconess Medical Center and associate professor at Harvard Medical School. “Combining these two methods yielded a major reduction in errors.”

“Our AI method is based on deep learning, a machine-learning algorithm used for a range of applications including speech recognition and image recognition,” said Beck. “This approach teaches machines to interpret the complex patterns and structure observed in real-life data by building multi-layer artificial neural networks, in a process which is thought to show similarities with the learning process that occurs in layers of neurons in the brain’s neocortex.”

Increasingly, the industry expectation is that deep learning can be useful to aid human analysis. This is a first step to pre-analyze and learn from massive quantities of data to offer insights that are subsequently reviewed by people.

Breast cancer is not the only new area where artificial intelligence and deep learning are being applied. MedyMatch Technology Ltd. and Capital Health are focusing on strokes. MedyMatch Technology, the data analytics healthcare company focused on providing physicians with artificial intelligence and real-time decision support tools, recently announced a partnership with Capital Health. This is the first of several partnerships with hospitals in the United States intended to improve stroke patient outcomes. As part of the agreement, Capital Health will provide anonymized data from its two-hospital health system to MedyMatch for use in the development of its first decision support tool directed towards stroke patients.

“MedyMatch is excited to be working with a leading institution like Capital Health which offers advanced radiology services that help support the critical diagnostics needed for its comprehensive stroke center. True innovation in healthcare only comes from the close collaboration between industry and world-class clinicians,” said Gene Saragnese, Chairman and CEO of MedyMatch Technology. “Our partnership with Capital Health brings together breakthrough technology in deep learning and clinical expertise that will accelerate the transformation we all desire in healthcare–better care at lower cost…The data Capital Health will provide will allow us to move closer to providing this decision support tool which can help ensure appropriate diagnosis critical for treatment.”

“MedyMatch is going to make clinical decision support in radiology faster and more accurate,” said Dr. Ajay Choudhri, Director, Vascular and Interventional Radiology and Assistant Director, Radiology, Capital Health, “MedyMatch’s AI capability augments physicians’ reading ability and provides a second set of eyes on the patient’s imaging study. In the area of stroke where time equals brain, it is critical to get a fast, spot on diagnosis.”

Additionally, MedyMatch will leverage medical imaging libraries across multiple imaging modalities including CT, X-ray, MRI, Ultrasound and PET, which will be utilized as part of its research and development efforts to train its next set of applications and deep learning algorithms. “The right data is at the core of application innovation,” said Robert Mehler, co-Founder and Chief Operating Officer.

Artificial intelligence and deep learning have a lot of potential especially with the wealth of data that is now available to researchers and clinicians. Diagnosing strokes and breast cancer is a great starting point to turn this potential into practical value.


Ryan Lahti is the founder and managing principal of OrgLeader, LLC. Stay up to date on Ryan’s STEM-based organization tweets here: @ryanlahti

(Photo: Harvard Medical School, Flickr)