Cloud-based radiology service company Imaging Advantage said today it launched a machine learning research initiative with faculty members from the Massachusetts Institute of Technology, Harvard Medical School and Massachusetts General Hospital.
The project, titled Singularity Healthcare, seeks to develop an artificial intelligence engine to pre-read digital x-rays and identify potential injuries or diseases while continuously learning from Imaging Advantage’s 7-billion image database, the company said.
“Inconsistency in testing and access to care contribute significantly to $1 trillion of waste in the $2.8 trillion U.S. healthcare industry. If successful, Singularity will introduce a solution with potential to transform radiology by providing faster, more accurate and less expensive diagnostic testing, representing an indispensable innovation for radiologists,” Imaging Advantage prez Brian Hall said in a press release.
The Phoenix, Ariz.-based company said the program will seek to improve X-ray management, saying that X-ray exams constitute 50% of all radiology tests in healthcare and serves as a “limiting factor” in hospital emergency department patient flow and treatment.
“We have a number of opportunities for research and innovation at MIT, but were particularly intrigued by the bold initiative proposed by Imaging Advantage. Given IA’s platform approach to healthcare delivery, national scale and significant imaging data set, and the contribution of Dr. Saini from MGH, one of the leading global radiology teaching and research institutions, the project is not only achievable, but also has potential to touch nearly every person in world. This is how we think artificial intelligence and deep learning should be developed and deployed,” project lead Dr. S.P. Kothari of MIT’s Sloan School of Management said in a prepared statement.
The Singularity Healthcare project is slated for launch in the 2nd quarter of 2016, according to the company.
“The proposed deep-learning solution combines all layers of machine learning into a single pipeline, and then optimizes and meshes with other machine-learning algorithms on top of it. Starting this endeavor with the enormous trove of meta data in Imaging Advantage’s archives, we can learn how decisions made at the initial, raw representation stage impact the final predicted accuracy efficacy,” Kalyan Veeramachaneni of MIT’s Institute for Data, Systems and Society said in a prepared release.
“Given the advances in the field of artificial intelligence that have taken place at MIT and elsewhere, and Imaging Advantage’s scale, we are not only optimistic about a successful outcome, but expect it to be realized on an accelerated schedule,” Dr. Kothari said in prepared remarks.