Digital

Amazon’s health care push expands to machine learning for the EHR

Alexa, is my patient taking her blood-pressure medications? 

That could be the kind of question physicians can answer with the help of Amazon.com. The online retailing and cloud-services giant is bringing its analytic expertise to health care and will be competing in the same $7 billion space as IBM’s Watson Health and United Health Group Inc.’s Optum. 

Related Coverage

Don’t fall for these 3 myths about AI, machine learning

“We are excited to announce Amazon Comprehend Medical, a new HIPAA-eligible machine learning service that allows developers to process unstructured medical text and identify information such as patient diagnosis, treatments, dosages, symptoms and signs, and more,” wrote Amazon artificial intelligence leaders Taha Kass-Hout, MD, and Matt Wood, PhD, on the Amazon Web Services Machine Learning Blog. 

They noted that the technology is being previewed at the Fred Hutchinson Cancer Research Center in Seattle and by Roche Diagnostics. 

Wood and Dr. Kass-Hout wrote that Amazon Comprehend Medical “may be able to one day help consumers with managing their own health.” This could include doing their own medication management, proactively scheduling care visits, or “empowering them to make informed decisions about their health and eligibility.” 

Amazon’s previous entries into the health care space include the $1 billion acquisition of the PillPack Inc. online pharmacy and a joint venture with JPMorgan Chase & Co. and Berkshire Hathaway Inc. Famed surgeon and writer Atul Gawande, MD, was tapped as the venture’s CEO. 

Why physicians’ insight on AI is critical 

The AMA has noted that augmented intelligence (AI) holds promise for health care improvement, but physicians’ perspective is needed in its development. The AMA House of Delegates adopted policy at the 2018 AMA Annual Meeting that seeks to seize the opportunities offered by AI to provide a transformative set of tools to help patients and physicians. 

The AMA is leveraging its ongoing engagement in digital health and improving patient outcomes and physicians’ professional satisfaction to help set priorities for health care AI.  

In addition, the AMA is promoting the transparent development of clinically validated health care AI that is thoughtfully developed with best practices in user-centered design, and takes steps to address bias and avoids introducing or exacerbating health care disparities including when testing or deploying new AI tools on vulnerable populations.  

“Combining AI methods and systems with an irreplaceable human clinician can advance the delivery of care in a way that outperforms what either can do alone," AMA Trustee Jesse M. Ehrenfeld, MD, MPH, said at the time. "But we must forthrightly address challenges in the design, evaluation and implementation as this technology is increasingly integrated into physicians’ delivery of care to patients.” 

In a Health Affairs blog post co-written by AMA Chief Medical Information Officer Michael L. Hodgkins, MD, MPH, and Shantanu Nundy, MD, director of the Human Diagnosis Project, the doctors noted that the real promise of health care AI is to enhance the work of physicians—not to replace them. 

“As we learned from the flawed implementation of the EHR, this transformation can only take place if physicians are instrumental to the design, validation and implementation of augmented intelligence systems,” Drs. Hodgkins and Nundy added. 

The two physicians outlined three ways that AI could be used to improve physicians’ workflow while simultaneously contributing to burnout relief and prevention.  

Point-of-care learning. AI can personalize content physicians need and want by analyzing practice data, online search queries and formal and self-completed assessments.  

Clinical documentation. They envision programs that can analyze a physician’s free-text narrative, extract relevant information and insert it into the appropriate structured data field.  

Quality-measurement reporting. AI could replace manual data-collection processes by reviewing clinical documents and extracting information for quality reports and to populate missing data fields.  

“By freeing doctors from clinical documentation and quality measurement as well as enhancing the value of practice-based learning, AI has the potential to augment the most foundational aspect of high-quality care: the doctor-patient relationship,” they wrote.