- Amazon Web Services is helping Beth Israel Deaconess Medical Center explore an array of machine learning applications in medical care, via a $2 million multi-year grant.
- Using AWS tools, the medical center will look at ways artificial intelligence and machine learning can improve clinical care and increase efficiencies, with the goal of creating data-based processes that can be scaled across the healthcare industry, Matt Wood, AWS general manager of deep learning and AI, wrote in a Monday blog.
- In one of its first projects, the Boston-based teaching hospital used machine learning to optimize scheduling for 41 operating rooms and improve patient flow within the hospital. Another project used machine learning to enhance operational flow in ORs.
The needle is starting to move on AI's potential to disrupt healthcare, as companies and health systems invest in deep learning systems to recognize patterns and predict actions that may help workflow efficiencies and patient care. Global spending on healthcare AI products and services is expected to exceed $34 billion by 2025, according to market research firm Tractica.
Among those mining and leveraging AI-powered data are Amazon, Siemens and IBM, as well as health systems like Mayo Clinic and Intermountain Healthcare.
Some of the biggest opportunities currently are in scheduling, operations and billing — repetitive, time-consuming tasks where AI can augment human efforts to reduce costs and improve results. AI can also be used to track medications and medical devices in hospitals and their supply lines.
But along with the promise comes risk and uncertainty. Last year, a STAT investigation found IBM Watson wrongly advised doctors on how to treat cancer patients, and suggested the super computer’s advice does not rely on real patient data as claimed by the company. No patients died as a result of the misleading advice.
One of the ways the hospital is using machine learning is to reduce delays and rescheduling of medical procedures when patients' history and physical forms aren’t easily located. To address the problem, the hospital deployed Amazon Comprehend Medical to pull out key medical terms and insights used to identify H&Ps, saving time and preventing schedules from going off track.
"Every minute spent on cumbersome clerical tasks and management adds up to millions in lost productivity and directly impacts patient care," John Halamka, executive director of the Health Technology Exploration Center at BIDMC, said in the blog, according to AWS.
Other projects underway include predicting which patients are likely to miss appointments and reaching out them with reminders and detecting where changes in OR scheduling can increase efficiency, lower costs, balance the workload during peak times and reduce unwanted effects on patient care.
Future plans including assessing machine learning’s impact on managing intensive care unit flow and anticipating surges in patient volume. BIDMC will use Amazon Cloud to store and process data from the projects, along with Amazon SageMaker to construct deep learning models to foresee where and when space will be available, Wood said.
Amazon has been making steady inroads into healthcare. Last year, the retailing giant acquired online pharmacy PillPack and launched a joint venture with J.P. Morgan and Berkshire Hathaway aimed at curbing healthcare costs for their employees.