According to a 2013 study published in the Journal of Patient Safety, 440,000 deaths each year can be traced back to preventable medical errors. If the U.S. government’s Centers for Disease Control and Prevention took these metrics into account, medical errors would be the third leading cause of death in the United States, behind heart disease and cancer. Julian Goldman, a physician at Massachusetts General Hospital and a co-leader of the health-care task group for the Industrial Internet Consortium (IIC), a member-led organization that is helping to accelerate the deployment of networked sensor-based technology in the industrial sector, hopes to change that.
Toward that end, the Massachusetts General Hospital Medical Device Plug-and-Play (MD PnP) Lab, a research center that Goldman established in 2004 to help improve patient safety through device interoperability, is partnering with global systems integration firm Infosys, middleware provider Real-Time Innovations (RTI) and manufacturing software developer PTC to run the Connected Care Testbed through the IIC.
“We want to make it harder for mistakes to be made,” Goldman says, adding that medical errors are defined as any preventable adverse effect of care. These mistakes often occur because medical devices, such as infusion pumps and heart and blood pressure monitors, operate in silos, which can lead to failures to respond to a crisis.
Say, for example, that a nurse receives multiple alarms throughout the day that a given patient is suffering from low blood-oxygen saturation, but each time he goes to the room to respond to the alarm, he finds that the patient is holding a phone or trying to reposition himself in bed, which is causing the finger-clamp pulse oximeter to dislodge. Eventually, the nurse becomes fatigued and does not respond as quickly to the next alarm from the blood-oxygen monitor.
However, the next time the alarm is triggered, it mignt be because the patient’s family member, concerned that he was in pain while trying to sleep, had accidentally overdosed him by pressing his morphine pump—either because she wanted to decrease his pain or because she mistook the device for the nurse call button. If the blood-oxygen monitor were linked to the morphine pump and other medical devices used to monitor patient well-being, the alarm system could offer the medical staff a more complete view of what is happening inside the patient’s room, and how much danger that person is in.
Medical device makers and the software platforms that medical staffers use to monitor patients, Goldman says, have “not done a good job of data fusion, so alarms happen when they don’t need to.”
To build more intelligence and situation awareness into alarm systems, and to enable interoperability between medical devices, the MD PnP Lab is developing a cloud-based open-source data-communications platform called Integrated Clinical Environment (ICE). ICE has been developed as an ASTM standards-based framework for integrating medical devices, the data they generate and health-care IT systems. The goal of the testbed, which will take place both in the MD PnP Lab (using devices utilized in clinical settings) and within a residence (using home health-care equipment), is to bring the ICE platform closer to deployment by assembling the types of technologies and expertise required to ensure proper data collection from disparate devices not designed to work together.
Infosys is leading the testbed, working with RTI to apply its middleware, and with PTC to use its ThingWorx application platform to run on the ICE platform, which will provide data modeling to enable a user interface for medical personnel or users of home health-care systems that will integrate information from multiple devices.
All testbed participants are currently in the planning stage of the testbed, but by the third quarter of this year, the companies plan to launch both residential and lab trials. Inside the home, the types of devices that will be linked through the ICE platform will include a fitness monitor, such as a Fitbit, an Internet-connected scale, a blood pressure monitor and a fall-detection system. Within the lab, the equipment will include a pulse oximeter, a heart-rate monitor and other equipment generally used at hospitals.
The testbed’s objective is to show that networking these devices, as well as collecting and analyzing data from all of them through a single interface, can make the process by which alarms or other alerts are sent to caretakers more accurate and insightful, thereby reducing medical errors related to their use. Goldman and his colleagues at the MD PnP Lab estimate that the ICE platform and the use of connected medical devices could reduce preventable adverse events (medical errors) by 35 percent. “This could be achieved by using current and new sensors, actuators and algorithms to provide better data access, improved data quality, real-time clinical decision support and enhanced vigilance,” he explains.
That 35 percent figure is based on the analysis of a number of past reports, including a 2010 study from the Office of the Inspector General for the Department of Health and Human Services, which shows that one of every seven Medicare beneficiaries who is hospitalized is harmed as a result of problems with his or her medical care. However, Goldman notes, this is a preliminary estimate. The potential impact of the ICE platform and the use of connected medical devices, he notes, will not be known until the lab is able to collect more data, which is why experiments such as the IIC testbed are important.
Data security is a major focal point of the testbed, since the MD PnP Lab would need to ensure that connecting medical devices via the Internet would not introduce new pathways for malicious actions that could result in harm or death to patients, let alone compromise patient data privacy.
RTI’s middleware is built on the Data Distribution Service (DDS) data-exchange standard that employs a common language to enable distributed systems, used in any industry, to send and receive data messages and commands. “We’ve modeled security risks,” says Brett Murphy, RTI’s director of business development, “and did an initial mapping on how they could be handled through the DDS security specification.”
In January 2016, the U.S. Food and Drug Administration released draft device interoperability guidelines that it wants medical device manufacturers to follow in order to improve the rate of innovation in the medical device industry, while also reducing the costs associated with developing and deploying new devices. The guidelines also include basic safety and security guidelines, upon which Goldman says the MD PnP Lab will build.
Developing robust cybersecurity tools is an obvious requirement, Goldman says, though doing so should not stand in the way of achieving device interoperability. “We can’t afford to not leverage [device] interoperability to improve health care, just because we’re afraid of cyber threats,” he states. While connecting devices on a single platform increases the attack surface, Goldman adds, it also improves visibility and control since, currently, air-gapped devices (those not networked to the Internet or an intranet) could be attacked without medical staffers’ knowledge.