1B3. Identify technology trends affecting healthcare IT
Growth of Consumer Technology
By 2015, nearly 80% of American households had both a computer, and internet service. Similarly, 86% of adult Americans also own a smartphone. It would be just a matter of time before regular folks decide to put the connectivity and computing power in their mobile devices and personal computers to work for their health.
Between 2014-2018, the adoption of health-related apps tripled. In 2019, there are more than 320,000 health & fitness apps in the major app stores, with 200 new healthy apps added every day. Consequently, the mobile health industry is worth $37,000,000,000 USD in 2019. The most popular mobile healthcare apps track fitness and healthy eating. However, the breadth of apps available also improve communication with providers and care coordination and assist with managing conditions. Healthcare providers have mobile apps as well. These include alerts and communication apps, supply chain management, reference and diagnostic aids, and consults.
Mobile apps have become such an integral part of healthcare, the US Food and Drug Administration (FDA) has begun regulating part of the market. The FDA has two classification of apps: 1) Wellness Apps, which enhance or track the overall health of the individual, and 2) Health Apps , which are mobile software that diagnose, track or treat disease. Most apps are Wellness Apps, but there are some interesting FDA Approved Health Apps worth noting. Health apps may also include a new sensor or other device that links with a phone. For example: Eko Digital Stethoscope, GoSpiro digital spirometer, SENSIMED Triggerfish for ocular monitoring, and the Neurometrix Quell smartphone controlled pain relief are taking us into a consumer-managed health device future.
A pocket EKG
AI Measuring blood loss
The US Department of Health and Human Services legacy definition of telemedicine is “ use of electronic information and telecommunications technologies to support and promote long-distance clinical health care, patient and professional health-related education, public health and health administration. Technologies include videoconferencing, the internet, store-and-forward imaging, streaming media, and terrestrial and wireless communications. ” Advances in the supporting technologies of telemedicine, particularly telepresence, have dramatically improved the capability and adoption. in 2019, 45 percent of outpatient facilities provide telepresence services of some type. Telemedicine adoption is facilitated through
- Widespread smartphone/mobile device adoption
- Improvements to device cameras and processors
- Greater broadband access and faster mobile data coverage
- A healthcare workforce comfortable with technology
- Reimbursement policies that provide payment for telemedicine visits.
Real Time Location Systems
Real-time location systems (RTLS) are technology systems that automatically identify and track the location of people or objects in real time. Typically, wireless RTLS tags are attached to objects or worn by people, and sensors throughout the building receive wireless signals from the tags to determine their location. The most common technology implementation uses powered Active RFID tags that have transmitting capability and RFID receivers ( as the sensors) in fixed known locations. Unpowered Passive RFID tags that have a much smaller range from the sensors, and GPS systems may also be used when precision can be low and satellite signals are strong enough.
RTLS systems enable organizations to more effectively keep track of where things are, which can help them improve processes they are currently struggling with, such as inventory management, room cleaning/turnaround, and item maintenance. When used as an “Item tracker” RTLS is designed to replace less efficient solutions such as spreadsheets and clipboards, by automating tasks that are manual. Consider Texas Health, who saved $412,000 in the first year by tagging & tracking equipment. Nurses waste several minutes each time they have to look for equipment like patient medication pumps and vital signs monitors. To make matters worse, when they come across equipment that they don’t currently need, they might move it to a room where they can find it later (called hording). Not only does RTLS decrease search times and make hording impossible, it helps with equipment maintenance turnaround times. If an organization faces theft losses of durable items, RTLS might help with that as well. However, unless you are using GPS, tracking is impossible once the tags are outside of the monitored area.
When patients and staff are tagged, RTLS becomes a quiet, real-time directory and rooming system. For decades, healthcare environments that cared for patients who may wander (e.g. dementia units) or need additional security (e.g. labor and delivery) tagged patients using RFID and simple sensor ‘gates’. Using RTLS instead of simple electronic gates at exits, the system knows exactly where the patients and the staff are and unlocks useful new real-time benefits. For example, you can easily find staff in noise-sensitive healthcare environments like NICUs, or even have patients check themselves in and get assigned a room by a kiosk.
Tracking people via RTLS can fight infections at least two ways: First, contact tracing (the exposure investigation that happens when anyone is discovered to have a particularity nasty bug) is automatic. With RTLS, you can push a button and know which patients and staff have come within feet of a patient with drug-resistant tuberculosis. Second, RTLS is an imperfect, but convenient way for measuring hand washing compliance. If you know where the sinks (or foamers), patients, and staff are, then you can generally tell whether a provider moved close to a sink before and after being close to a patient.
RTLS systems are also valuable to businesses because they generate enormous amounts of potentially useful data. Business intelligence can be generated about asset movement within the facility, how quickly processes are being completed, and what organizations such as hospitals can do to remove bottlenecks and speed up services. Data gathered by RTLS systems can be viewed in real-time, -like on a floor plan, or stored, analyzed, and audited as needed.
The technology behind RTLS is fairly mature and has benefited other industries for years. Perhaps like other IT innovations, RTLS has taked a back seat to the priority and resource demands of EHR implementations. RTLS can be expensive and very accurate (like 6mm accurate) using precise positioning equipment. However, If you simply need to know the room or general area to find a person or thing, then you can get extremely cost effective RTLS cost deployments. RTLS deployments, -particularity those that track people, can be difficult. RTLS will likely add to your management burden through staff resistance and uncovering existing issues you didn’t know about. You should expect to manage through resistance to Big Brother micromanaging the staff. However, RTLS will uncover staff behaviors (such as hording mentioned above), that you will have to deal with.
Before deploying a RTLS system, it is important to also consider the information infrastructure required. A single RTLS tag, -like any machine-generated data source, can produce hundreds of data points per second. Every second. For every person and thing with a tag. Although the data itself is small: a location and a unique ID for the tag, the high number of tags and high rate of reporting their location will generate a surprisingly high amount of streaming data. This streaming RTLS data can slow down networks that are also being used for EHR data, and will need a lot of storage at rest.
A.I. and Machine Learning
Artificial Intelligence, or A.I. is the ability for a computer to be an autonomous entity, to learn and do things without having been specifically programmed to do so. A.I. computer algorithms can make conclusions without direct human input. This is called machine learning. A.I. also powers the technology behind a computer’s understanding of human language (called natural language processing). Autonomy (the ability to be independent) is the A.I. of last century. It causes me a little bit of anxiety to think about it, but sentience (the ability to perceive, feel and experience) is expected on your desktop and mobile device within two decades.
Healthcare A.I. is currently booming in big data areas where there is more data than time, and limited understanding of how and where to start looking. For example, Google aims to accelerate drug discovery through virtual drug screening that creates and simulates of millions (and perhaps billions) of drug compounds to narrow down the number of drug treatment candidates to non-virtually create. Imaging is another healthcare A.I. sector with significant progress. Microsoft has developed set of diagnostic radiology tools that can identity and measure structures in medical imaging. Called ” InnerEye ” it can help radiologists identify and describe abnormalities in images. The following video is a quick overview of the InnerEye A.I. project:
Once A.I. makes the correct diagnosis, it (or she) may then provide the biologically-limited doctor with a treatment plan and prognosis. IBM’s Watson has been showing off its ability to read, talk, and think ( better than me at least) since it beat Ken Jennings on Jeopardy in 2011. Watson was designed to 1) ingest enormous amounts of unstructured healthcare knowledge in the form of journals, textbooks, and medical records. 2) Identify trends, make new associations, and synthesize new knowledge, and then 3) Recommend the best treatment for a specific patient. Watson is in use ( in a limited capacity ) by biggies such as Memorial Sloan Kettering’s Oncology department, who feeds him terabytes of data on cancer patients and treatments used over decades, so Watson can see patterns in patients , treatments, and outcomes.
I just asked Amazon Alexa “Why has it been so difficult for A.I. to get traction in Healthcare?” and she didn’t know. In fact, she didn’t even know what I was asking. I bet if you asked Watson, he might explain a few reasons:
- ‘Black box’ systems that doctors do not understand are not easy for them to trust
- Insurance reimburses for a doctors work, not a robots
- If a doctor is going to accept the liability for a treatment recommendation, they will take the time to work up the patient themselves, probably loosing any efficiency gained by the computer.
- Medicine is more than math. Remember “the art of medicine” ?
The irony is that A.I. technology adoption is accelerating in an industry that still depends on the fax machine. Although there is no doubt we will see wondrous things from A.I. and machine learning , I would not throw out the fax machine just yet. An interview with CMS Administrator Seema Verma indirectly reminded me of of this recently when she announced that A.I. plays “a key role” in patient empowerment and reducing healthcare spending, and “The timing is right for true digital transformation.” If you think you might have heard that somewhere before, you probably have (during the Health IT push in early 2000s, and again for meaningful use 2009, and again in 2015…)