The term Artificial Intelligence (AI) tends to conjure images of super futuristic technology where human workers are ultimately replaced
But in reality, AI will likely serve as a way to augment and support the capabilities of highly specialized human experts. AI reflects a wide range of analytic methods, some of which are already well established, accessible and easy to implement. There’s a wide range of practical use cases that can assist in healthcare delivery today.
Remote patient monitoring (RPM) is an important development in healthcare that has been around for many years. It has taken a variety of forms and continues to evolve as technology evolves. The importance of RPM in a post-COVID world has also become clear, as efforts are underway to shift non-acute care from the hospital back into the community and the home. This new urgency, coupled with advances in technology, will bring RPM further into the mainstream in 2021. AI is one of the key enabling technologies that can help accelerate this progress.
It is now possible to capture real time, digital data points to facilitate RPM and to combine it with other clinical, administrative or contextual data. But, it’s only effective if that digital data is translated into actionable insights. This convergence of data, now available in real time, can enable predictive and proactive biomarkers so healthcare providers can intervene earlier in the diagnostic and care cycles, and appropriately triage patients so that the right caregivers and experts can be engaged. There’s additional value to be gained by consolidating this data across many patients — leading to a better understanding of disease patterns at the population level. This can lead to the initiation of targeted public health and prevention programs where they are needed most.
Actionable and Accessible
So how do RPM programs ensure the capture of quality, actional data? To start, it’s imperative to make it easy for patients to participate. Patients and their families want simple, convenient, connected devices that don’t interfere with their day-to-day lives. When it’s easy to wear the sensors, it’s more likely that compliance will continue and lead to the capture of continuous, quality data points. Key to this is continuous data collection via medical-grade sensors, such as those provided by Vivalink. Sensors like those offered by Vivalink allow for passive, continuous monitoring of patients. This type of high volume, quality data is necessary for the application of AI in RPM.
There are several examples of how data from RPM programs and AI can be used in different clinical specialties, including oncology, cardiology, psychology and acute care:
In the oncology space, when a patient is discharged following in-clinic chemotherapy, RPM can help providers assess conditions indicative of serious side effects such as neutropenic fever by monitoring and correlating three basic vitals: heart rate, respiratory rate, and body temperature. When a certain threshold is reached, the provider can be alerted immediately. It’s a fairly simple formula, but insights derived from the data can potentially indicate complications or deterioration in the patient’s condition that can be addressed immediately with telemedicine visits, meaning the patient won’t need to leave home. Without this, the patient’s condition could worsen over time resulting in emergency room visits or further complications.
In cardiology, data collected via RPM supports algorithm-based biomarker identification. In a 10-year, 3,000 patient longitudinal study by University of California, San Francisco (UCSF) designed to detect biomarkers of early atrial transformation in atrial fibrillation (AF), subjects used VivaLINK wearable ECG monitors to continuously capture data. Monitoring included ECG recording for rhythm, heart rate and RR-interval with the goal of determining incident AF as well as progression and recurrence of AF. Participants were expected to wear the ECG monitor for one week at a time each month, on an ongoing basis. Data captured from the study is retrospectively analyzed using ECG software that helps identify possible development and progression of AF. This use of AI can help predict the onset of AF before it happens.
In a behavioral science study by Stanford University on the link between stress and teenage depression, wearable sensors were used to continuously capture physiological stress levels over the course of multiple days in order to discover correlations between stress and depression.
In addition to chronic conditions, RPM plays a growing role in virtual acute monitoring. The idea of allowing patients to return home sooner following an acute event has gained traction over the last year. Hospitals are setting up “hospital at home” and various types of virtual hospital programs. These all leverage the same capabilities used for less acute remote patient monitoring — easy to collect, accurate, real time, and continuous data capture coupled with advanced algorithms that allow for a more predictive and proactive approach to patient care.
The idea of a virtual ICU that serves a community hospital or a hospital in a remote area or even another country is another viable way to leverage RPM devices and AI to ensure even if an expert isn’t physically at the facility, the patients have access to exactly the right experts.
Successful RPM addresses not only technical challenges, but also human factors involved with user adherence, requiring both a quantitative and a qualitative approach. While there’s still tremendous future upside to further technological advancements in AI and RPM, it’s critical for providers and patients to understand that quality devices, technology, and interface solutions are available now. There are secure, viable delivery systems that take complex data sets and make them accessible to providers.
About the Author
Jennifer Esposito is a member of Vivalink’s board of directors. She has spent most of her career at the intersection of technology and healthcare, currently at Magic Leap and previously at Intel Corporation and GE Healthcare.