Author: Terry Duesterhoeft, Chief Product and Commercial Officer, EarlySense
Looking back at the evolution of the RPM market over the last 16 years, I have seen massive change in the technology, economics, applications and the market expansion. The first generation of RPM solutions were expensive, purpose-built systems – initially costing over $5K pre-2005, for a kit containing 3 hard wired vital sign sensors and the monitor – with limited connectivity options (POTS with dial-up modems, eventually moving to utilizing the wireless paging network and cellular modems), connecting to premises-based workstations and servers. The technology and especially the software, often evolved from acute care patient monitoring, initially including very basic patient data management and dashboard applications. Evolving to introduce rules-based algorithms to support patient health surveys, clinical pathways, and alert management. With some systems having to support integration with 15 – 20 different EHR’s.
A second-generation of RPM solutions began to leverage fast moving consumer electronics and emerging wireless consumer health products. These solutions were much lower cost (<$1000 for 3 wireless bluetooth sensors, and patient facing app on a commercially available tablet, and utilizing commercially available wireless vital sign sensors), cellular connectivity was generally standard and software applications and data services were cloud-based. Software applications were more comprehensive and saw the introduction of machine learning to help extract more valuable insights from the data and enhance alert management. These second-generation systems were introduced around 2010. While the overall processing power, wireless communications, interoperability, and logistics had improved greatly, the same asynchronous, once (to a couple of times) per day patient data acquisition, remained. As did some of the limitations of traditional RPM monitoring devices and solutions. Including patient compliance and user error; data reliability (stable readings from devices validated against a standard and being used consistently); and limitations in caregiver/clinician productivity. However, RPM was also still limited to one (or a small number of) bio-metric reading (s) a day. Nurse and other caregiver resource shortages, along with patient compliance, became practical barriers to increasing the number of biometric readings (and the subsequent increase in data). Thus, improving the ease in which caregivers can digest increasing amount of data along with maximizing patient compliance remains key steps to taking RPM to the next level.
Looking at RPM today, both expanding reimbursement and the COVID-19 pandemic have been catalysts for RPM adoption. However, while adoption is accelerating, some of the same limitations remain.
Is there a third-generation of RPM solutions on the horizon? If so, what would it look like? In my discussions with providers and clinicians, and understanding their pain points in receiving actionable insights from RPM, I strongly believe that passive monitoring and continuous data powered by AI will be key components helping solve for the issues of patient compliance, data accuracy, data overload, and provide the catalyst for a new generation of solutions.
So to help understand why, let’s take a closer look.
Providers want to make sure data collected remotely is accurate and standardized, but also that patients comply with the correct use of the RPM solutions. Also, that these solutions can work with the largest percentage of the population of interest. Without that, the foundation of any good remote care model falls apart. Thus, the terms passive and continuous are important to this discussion. As used here, passive means that the patient isn’t required to do anything different in their daily life to collect biometric data. The system is invisible, behind the scene gathering health data. Continuous means that data is being gathered with high temporal resolution (i.e. gathered more often) than the more typical once per day RPM resolution. For example, gathering data once a minute vs once a day provides much greater information on the actual health condition of an individual. Thus, putting passive sensing and continuous data monitoring together provides both guaranteed compliance and a major step in more accurately determining the actual health condition of an individual.
Of course, it is critical to note the importance of using continuous monitoring devices that align this data to existing “gold standards” to ensure proper validation, regulatory clearance and the protocols used are verified by a qualified third-party to ensure consistency and quality.
As a first step, we need to collectively work on improving patient compliance to enable RPM and telehealth to fully realize their potential outcome improvements . The optimum way to do this is by introducing technology solutions into RPM that are built around passive and where possible, contact-free, patient involvement. Sensing technologies which are seamless enough to fit into the backdrop of everyday life, so the collection of data can be passive, continuous and automatic.
Once we are able to passively gather continuous vital sign and other biometric data it is important to be able to use it effectively. Having this level of data alone is great, but, that amount of data is not realistically consumable by a caregiver in its raw form. Thus, the importance of using data science tools to better understand the critical information available within the data, alerting a caregiver with only relevant changes. Thus, continuous vitals data powered by AI and other rules based algorithms can address and contextualize the massive amounts of data generated by continuous digital health technologies.
Rather than simply pushing more data into a clinician’s in-box, applying machine learning to a continuous stream of vitals data is an opportunity for us to help care providers realize the full potential of digital health. With providers the capability to continuously evaluate patient data while looking for meaningful patterns (or markers) and deliver actionable health insights in a way that easily integrates into existing telehealth and related virtual care management systems is critical.
One potential first step towards this, is to establish a scoring method to show stability and/or change in one’s health status. Such an indicator of health status which can be easily trended would fit into existing remote care management workflows. Then with a change in health status a caregiver can decide to go deeper and visualize which biomarkers are behind the change and even review important regions of Interest. However, to do this right, it will be Important to have the right data sets.
One initial application would be to provide this health status as a way to establish a baseline. The capability to capture continuous data allows for the creation of a dynamic baseline, plus the ability to dial in a baseline score for a patient as a gauge of how their overall health status is progressing or declining. Then as the score changes, allowing the caregiver to be able to drill down and easily visualize the biomarkers which are behind the change could provide a major productivity and quality improvement. Additionally, being able to filter out other factors which could be impacting the change [e.g. medication changes, environmental factors, other recent medical events etc.] would now be possible. This unlocks major improvements in providing a clearer, more reliable picture of an individual’s condition and also in enhanced caregiver productivity.
These two improvements alone can help anchor a new generation of RPM solutions. However, it still only provides a first step as it relies on the caregiver or clinician to determine the potential meaning of these insights. The second step may be to provide additional guidance as to what the presence of these biomarkers and external filter data really represent.. Of course, it is still early days with these emerging applications. But the ability to close the loop and provide ‘smart’ input to caregivers as to the possible conditions or condition changes behind the presence of these biomarkers could provide a real ‘Sea Change’ in the application of RPM.