Healthcare: Managing the information flow from wearables and remote monitoring devices
Sensors at the heart of the system
Sensors are an integral component of wearable systems. Valencell makes an optical heart monitor sensor that is embedded in 50 different devices, mostly in consumer wearables. However, interest from the medical and corporate community is also increasing. “We are seeing greater interest from additional healthcare and wellness communities as well as from pharmaceutical and health companies,” said Ryan Kraudel, vice president of marketing at Valencell. The sensors can be put into many products, including devices such as Fitbits and smart watches, clothing, earbuds, hearing aids, or even into patches in objects such as hard hats or clothing to track heat stress and prevent heatstroke.
Sensors can detect heart rate, heart rate variability, cardiac efficiency, and other biometrics. For example, measuring how many steps a user takes between heartbeats can be a proxy for cardiovascular fitness, commented Kraudel. “A decreasing number of steps per heartbeat can be a signal of worsening cardiovascular condition.” Blood pressure is measured by use of a reflected waveform that measures blood volume variability.
In clinical settings, devices using medical heart monitors track the same heart measures that are tracked in consumer devices. They can also be used for pain management in conjunction with neuromodulation therapy, which is a form of visual light wave therapy used to alleviate pain. In this scenario, the sensor helps measure the effectiveness of the treatment by providing biofeedback that identifies the wavelengths that maximize the impact of the therapy.
Data may go to a mobile phone app that is the presentation layer for the user, onto a sports watch, or into a medical system. “In a clinical setting, the data may go to a clinical database or to the patient’s EHR and be integrated into a treatment protocol,” commented Kraudel. Part of the engineering challenge with optical sensors is to do as much processing at the edge as possible to produce the relevant datapoints and provide more ready-to-analyze data. Inputs from sensors very quickly turn into big data scenarios, so managing the data upstream is a valuable capability.
Numerous anecdotal reports indicate that devices such as the Fitbit have alerted users who were unaware of having medical issues that they had an abnormal heartbeat, which turned out to indicate blocked arteries, kidney failure, or other serious problems.
Processing the data stream
Wearables and remote monitoring devices produce a steady stream of data that becomes most valuable when it is analyzed. Validic offers a platform that can consume inputs from more than 400 wearables and in-home devices. The Validic Inform connectivity platform is able to continuously stream personal health data, and Validic Impact, built on the connectivity platform, integrates directly into clinical systems such as the electronic health records, and analyzes patient data from home health devices, including wearables. Users include participants in wellness programs and health clubs, healthcare systems, insurance companies, and pharmaceutical companies.
The platform solution was developed first to serve wellness programs, and a few years later, Validic repositioned its solutions toward collecting and analyzing clinical data such as blood glucose and blood pressure measurements. Healthcare systems were interested in understanding the data but did not have solutions that could easily ingest or analyze it. “The companies had challenges with the amount of data coming from clinical systems,” said Brian Carter, senior vice president of product at Validic. “They wanted to be able to monitor the data as it flowed, flag issues of concern, view the data on a graph, and perhaps change the rules around the data.” An emerging use is remote collection of clinical trial data by pharmaceutical companies and contract research organizations. “This data can be used both for measuring the effectiveness of the drug being tested and also to detect side effects,” noted Carter.
Validic uses a MongoDB cluster and Amazon Simple Storage Service to store historical data, but for the streaming data, big data solutions such as Apache Kafka are required. “We process billions of datapoints per week,” said Carter. “We also have invested in AI and machine learning that let us look not only at metrics as they are happening, but also can be used for prediction.”
Wearables are serving an increasingly wide group of users and serving multiple purposes. Carter concurs that sleep metrics and activity metrics provided by wearables in healthy individuals can be indicators of many pending conditions. “These are good proxies for heart failure and other events,” Carter commented. “In addition, when a patient recovering from a heart attack is in rehab and has target levels for the amount of activity and the heart rate, wearable inputs are more reliable than self-reporting and more convenient, since the patient does not have to enter the data.”
Remote monitoring is a component of telemedicine but has taken off to a greater degree than other areas of telemedicine such as videoconferencing. Telemedicine as a primary means of interaction has seen only moderate uptake. “A major issue is that adding telemedicine to an existing healthcare system is not a simple change,” observed Antin. “The infrastructure requirements are significant, and there is also a change management issue, especially for large systems. There also needs to be an economic incentive, with proof that it will be both more efficient and at least equally as effective to justify the investment.” Restrictive telemedicine regulations also limited its use, but these were recently modified in response to COVID-19. This technology has already seen a jump in use that is likely to continue. With broader use of wearables, routine monitoring of even asymptomatic high-risk groups such as the elderly may become the norm, allowing earlier intervention and improved outcomes.