Author: David Futoran, Somatix
Now more than ever, patients are depending on healthcare systems to keep them healthy. The global COVID-19 pandemic has led to massive staff shortages and quarantine protocols that have left many patients, both at home and in care facilities, without the care they need. Somatix is leveraging wearables, gesture detection, and Big Data analytics to tackle this problem.
Somatix has developed a real-time, AI-powered remote patient monitoring platform called SafeBeing, which is rooted in data science to keep patients safe, monitor their health and wellbeing, and enable healthcare providers to observe at-risk patients.
A world of clinical and predictive insights
While many technologies now use vital signs to track and evaluate patients remotely, we can use data science to find clues – a world of clinical and predictive insights – in places that were historically rarely analyzed: people’s daily activities.
A patient’s day, ranging anywhere from their daytime patterns (e.g. drinking, activity levels) to their nighttime patterns (e.g. sleeping, bathroom use), sheds light on signs of an infection and risks for adverse health events.
Especially due to the pandemic, patients and namely seniors have drastically altered their activity patterns. For example, seniors who would previously spent their week engaging with senior day care, recreational activities, doctor’s visits, therapy sessions and family have been in isolation in one place. There are downstream adverse effects of such a transition, including both cognitive decline, muscle atrophy and increased risk of falling or hospitalization. Using our technology, we can monitor for these risks and detect them in real-time.
There have been three main factors that have enabled us to offer an impactful solution to healthcare organizations beyond the standard measurement of vitals.
(1) Collecting unique and relevant data
Somatix is powered by our real-time, patented gesture detection technology. Using machine learning and artificial intelligence algorithms, we can differentiate among gestures (e.g. drinking, smoking, taking a pill) using only a smart band, with high specificity and sensitivity.
Figure 1: An example of data we collect when an individual has a drink.
We analyze movement data 24/7 and build a map of an individual’s Activities of Daily Living (ADLs), including their general movements and sleep patterns.
(2) Analyzing the data for clinical and predictive insights
The data enables us to scan for clinical and predictive insights based on changes in activity patterns. Specifically, we can connect certain activities or deviations from baselines with risk factors for events such as urinary tract infections, falling, or hospitalizations.
Somatix’ analytics have resulted in promising early clinical results, with a case study on an elderly patient population showing a 17% reduction in hospital readmissions. During COVID-19, identifying preventable readmissions is critical for keeping vulnerable patients out of the hospital to reduce their risk of disease transmission and free hospitals for COVID-19 cases.
The insights we have gleaned from activity data have proven to be extremely powerful predictors of clinical deterioration. For example, the movement heatmap below depicts the data for a patient who had normal blood pressure, normal oxygen saturation levels, a healthy heart rate, and no fever. To the human eye, the patient appeared to be stable. However, SafeBeing presented a different picture to the care staff, which you can see quite clearly in the visual below.
Figure 2: A movement heatmap showing a change in condition on May 21st, easily noticeable to the sensors but not visible to the human eye.
Our algorithms accurately detected a change in condition, deterioration in this case, based on the patient’s movement patterns alone. Five days later, the patient began showing symptoms of decline and infection, and vital signs fluctuated accordingly.
This purpose of this anecdote is not to downplay the importance of vitals. In fact, our smart band is built to continuously measure oxygen saturation and heart rate in addition to gesture detection and movement analytics. Rather, this example is included to demonstrate that there are telltale signs of decline, subtler symptoms, that we can now detect that go beyond, and sometimes even precede, any indication given by vital signs.
(3) Creating a solution that is practical to use
This last component is extremely important. Data science is only as good as the data collected. Technological literacy, implementation cost, and engagement levels have historically been challenges to collecting meaningful data even before COVID-19. In today’s environment, quick, simple, and remote deployments are necessary.
Our solution can collect quality data because it is user-friendly and does not require hardware (sensors, beamers, base stations). There are no installations that require technicians or maintenance. To collect data, there is no reliance on the user to answer questions or engage with the device. All data are passively collected from the smart band and transmitted to our cloud-based platform, which enables us to be remotely deployable within 48 hours.
Likewise, to be impactful, the data need to be easily actionable. Besides the use of standard notification channels (SMS, app notification, email), we are integrated with Electronic Health Records. We also have two-way audio and video communication functionality for telemedicine, ensuring the data get to the right place and all staff can intervene from the same platform.
With the right user experience, data science is well-positioned to make lasting improvements to our healthcare delivery system for COVID-19 and beyond.
* * *
Somatix® is a provider of wearable-enabled, AI-powered Remote Patient Monitoring (RPM) solutions for healthcare providers. Somatix serves elderly care facilities, hospitals (monitoring discharged patients), and substance abuse rehabilitation centers. Its cost-effective platform uses patented gesture detection technology, machine learning algorithms and advanced analytics to remotely and passively analyze user's activities, parameters and related events in real-time. This data delivers important clinical insights to healthcare providers, helping them maintain continuous contact with and improve the well-being of those under their care.