Wearable Sensors: It’s Not Just About Steps During the past decade, led by FitBit, Apple, Jawbone, Samsung, and many others, use of wearable sensors has become widespread in the consumer market. Although the sensors in these wearables generate millions of data points every day, most of the focus has been on counting steps. Doing 10,000 steps a day has become a common benchmark for staying active and fit. For consumers, this is a reasonable approach. Research suggests that step counters encourage individuals to make better choices. In the healthcare community, physical activity, including walking, is related to positive health outcomes1, while a sedentary lifestyle is associated with negative health outcomes2.
have developed applications to track rehab both to ensure it is performed and to measure progress in key metrics like range of motion. These applications can also be used for sports training, and many other companies are developing products in this area. Tremor Classification in Parkinson’s Disease Levodopa-induced dyskinesia (LID) is a major side effect of Levodopa treatment for Parkinson’s disease. When assessing treatment effect, the challenge is how does one distinguish between Parkinson’s tremors and movement generated by LID? Clearsky Medical used multiple sensors and machine learning to be able to discern the difference. Doctors using this diagnostic can now more effectively set the Levodopa dosage and reduce the impact of LID dramatically.
Counting steps may work for consumers, but medical applications require a completely different level of specificity and accuracy. Counting steps does not help much in assessing Parkinson’s tremors or epileptic seizures. Often, multiple measures must be used. According to the Clinical Trials Transformation Initiative (CTTI), “no single OA [outcome assessment] is sufficient to characterise and assess a disease state on its own. Rather, different classes of outcome measures should be used in combination to provide complementary information.”3
These three examples provide an indication of just how much valuable information can be collected by the simplest wearable sensors. There are hundreds more examples in the academic literature. But measuring motion is just one benefit of employing wearable sensors. There are many applications for electrocardiograph (ECG) sensors, which are well understood. But there are lesser known sensors, which monitor electromyography (EMG) and galvanic skin response (GSR) for example, that open up whole new applications.
During the past decade, thousands of researchers have worked to develop hundreds (probably thousands) of algorithms and created new metrics based on wearable sensor data. It is impossible to catalogue all of these developments in a single article, but we have listed a few illustrative examples below. The first three examples focus on motion analysis only, using inertial measurement units (IMUs) containing the same type of sensors that are ubiquitous in consumer wearables.
Using EMG and IMUs to Characterise Soft Tissue Injuries Soft tissue injuries are difficult to diagnose, and magnetic resonance imaging cannot help evaluate sprain and strain injuries. Emerge Diagnostics uses EMG and IMU sensors to conduct the electrodiagnostic functional assessment (EFA) test to identify the location and objectively diagnose the extent, nature and age of soft tissue injuries. Its EFA technology is registered with the US Food and Drug Administration as a Class II Medical Device for musculoskeletal disorders. One of its primary applications is to manage workers’ compensation injuries, reducing time lost due to injuries and the frequency and expense of soft tissue claims. It also improves treatment solutions and quality of life for workers.
Detailed Gait Analysis for Fall Prevention Falls are a major cause of injury and hospitalisations in the elderly, at a great cost to quality of life, not to mention annual healthcare costs of more than $50 billion in the US alone4. Research has demonstrated that an appropriate intervention can reduce the number of falls by 30–40%5, if researchers can identify in advance who is at risk. By measuring more than 20 detailed gait parameters and comparing them to a database of individuals that represent the general population, the Kinesis Quantified Timed Up and Go (QTUG™) solution can accurately estimate fall risk. In addition, it generates a complete mobility assessment for each individual based on comparisons with the reference database. These capabilities will make it valuable for tracking a wide variety of conditions, including Parkinson’s disease, multiple sclerosis, and any mobility issue. Joint Angle Measurement for Rehab Getting patients to perform their prescribed rehabilitation exercises and monitoring their progress are major challenges. Much of the assessments for musculoskeletal disorders are based on individual self-report and surveys, methods that are prone to error and are qualitative at best. Using IMU sensors, however, it is possible to count repetitions and measure joint angle automatically. A number of companies, including Telefonica (Rehabitic) and Dycare™ Lynx, 46 Journal for Clinical Studies
Using GSR to Identify Early Indicators of Emerging Mental Health Problems Many mental health problems begin in early childhood, sometimes as early as six months. Diagnosing them at that young age is extremely difficult. However, the ‘watch me grow for real’ longitudinal study being conducted by researchers at the University of Sydney is taking a very creative approach. They are examining whether it’s possible to identify early indicators of emerging mental health problems in children three years of age and younger using GSR and an optical pulse sensor. Specifically, the study hypothesises that children’s responsiveness (as measured through psychological arousal using the GSR sensor), emotional attention (captured using eye tracking), and learning can differentiate between common mental health problems, such as anxiety, autism and aggression. The experimental paradigm included play sessions, stressful interaction, and a returnto-baseline period with the primary caregiver. The goal is to capture GSR differences in how these children respond to the above listed interactions and then return to normal functioning. Volume 11 Issue 3