Case Studies

Shimmer has over 10 years’ experience providing wearable sensing equipment for clinical and other research needs. Shimmer has supported thousands of researchers in thousands of studies in over 75 countries around the world. We provide four examples below. 

*Note: In some cases, confidential information has been redacted.

Case Study 1: Remote Monitoring of Activity

Boston University (BU) is conducting an ongoing clinical trial to assess the impact of exercise on cognitive abilities. In this trial, subjects have an initial battery of cognitive tests to assess their baseline. Subjects are then put in one of two groups: an exercise group and a control group. The exercise group are enrolled in an exercise plan. There is no intervention in the control group. At the end of a 12-week test period, both groups were re-evaluated using the same battery of cognitive tests. During the study period, both groups were called weekly by the researcher to check on their status. The activity level of both groups was monitored during all waking hours using Shimmer3 IMUs. IMU data for the entire 12-week period was collected on the sensor itself and uploaded to a local computer at the completion of the 12-week period.

This study demonstrates a number of Shimmer’s capabilities that are directly relevant to the current study proposal:

  • It involved remotely monitoring unsupervised subjects over an extended period of time (12 weeks instead of 2 weeks).

  • Shimmer customized its activity algorithms for the study.

  • Shimmer processed the data for the first several participants and validated the process.

  • Shimmer then integrated the algorithms into its standard software so that BU is now able to run the automated algorithms completely independently.

  • This study stored the data on the sensor for the entire remote period, then the data was uploaded when the subjects returned the sensors to the study site at completion of the study.

Although this study is ongoing, data collection has been occurring successfully over the past two years.

Case Study 2: Verisense Beta Trials

Shimmer is conducting three beta trials with the Verisense IMU system. 

  1. The first trial is being conducted by the Letterkenney Institute of Technology and is studying fatigue in cancer patients using a wrist worn Verisense system. Participants in this trial will wear the Verisense IMU sensors for 12 months of continuous monitoring.

  2. The second trial is being conducted by the Spaulding Rehabilitation Hospital’s Motion Analysis Laboratory to analyze recovery from stroke. Each participant will wear two Verisense IMU sensors, one on each wrist, allowing the investigators to analyze the differential motion on each side and how it changes over time. 

  3. The third beta trial is being conducted by Boston University at their Alzheimer’s Disease Center. Each participant will wear a single Verisense IMU sensor on their wrist. Activity and sleep patterns will be analyzed over time. This study is designed to be a prototype for continuous monitoring of Alzheimer’s patients. 

Case Study 3: Innerscope Research Data Collection System

Innerscope Research, Inc. hired Shimmer to develop “bridge” software to integrate the Shimmer3 GSR+ and ECG sensors into Innerscope’s automated data collection system. This bridge software allowed Innerscope to access data from the sensors in real time.  Shimmer’s development was done on time and on budget. The Innerscope system was used in ~100 studies to collect biometric data on tens of thousands of people. It was exceptionally reliable and was able to be successfully used by minimum wage workers in a shopping mall environment.

Case Study 4: Comparing Impact Accelerations

A Shimmer customer compared impact accelerations during un-planned and pre-planned lateral cutting. The goal of this was to provide insight into the role that the physical and cognitive demands of sports play in acute musculoskeletal injury risk.

15 males and 15 females, ages 18-30 years old, who had demonstrated experience competing in sports involving frequent landing and cutting participated in this study. Prior to the warm-up/testing, a Shimmer3 IMU sensor was secured to the anterior-medial tibia of the subject’s non-dominant limb, 8 cm above the medial malleolus, where it was set to sample at 500 Hz (range of ± 16 g).

Subjects then completed a warm-up which involved alternating between bodyweight squats and vertical jumps. Immediately following the warm-up, the subjects began the testing protocol, which required the execution of lateral cuts in two randomly ordered conditions: pre-planned and un-planned. For the pre-planned trials, subjects would stride forward from a distance of 1.5 m, land with their nondominant limb, and immediately cut laterally in the opposite direction. The initial cut (non-dominant limb) and subsequent landing (dominant limb) needed to occur within 40 cm x 50 cm areas outlined on the floor to ensure consistency among trials, conditions, and subjects. For the un-planned condition, subjects were unaware of the maneuver to perform until after initiating a trial. As a result, they were unable to pre-plan their movement, which imposed temporal constraints on decision-making.

As expected, tibial impact accelerations were higher when subjects could not pre-plan their movement. There was also a small-medium increase in impact variability for the un-planned trials. In general, it appears that the subject’s ability to attenuate impact accelerations during cutting may have been compromised when they could not pre-plan their movement.

Case Study 5: Mobile Life Logging

A Shimmer customer used Shimmer ECG and IMU sensors to develop a lifelogging platform to measure negative emotional states during real-life driving. The data that has been collected includes acceleration (to calculate speed), photographs of the road in front and ECG signals (to calculate cardiovascular activity). Each participant undertook the study for five working days and recorded data during each of their driving journeys to and from work. A minimum of 20 minutes of continuous driving per journey was required to ensure a sufficient period of data collection. Before commencing the study, participants were briefed with a description of the task and received a demonstration of how to use the equipment. Once the participants had reached their destination, they were then required to stop recording data and remove the equipment.

Once finished, all of the collected data was analyzed using three-minute windows (minimum epoch necessary for analysis of HRV). Before calculating HR, the ECG signal was filtered using a Chebyshev Type I 2nd order highpass/lowpass filter with a cut-off frequency between 0.5 and 200 Hz. Once filtered, the R-R intervals were calculated to subsequently calculate HR in BPM and HRV. The acceleration signal was also filtered using a 1st order Butterworth low-pass filter; speed was then calculated from this signal by combining the acceleration vectors (x,y, and z) into one vector and converting this acceleration signal (captured in m/s2 into velocity).

Significant correlations between mean speed and either HR or HRV were found for ten of the thirteen participants (77%). When the relationship was significant, both HR and HRV had a negative (inverse) relationship with mean speed. Therefore, low mean speed was associated with increased HR, which was indicative of anger due to journey impedance.