SHIMMER Sensing Validation
The operation of the SHIMMER baseboard and daughterboards has been tested through a number of validation processes to determine the accuracy of the boards' function and their usefulness for biomedical-oriented research applications.
Electrocardiography (ECG) Validation
A number of tests were carried out to validate the SHIMMER ECG daughterboard as a valid tool for acquiring ambulatory ECG. The tests consisted of validating the ECG amplifier and ADC performance by using calibrated input signals and a resting ECG from a healthy subject in normal sinus rhythm. Simulated ECG signals as well as an ECG recording from a healthy non-resting subject were used to validate the performance SHIMMER ECG daughterboard for use in ambulatory monitoring. Figures 6 and 7 show a 1mV QRS amplitude (60 BPM) waveform generated by the Fluke MPS450 Patient Simulator captured by a SHIMMER ECG and a MAC 3500 ECG Analysis System (GE Medical Systems). The plotted waveform (Figure 6) was recognizable by a clinician as Normal Sinus Rhythm. Figure 7 shows the MAC 3500 ECG Analysis ECG System waveforms for comparative signal quality purposes. Visual examination of the waveforms indicate that they compare well.
QRS Detection of Normal Sinus Rhythm for a Non-resting Healthy Subject
A 5.9 minute ECG recording containing 503 heart beats from a non-resting healthy subject during a moderate walk was captured by the SHIMMER ECG and also captured by a Medilog Holter monitoring system. The R-R intervals and instantaneous heart rate (HR) identified by using the Medilog Holter ECG monitor software were compared against the R-R intervals calculated by using previously reported QRS detection and R-R interval correction algorithms [61, 62]. Each automatically detected QRS point on the SHIMMER ECG was manually verified to ensure correct detection by using the QRS detection algorithm.
The mean R-R interval for the SHIMMER acquired ECG was 0.7049 seconds, while the mean R-R interval for the Medilog acquired ECG was also 0.7049 seconds. The percentage difference between the R-R intervals for each acquisition was calculated on a beat-beat basis. The mean percentage error between the R-R intervals calculated, by using each acquisition, was found to be 0.0192 percent which can be considered negligible. These results indicate that the SHIMMER ECG can be used to acquire ambulatory ECG from resting and non-resting human subjects for research application purposes.
To validate the SHIMMER platform for use in studies of human gait analysis, temporal gait parameters derived from a tri-axial gyroscope on the SHIMMER platform were compared against those acquired simultaneously, by using the codamotion analysis system from Charnwood Dynamics Ltd., UK.
The gait of one normal healthy adult male (age 25) was measured simultaneously by using two SHIMMER sensors placed on each shank and the Cartesian Optoelectronic Dynamic Anthropometer (CODA) motion analysis system. Data were recorded whilst the subject performed multiple over-ground walking and running trials along a 15-meter walkway in a motion analysis laboratory. In all, ten walking trials at a self selected comfortable walking pace and four running trials at a self selected jogging pace were completed. Heel strike and toe-off points were calculated from the medio-lateral angular velocity derived from the gyroscope signal by using an algorithm reported by Salarian et al.  as in Figure 8.
The heel strike and toe-off characteristic points derived from the SHIMMER and CODA systems were used to calculate the three temporal gait parameters listed below:
- Stride time
- Stance time
- Swing time
Results show an intraclass correlation coefficient (ICC(2,k))  greater than 0.85 in stride, swing, and stance times for ten walking trials and four running trials. These results suggest that the SHIMMER platform is a versatile cost-effective tool for use in temporal gait analysis. Full results from this study are reported elsewhere . Previous studies [66, 67] have validated the CODA system as a reliable platform for gait measurements, so these results are very promising.
The SHIMMER galvanic skin response (GSR) sensor contains an internal resistor network that works as a potential divider and provides a voltage that can be converted by the SHIMMER's ADC to a 12-bit value, used to measure external skin resistance. All skin resistance values were calculated in the SHIMMER platform firmware and transmitted to a BioMOBIUS patch for real-time display and persistence to file. The sensor performance was correlated with a commercial Nexus-10 system (Mind Media BV) utilizing a series of known resistors from 10K Ω to 2.2M Ω. SHIMMER GSR demonstrated an average mean percentage error of 2.3 percent versus the commercial Nexus-10 that had an average mean error of 4.1 percent.