V3 Validation

The Verisense IMU has been validated following the V3 framework championed by the Digital Medicine Society (DiMe).

Following the V3 framework is a rigorous process. The initial verification stage involves evaluating and demonstrating the performance of the sensor technology, and the sample-level data it generates, against a pre-specified set of criteria. During the analytical validation stage, the performance of the algorithm is evaluated, together with its ability to measure, detect, or predict physiological or behavioral metrics. Then, in the clinical validation stage, the biometric monitoring technology is evaluated to determine whether it acceptably identifies, measures, or predicts a meaningful clinical, biological, physical, functional state or experience, in the stated context of use (which includes a specified population).

The first version of a white paper detailing the V3 validation of the Verisense IMU can be found here.

This paper is designed not only to validate Verisense but to help provide a practical example of employing the V3 framework. Shimmer is excited to help this process forward. We chose the white paper format because the amount of validation material is growing so rapidly that we anticipate releasing new versions frequently.

Metrics

Worn on the wrist, the Verisense IMU provides:

• Raw data: 3-axis accelerometer, 3-axis gyroscope

• Time spent in sedentary, light, moderate, or vigorous activity

• Sleep metrics: total time in bed and sleep efficiency

• Non-wear detection: total time device was not worn

The primary deliverable of the Verisense platform is continuous raw data from all participants.

Accelerometer

• Sample Rate: 12.5Hz, 25Hz, 50Hz, 100Hz, 200Hz, 1600Hz.

• Range: +/-2g, +/-4g, +/-8g, +/-16g

Gyroscope

• Sample Rate: 12.5Hz, 26Hz, 52Hz, 104Hz, 208Hz, 1666Hz, 3332Hz.

• Range: +/-125dps, +/-250dps, +/-500dps, +/-1000dps, +/-2000dps

Activity and sleep metrics are calculated using the GGIR algorithm.

Validation

Verisense is an open system providing raw data and it utilizes GGIR – an open source analysis package used in over 150 peer reviewed papers.

The wrist-based activity and sleep algorithm, GGIR, has been scientifically validated and used in over 150 peer reviewed publications. A wide variety of patient populations have been investigated including stroke, dementia, obesity, cardiovascular disease and muscular dystrophy to name a few. We have developed a guide to break down those publications by year, population and disease area.

For more validation, please see a list of publications using the Shimmer sensing technology.