Fig. 2: The “Raw” data dilemma: defining sample-level data in the data supply chain in a uniaxial MEMS accelerometer. | npj Digital Medicine

Fig. 2: The “Raw” data dilemma: defining sample-level data in the data supply chain in a uniaxial MEMS accelerometer.

From: Verification, analytical validation, and clinical validation (V3): the foundation of determining fit-for-purpose for Biometric Monitoring Technologies (BioMeTs)

Fig. 2: The “Raw” data dilemma: defining sample-level data in the data supply chain in a uniaxial MEMS accelerometer.

Acceleration results in physical motion of the equivalence of a spring and proof mass, which in turn results in changes of electrical properties that can be captured by electrical property sensors. Electrical signals are then converted from analog to digital signals and stored and transmitted via the microprocessor on a wristband or mobile device. Through BLE, data are then processed and compressed multiple times for transmission and storage through mobile devices or cloud storage. This figure summarizes the steps of data collection and manipulation into a daily step count metric and illustrates that “raw” data could refer to different stages of the data collection and manipulation process and have different meanings. For more details of the data types and technologies involved in each step, please refer to Supplementary Table 2. Here, two arrows are highlighted with asterisks, which signify steps in the data supply chain where the “raw data dilemma” usually occurs. What is defined and clarified as “sample-level data” are the primary and processed digital signals marked by asterisks.

Back to article page