Back and spine-related issues are frequent maladies that most people have or will experience during their lifetime. A common and sensible observation that can be made is regarding the posture of an individual. We present a new approach that combines accelerometer, gyroscope, and magnetometer sensor data in combination with permanent magnets assembled as a wearable device capable of real-time spine posture monitoring. An independent calibration of the device is required for each user. The sensor data is processed by a probabilistic classification algorithm that compares the real-time data with the calibration result, verifying whether the data point lies within regions of confidence defined by a computed threshold. An incorrect posture classification is considered if both accelerometer and magnetometer classify the posture as incorrect. A pilot trial was performed in a single adult test subject. The combination of the magnets and magnetometer greatly improved the posture classification accuracy (89%) over the accuracy obtained when only accelerometer data were used (47%). The validation of this method was based on image analysis.

In this work we propose an artificial neural network (ANN) for magnetic micro cytometry pattern recognition and automated counting. The method is tested for detecting analytes in the 2-3 μm range. The cytometer is composed of a disposable cartridge and an acquisition platform. The disposable cartridge contains microfluidic channels with a 10×100 μm ² cross section on top of a substrate with magnetoresistive (MR) sensors. The custom analog signal chain performs with an integrated noise of 2.99 μV _RMS in a 10 kHz bandwidth. To employ the ANN we synthesize a training data set based on the magnetic-dipole equation and several data set expansion methods. The ANN is tested on an experiment with 2.8 μm magnetic particles (MP) and compared with an improved threshold-based method with reduced false positive. The ANN produces a maximum of 90% detection rate, improving on the 30-50% detection rates of other single sensor methods published in literature.

Magnetic flow cytometers are promising tools to address cell detection and counting in a seamless and compact manner. Such devices integrate magnetic sensors with microfluidics for sample transportation and magnetic particles as reporters. This work presents a signal analysis method developed to address one of the challenges in magnetic flow cytometry: the ability to distinguish from each single detected signature the number of bacteria effectively detected. Comparing the sensor output signal (pulse amplitude vs. pulse width analysis) with simulations we develop a reliable classification methodology for the identification of the number of magnetically labelled bacteria based on a custom Matlab algorithm. The discrimination of the number of bacteria from the sensor event signatures will improve the detection system accuracy.