Abstract: Continuous monitoring of acetone is a challenge using related art sensing methods. Though real-time detection of acetone from different biofluids is promising, signal interference from other biomarkers remains an issue. A minor fluctuation of the signals in the micro-ampere range can cause substantial overlapping in linear/polynomial calibration fittings. To address the above in non-invasive detection, principal component analysis (PCA) can be used to generate specific patterns for different concentration points of acetone in the subspace. This results in improvement of the problem of overlapping of the signals between two different concentration points of the data sets while eliminating dimensionality and redundancy of data variables. An algorithm following PCA can be incorporated in a microcontroller of a sensor, resulting in a functional wearable acetone sensor. Acetone in the physiological range (0.5 ppm to 4 ppm) can be detected with such a sensor.

Abstract: Continuous monitoring of acetone is a challenge using related art sensing methods. Though real-time detection of acetone from different biofluids is promising, signal interference from other biomarkers remains an issue. A minor fluctuation of the signals in the micro-ampere range can cause substantial overlapping in linear/polynomial calibration fittings. To address the above in non-invasive detection, principal component analysis (PCA) can be used to generate specific patterns for different concentration points of acetone in the subspace. This results in improvement of the problem of overlapping of the signals between two different concentration points of the data sets while eliminating dimensionality and redundancy of data variables. An algorithm following PCA can be incorporated in a microcontroller of a sensor, resulting in a functional wearable acetone sensor. Acetone in the physiological range (0.5 ppm to 4 ppm) can be detected with such a sensor.