Optical spectroscopy is based on the interaction between electromagnetic radiation and matter. As measured spectra of single substances are characteristic like a fingerprint, they can be used to identify atoms and molecules. Concentrations can be determined by the magnitudes of the individual features of the spectra. However, if the investigated sample is composed of different substances, the individual spectra may overlap and are sometimes hard to differentiate. This often applies to larger molecules where the spectra differ only slightly in their characteristic features. In this case multivariate analysis methods can be instrumental to decompose the superimposed spectra. The regression analysis focuses on the correlation between spectral features and concentrations of individual substances. To increase the prediction accuracy of regression, the data set can be prepared by a feature selection or feature projection, which helps to reduce the influence of noise in the data set by reducing its dimension.

We analyzed photoacoustic spectra of mixtures of different volatile organic compounds (VOCs) in the infrared wavelength region. The spectral features of the single substances are broad and overlap strongly. For trace gases with weak absorption, the features of the photoacoustic spectrum are quasi linearly proportional to the concentration, hence linear methods can be applied in this case. Different combinations of feature selection, feature projection and regression are compared to demonstrate their strengths and weaknesses and to determine the combination with the highest detection selectivity.

Affective Computing is a multidisciplinary field with high potential in many human computer interaction applications including the medical field. One growing application is the emotion and stress recognition for early intervention of depression, stress management, risk prevention as well as monitoring individuals’ mental health. In this context, various modalities ranging from facial, speech, text and biosignal analysis have been adopted for the purpose of emotion and stress recognition. Among these modalities, psychophysiological signals have the valuable advantage as “honest signals”: they cannot be easily triggered by any conscious or intentional control and are continuously available. Biosignals acquired through wearable sensors, add the convenience of mobile implementation in real-life in-the-wild applications. This paper presents an automated processing workflow for the psychophysiological recognition of emotion and stress states. Our proposed workflow allows processing biosignals in their raw state as obtained from wearable sensors. It consists of five stages, for which various Matlab-based methods have been implemented allowing 1) Biosignal Preprocessing: raw data conversion, relevant information selection, artifact and noise filtering, sliding window decomposition, 2) Feature Extraction: from different mathematical groups including amplitude, frequency, linearity, stationarity, entropy and variability, 3) Feature Selection: dimension reduction and computation enhancement using forward selection, backward elimination and brute force feature methods, 4) Classification: machine learning using Support Vector Machine, k-Nearest Neighbor and Random Forest algorithms, 5) Evaluation: performance matrix computation based on k-cross, leave-one-subject-out cross and split validations. All workflow stages are integrated into embedded functions allowing an automated execution of the recognition process. The next steps include further development of the algorithms and the integration of the various tools into an easy-to-use system with graphical interface, satisfying the needs of medical and psychological staff. Our automated workflow was evaluated using our Multimodal Affective Corpus (the uulmMAC Database) previously published for Affective Computing in Human-Computer Interaction.

The smallest concentrations of heavy metal ions can be harmful to both the environment and human health. They are non-biodegradable and can accumulate all along the food chain, thus their onsite monitoring and removal is of great importance. In this work, a novel material based on (3-aminopropyl)triethoxysilane (APTES) coated iron oxide (Fe₃O₄) nanoparticles functionalized with guanine hydrazide (GH) was elaborated. Fourier transform infrared spectroscopy, energy-dispersive X-ray analysis and X-ray diffraction were used to control the synthesis and functionalization steps of the nanoparticles. The morphology and particle size were studied by scanning electron microscopy. Spherical nanoparticles with an average diameter of 45 nm were obtained. A boron-doped diamond electrode coated with GH-APTES-Fe₃O₄ nanoparticles was used to evaluate the electrochemical interaction of some divalent heavy metal ions with guanine hydrazide. Adsorption isotherms were investigated electrochemically and it was shown that the adsorption capacity of the nanoparticles towards heavy metals decreased in the following order: Cu²⁺>Pb²⁺>Cd²⁺. Moreover, the signals generated by square wave voltammetry exhibited two distinct linear response ranges; the first linear plot lies in the range of 0.209 to 1.03 μM with a sensitivity of 171.6 µA/µM for Cu (II), 0.232 to 0.809 μM with a sensitivity of 156 µA/µM for Pb (II) and 0.483 to 4.97 μM with a sensitivity of 101.4 µA/µM for Cd (II). Furthermore, an excellent reproducibility was achieved with relative standard deviation (RSD) values of 4%, 5% and 10% respectively over five independent measurements.

Early detection of arrhythmia is very important. Recently, a wearable device is used to monitor the patient’s heartbeat to detect arrhythmia. However, there are not satisfying algorithms for real-time monitoring arrhythmia in a wearable device. In this work, A novel Fast and Simple Arrhythmia detection algorithm based on YOLO is proposed. The algorithm can detect each heartbeat on long duration ECG signals without R-peak detection and can classify arrhythmia simultaneously. The model replaces the 2D CNN with 1D CNN and a bounding box with a bounding window to utilize Raw ECG signal. Results demonstrate that the proposed algorithm has high performance on speed and mAP in detecting Arrhythmia. Furthermore, the bounding window can predict different window lengths on different types of arrhythmia. Therefore, The model can choose optimal heartbeat window length for Arrhythmia classification. Since the proposed model is a compact 1D CNN model based on YOLO, it can be used in a wearable device and embedded system.

This paper proposes a new readout method for a sensor to detect minute variations into the capacitance. A sensor is based on the weakly coupled electrical resonators that use an amplitude ratio (AR) as an output signal. A new readout scheme with a relatively higher output sensitivity is proposed to measure the relative changes in the input capacitor. A mathematical model is derived to express the readout output as a function of change into the capacitance. To validate the theoretical model, a system is modeled and designed using an industry-standard electronic circuit design environment. A SPICE simulation results are presented for a wide range of design parameters, such as varying coupling factors between the two electrical resonators. A sensitivity comparison between the existing and the proposed AR readout is presented to show the effectiveness of the method of detection proposed in this work.

Impact factor as efficiency or sustainability are entirely correlated to the continuous development of the smart city concept technology application. Intelligent transportation systems (ITS) and particularly autonomous vehicles are expected to play an important role in this challenging environment. Fast and secure connections will be pivotal in order to achieve this new vehicular communications’ application era. The use of millimeter-wave (mmWave) frequency range is the most promising approach to allow these real-time high-demand applications which requires higher bandwidth with the minimum possible latency. However, an in-depth mmWave's channel characterization of the environment is required for a proper mmWave based solution deployment. In this work, a complete radio wave propagation channel characterization for a mmWave V2X smart parking solution deployment in a complex outdoor environment has been assessed, at 28 GHz frequency band. The considered scenario is a parking lot placed in an open free university campus area surrounded by inhomogeneous vegetation. Vehicle and vegetation density within the scenario, in terms of inherent transceivers density and communication impairments, leads to overall system operation challenges, given by multiple communication links operation at LOS and NLOS conditions. By means of an in-house developed 3D ray launching (3D-RL) algorithm, the impact of variable vehicle and vegetation density are addressed, providing coverage/capacity analysis and precise modelling estimations of small-scale and large-scale multipath propagation effects in terms of received power levels and time domain characteristics. The obtained results along with the proposed simulation methodology, can aid in an adequate characterization of the mmWave communication channel for new vehicular communications networks, applications and deployments, considering outdoor conditions as well as the impact of different vehicles and vegetation densities, for current as well as for future wireless technologies.

Molecularly imprinted polymers (MIPs) are synthetic receptors with complimentary cavities towards a chosen template molecule (the target analyte), able to rebind it with high affinity and specificity. Those interactions are similar to the ones between the antibodies and antigens, but with superior chemical, mechanical, thermal and pH stability, and reusability. Thus, with the goal of obtaining of low-cost artificial receptors with high selectivity towards a desired analyte, highly suitable for applications in diverse many fields, such as the environmental, medical or agro-alimentary. In this regard, MIPs have become a significant research hotspot in the development of electrochemical sensors given their ability to selectively rebind to the target analytes with high specificity even in the presence of complex matrix; thus simplifying the analysis process and improving chemosensors performance.

In this work, MIP films are in-situ electro-synthesized from a monomer (pyrrole) solution, in the presence of the template molecule and different doping anions as a facile approach for the tuneability of the MIP morphology. A systematic evaluation on the effect of a series of anions as counter ion dopant integrated into the polypyrrole (PPy) backbone was carried out, including perchlorate (ClO₄^-), p-toluene sulfonate (pTS^-), dodecyl sulfonate (DS^-) and dodecyl benzene sulfonate (DBS^-). The target compounds being evaluated were acetaminophen (AP), uric acid (UA) and ascorbic acid (AA), and the performance of the resulting MIPs modified electrodes was evaluated by means of cyclic voltammetry (CV) and differential pulse voltammetry (DPV). Finally, combination of the different MIP modified electrodes as well as the NIP (non-imprinted polymer) into a sensor array will be evaluated to carry out the analysis of mixtures of the above-mentioned compounds, with the aid of chemometric methods such as principal component analysis (PCA) and artificial neural networks (ANNs).

The photoplethysmogram (PPG) signal is widely measured by smart watches and fitness bands for heart rate monitoring. New applications of the PPG are also emerging, such as to detect irregular heart rhythms, identify sleep disturbances, and assess stress. PPG signal processing algorithms must be robust to variable signal quality due to motion artifact and poor sensor contact. Consequently, datasets of PPG signals acquired in daily life are valuable for algorithm development. The aim of this pilot study was to assess the feasibility of acquiring PPG data in daily life.

A single subject was asked to wear a wrist-worn PPG sensor (SmartCare Analytics) for as much time as possible, for six days a week, for four weeks. They kept a diary of daily activities and any sensor maintenance or troubleshooting required. PPG data were transmitted to a smartphone for storage.

The sensor was worn for approximately three quarters of the time, and was primarily removed for charging, as well as activities involving water, and having a break or forgetting the sensor. Whilst wearing the sensor, the most common reason for data loss was Bluetooth disconnection. Signal quality was high for approximately half of the time: it was highest during sleep, followed by sedentary activities, and was particularly low during exercise or activities involving hand movement such as cooking.

In this study it was possible to acquire PPG data during daily living for four weeks. Key lessons were learnt for future studies: the sensor should be waterproof to reduce the need for removal; data should be stored on the device to avoid loss due to disconnections; and data should preferably be acquired during sleep or periods of low activity to maximise quality, although further research should investigate how this would limit utility. The dataset is being made available at https://peterhcharlton.github.io/ppg-diary/ .

INTRODUCTION

Predicting human intentions is a challenging task, which is gaining importance as well as collaborative robots (cobots) use and human-robot interaction, in several contexts, like industrial and clinical. The use of wearable sensors allows gathering data reducing instrumental or motion constraints, whereas machine learning techniques (MLT) can face the limit of small data amounts typical of these contexts. This study aims to compare MLT performance for the subject intention prediction in the illustrative scenario of a reaching movement, using three-dimensional position data gathered from wearable electromagnetic sensors.

MATERIALS AND METHODS

Nine subjects (4 males; mean age: 51[years], range [29;71]) were recruited and asked to perform 3 repetitions of a sitting reaching movement for combinations of direction (left, center, right), quote (high, low), and distance (proximal, distal) [1], touching the goal and returning. An electromagnetic tracking system (Polhemus Fastrak) was used, with four sensors placed on acromion, upper third of humerus, wrist dorsum and manubrium. Data were elaborated in MATLAB environment to predict the goal position using only the information coming from a first sample of the acquired data. Linear Discriminant Analysis (LDA) and Random Forest (RF) algorithms were implemented, testing several sample dimensions.

RESULTS

LDA and RF prediction accuracy is computed and compared with respect to the data sample dimension. Considering a sample equal to 1/10 of the total movement (average time length t: 0.27[s]), LDA presents an accuracy of 81% (Standard Deviation SD 0.044), and RF of 73% (SD 0.012). Increasing the sample at 1/7 (t: 0.37[s]), accuracy rises at 89% (SD 0.034) with LDA and 83% (SD 0.011) with RF.

DISCUSSION AND CONCLUSIONS

Both algorithms achieved good accuracy, which improves as the sample dimension increases. LDA presents better results. Both MLT give encouraging results and could be exploited in a collaborative scenario.

Reference

[1] J. V. G. Robertson et al. (2012), “Influence of the side of brain damage on postural upper-limb control including the scapula in stroke patients”.

Several CMOS imager sensors were proposed to obtain a high dynamic range imager (>100dB). However, as a drawback these imagers implement a large number of transistors per pixel resulting in a low fill factor, high power consumption, and high complexity CMOS image sensors. In this work, a new operation mode for 3T CMOS image sensors is presented for high dynamic range (HDR) applications. The operation mode consists of biasing the conventional reset transistor as an active load to photodiode generating a reference current. The output voltage achieves steady-state when the photocurrent becomes equal to the reference current similar to the inverter operation in the transition region. At a specific bias voltage, the output swing from o to Vdd in a small light intensity range, however, a high dynamic range is achieved using multiple readouts at different bias voltage. For high dynamic range operation, different values of bias voltage can be applied from each one the signal can be captured and then compose a high dynamic range image. Compared to high dynamic range architectures in the literature the proposed CMOS image pixel shows as an advantage high fill-factor and lower complexity. Moreover, the pixel does not operate in integration mode allowing higher speed operation. A prototype was fabricated at 0.35µm CMOS technology. Experimental results show that by applying five different control voltages it is possible to obtain dynamic range about 100dB.