Real-time Sensor Research Articles

Patient-Specific Readmission Prediction and Intervention for Health Care

Hospital readmission is often associated with unfavorable patient outcomes and a large cost of resources. Therefore, preventing avoidable re-hospitalizations is imperative. To target this problem, one important metric that researchers and practitioners strive to reduce is the 30-day hospital readmission rate. In this paper, we introduce a general decision support system that utilizes machine learning (ML) based patientspecific prediction to guide the suggestion of patient intervention program assignment, with the objective of minimizing the readmission cost for hospitals. This work has three major contributions. First, the proposed solution is highly scalable by using PySpark. Second, we outline solution architecture components including (1) data injection (both real-time sensor reading and data at rest), processing, and analysis, and (2) ML model building, evaluation, deployment and scoring. Third, we discuss how the ML prediction results can be taken into account in a decision support system by presenting a rich visualization.

Di-CNN: Domain-Knowledge-Informed Convolutional Neural Network for Manufacturing Quality Prediction.

In manufacturing, convolutional neural networks (CNNs) are widely used on image sensor data for data-driven process monitoring and quality prediction. However, as purely data-driven models, CNNs do not integrate physical measures or practical considerations into the model structure or training procedure. Consequently, CNNs' prediction accuracy can be limited, and model outputs may be hard to interpret practically. This study aims to leverage manufacturing domain knowledge to improve the accuracy and interpretability of CNNs in quality prediction. A novel CNN model, named Di-CNN, was developed that learns from both design-stage information (such as working condition and operational mode) and real-time sensor data, and adaptively weighs these data sources during model training. It exploits domain knowledge to guide model training, thus improving prediction accuracy and model interpretability. A case study on resistance spot welding, a popular lightweight metal-joining process for automotive manufacturing, compared the performance of (1) a Di-CNN with adaptive weights (the proposed model), (2) a Di-CNN without adaptive weights, and (3) a conventional CNN. The quality prediction results were measured with the mean squared error (MSE) over sixfold cross-validation. Model (1) achieved a mean MSE of 6.8866 and a median MSE of 6.1916, Model (2) achieved 13.6171 and 13.1343, and Model (3) achieved 27.2935 and 25.6117, demonstrating the superior performance of the proposed model.

Correlating nanoscale motion and ATP production in healthy and favism erythrocytes: a real-time nanomotion sensor study.

Red blood cells (RBCs) are among the simplest, yet physiologically relevant biological specimens, due to their peculiarities, such as their lack of nucleus and simplified metabolism. Indeed, erythrocytes can be seen as biochemical machines, capable of performing a limited number of metabolic pathways. Along the aging path, the cells' characteristics change as they accumulate oxidative and non-oxidative damages, and their structural and functional properties degrade. In this work, we have studied RBCs and the activation of their ATP-producing metabolism using a real-time nanomotion sensor. This device allowed time-resolved analyses of the activation of this biochemical pathway, measuring the characteristics and the timing of the response at different points of their aging and the differences observed in favism erythrocytes in terms of the cellular reactivity and resilience to aging. Favism is a genetic defect of erythrocytes, which affects their ability to respond to oxidative stresses but that also determines differences in the metabolic and structural characteristic of the cells. Our work shows that RBCs from favism patients exhibit a different response to the forced activation of the ATP synthesis compared to healthy cells. In particular, the favism cells, compared to healthy erythrocytes, show a greater resilience to the aging-related insults which was in good accord with the collected biochemical data on ATP consumption and reload. This surprisingly higher endurance against cell aging can be addressed to a special mechanism of metabolic regulation that permits lower energy consumption in environmental stress conditions.

Close the Optical Sensing Domain Gap by Physics-Grounded Active Stereo Sensor Simulation

In this article, we focus on the simulation of active stereovision depth sensors, which are popular in both academic and industry communities. Inspired by the underlying mechanism of the sensors, we designed a fully physics-grounded simulation pipeline that includes material acquisition, ray-tracing-based infrared (IR) image rendering, IR noise simulation, and depth estimation. The pipeline is able to generate depth maps with material-dependent error patterns similar to a real depth sensor in real time. We conduct real experiments to show that perception algorithms and reinforcement learning policies trained in our simulation platform could transfer well to the real-world test cases without any fine-tuning. Furthermore, due to the high degree of realism of this simulation, our depth sensor simulator can be used as a convenient testbed to evaluate the algorithm performance in the real world, which will largely reduce the human effort in developing robotic algorithms. The entire pipeline has been integrated into the SAPIEN simulator and is open-sourced to promote the research of vision and robotics communities.

QoS-Based Contract Design for Profit Maximization in IoT-Enabled Data Markets

The massive deployment of Internet of Things (IoT) devices, including sensors and actuators, is ushering in smart and connected communities of the future. The massive deployment of Internet of Things (IoT) devices, including sensors and actuators, is ushering in smart and connected communities of the future. The availability of real-time and high-quality sensor data is crucial for various IoT applications, particularly in healthcare, energy, transportation, etc. However, data collection may have to be outsourced to external service providers (SPs) due to cost considerations or lack of specialized equipment. Hence, the data market plays a critical role in such scenarios where SPs have different quality levels of available data, and IoT users have different application-specific data needs. The pairing between data available to the SP and users in the data market requires an effective mechanism design that considers the SPs’ profitability and the quality-of-service (QoS) needs of the users. We develop a generic framework to analyze and enable such interactions efficiently, leveraging tools from contract theory and mechanism design theory. It can enable and empower emerging data sharing paradigms such as Sensing-as-a-Service (SaaS). The contract design creates a pricing structure for on-demand sensing data for IoT users. By considering a continuum of user types, we capture a diverse range of application requirements and propose optimal pricing and allocation rules that ensure QoS provisioning and maximum profitability for the SP. Furthermore, we provide analytical solutions for fixed distributions of user types to analyze the developed approach. For comparison, we consider the benchmark case assuming complete information of the user types and obtain optimal contract solutions. Finally, a case study based on the example of virtual reality application delivered using unmanned aerial vehicles (UAVs) is presented to demonstrate the efficacy of the proposed contract design framework.

Multigateway Designation for Real-Time TSCH Networks Using Spectral Clustering and Centrality

This letter proposes a multi-gateway designation framework to design real-time wireless sensor networks (WSNs) improving traffic schedulability, i.e., meeting the traffic time constraints. To this end, we resort to Spectral Clustering unsupervised learning that allows defining arbitrary k disjoint clusters without knowledge of the nodes physical position. In each cluster we use a centrality metric from social sciences to designate one gateway. This novel combination is applied to a time-synchronized channel-hopping (TSCH) WSN under earliest-deadline-first (EDF) scheduling and shortest-path routing. Simulation results under varying configurations show that our framework is able to produce WSN designs that greatly reduce the worst-case network demand. In a situation with 5 gateways, 99% schedulability can be achieved with 3.5 times more real-time flows than in a random benchmark.

Robust Real-Time AUV Self-Localization Based on Stereo Vision-Inertia

Autonomous underwater vehicles (AUVs) play an important role in deep-sea exploration, in which AUV self-localization is a key component. However, due to poor visibility caused by challenging marine environments, AUVs are often equipped with high-cost and heavy-weight acoustic sensors to accomplish localization tasks. We propose a robust real-time AUV self-localization method based on stereo camera and inertial sensor, which merges point and diagonal features, as well as inertial measurements to overcome the challenges of poor visibility. Our method also includes an underwater loop detection algorithm based on the combination of points and diagonal segments, which can extract effective binary descriptors in low-textured underwater scenarios. Furthermore, we develop an AUV self-localization system based on a real-time, portable, low-cost, and small volume sensor suite. Finally, we test the proposed method in a real underwater environment using our sensor suite, and the experimental results demonstrate the effectiveness of the proposed method under dramatically changing underwater scenarios.

Enabling battery digital twins at the industrial scale

Digital twins are cyber-physical systems that fuse real-time sensor data with models to make accurate, asset-specific predictions and optimal decisions. For batteries, this concept has been applied across length scales, from materials to systems. However, a holistic approach with a strong conceptual and mathematical framework is needed for battery digital twins to achieve their full potential at the industrial scale. Developing a standardized and transparent approach for data sharing between stakeholders that respects confidentiality is essential. Industrial battery digital twins also need principled methods to quantify and propagate uncertainty from sensors and models to predictions. Ensuring retention of physical understanding is important for the identification of “stiff” parameters, which require careful measurement. Combined with uncertainty analysis, this can unlock optimal data-driven sensor selection and placement and improved root-cause analysis. However, better physical modeling and sensing approaches for battery manufacturing and thermal runaway are needed. Furthermore, immutability of data is also necessary for industrial uptake, with digital ledger technology providing new avenues of research. We believe that digital twins could be transformative for the current lithium-ion battery technologies and also as an enabler for emerging new battery technologies, optimizing lifetime and value through asset-specific control. Digital twins are cyber-physical systems that fuse real-time sensor data with models to make accurate, asset-specific predictions and optimal decisions. For batteries, this concept has been applied across length scales, from materials to systems. However, a holistic approach with a strong conceptual and mathematical framework is needed for battery digital twins to achieve their full potential at the industrial scale. Developing a standardized and transparent approach for data sharing between stakeholders that respects confidentiality is essential. Industrial battery digital twins also need principled methods to quantify and propagate uncertainty from sensors and models to predictions. Ensuring retention of physical understanding is important for the identification of “stiff” parameters, which require careful measurement. Combined with uncertainty analysis, this can unlock optimal data-driven sensor selection and placement and improved root-cause analysis. However, better physical modeling and sensing approaches for battery manufacturing and thermal runaway are needed. Furthermore, immutability of data is also necessary for industrial uptake, with digital ledger technology providing new avenues of research. We believe that digital twins could be transformative for the current lithium-ion battery technologies and also as an enabler for emerging new battery technologies, optimizing lifetime and value through asset-specific control.

Semiconducting MOFs on ultraviolet laser-induced graphene with a hierarchical pore architecture for NO2 monitoring

Due to rapid urbanization worldwide, monitoring the concentration of nitrogen dioxide (NO2), which causes cardiovascular and respiratory diseases, has attracted considerable attention. Developing real-time sensors to detect parts-per-billion (ppb)-level NO2 remains challenging due to limited sensitivity, response, and recovery characteristics. Herein, we report a hybrid structure of Cu3HHTP2, 2D semiconducting metal-organic frameworks (MOFs), and laser-induced graphene (LIG) for high-performance NO2 sensing. The unique hierarchical pore architecture of LIG@Cu3HHTP2 promotes mass transport of gas molecules and takes full advantage of the large surface area and porosity of MOFs, enabling highly rapid and sensitive responses to NO2. Consequently, LIG@Cu3HHTP2 shows one of the fastest responses and lowest limit of detection at room temperature compared with state-of-the-art NO2 sensors. Additionally, by employing LIG as a growth platform, flexibility and patterning strategies are achieved, which are the main challenges for MOF-based electronic devices. These results provide key insight into applying MOFtronics as high-performance healthcare devices.

Fabrication of CuO-NP-Doped PVDF Composites Based Electrospun Triboelectric Nanogenerators for Wearable and Biomedical Applications.

A flexible and portable triboelectric nanogenerator (TENG) based on electrospun polyvinylidene fluoride (PVDF) doped with copper oxide (CuO) nanoparticles (NPs, 2, 4, 6, 8, and 10 wt.-% w.r.t. PVDF content) was fabricated. The structural and crystalline properties of the as-prepared PVDF-CuO composite membranes were characterized using SEM, FTIR, and XRD. To fabricate the TENG device, the PVDF-CuO was considered a tribo-negative film and the polyurethane (PU) a counter-positive film. The output voltage of the TENG was analyzed using a custom-made dynamic pressure setup, under a constant load of 1.0 kgf and 1.0 Hz frequency. The neat PVDF/PU showed only 1.7 V, which further increased up to 7.5 V when increasing the CuO contents from 2 to 8 wt.-%. A decrease in output voltage to 3.9 V was observed for 10 wt.-% CuO. Based on the above results, further measurements were carried out using the optimal sample (8 wt.-% CuO). Its output voltage performance was evaluated as a function of varying load (1 to 3 kgf) and frequency (0.1 to 1.0 Hz) conditions. Finally, the optimized device was demonstrated in real-time wearable sensor applications, such as human motion and health-monitoring applications (respiration and heart rate).

Artificial neural network implementation for dissolved organic carbon quantification using fluorescence intensity as a predictor in wastewater treatment plants

Although spectroscopic methods provide a fast and cost-effective means of monitoring dissolved organic carbon (DOC) in natural and engineered water systems, the prediction accuracy of these methods is limited by the complex relationship between optical properties and DOC concentration. In this study, we developed DOC prediction models using multiple linear/log-linear regression and feedforward artificial neural network (ANN) and investigated the effectiveness of spectroscopic properties, such as fluorescence intensity and UV absorption at 254 nm (UV254), as predictors. Optimum predictors were identified based on correlation analysis to construct models using single and multiple predictors. We compared the peak-picking and parallel factor analysis (PARAFAC) methods for selecting appropriate fluorescence wavelengths. Both methods had similar prediction capability (p-values >0.05), suggesting PARAFAC was not necessary for choosing fluorescence predictors. Fluorescence peak T was identified as a more accurate predictor than UV254. Combining UV254 and multiple fluorescence peak intensities as predictors further improved the prediction capability of the models. The ANN models outperformed the linear/log-linear regression models with multiple predictors, achieving higher prediction accuracy (peak-picking: R2 = 0.8978, RMSE = 0.3105 mg/L; PARAFAC: R2 = 0.9079, RMSE = 0.2989 mg/L). These findings suggest the potential to develop a real-time DOC concentration sensor based on optical properties using an ANN for signal processing.

Adequacy of stationary measurements as proxies for residential personal exposure to gaseous and particle air pollutants

People are exposed to myriad of airborne pollutants in their homes. Owing to diverse potential sources of air pollution and human activity patterns, accurate assessment of residential exposures is complex. In this study, we explored the relationship between personal and stationary air pollutant measurements in residences of 37 participants working from home during the heating season. Stationary environmental monitors (SEMs) were located in the bedroom, living room or home office and personal exposure monitors (PEMs) were worn by the participants. SEMs and PEMs included both real-time sensors and passive samplers. During three consecutive weekdays, continuous data were obtained for particle number concentration (size range 0.3–10 μm), carbon dioxide (CO2), and total volatile organic compounds (TVOC), while passive samplers collected integrated measures of 36 volatile organic compounds (VOCs) and semi volatile organic compounds (SVOCs). The personal cloud effect was detected in >80% of the participants for CO2 and >50% participants for PM10. Multiple linear regression analysis showed that a single CO2 monitor placed in the bedroom efficiently represented personal exposure to CO2 (R2 = 0.90) and moderately so for PM10 (R2 = 0.55). Adding a second or third sensor in a residence did not lead to improved exposure estimates for CO2, with only 6–9% improvement for particles. Selecting data from SEMs when participants were in the same room improved personal exposure estimates by 33% for CO2 and 5% for particles. Out of 36 detected VOCs and SVOCs, 13 had at least 50% higher concentrations in personal versus stationary samples. Findings from this study aid improved understanding of the complex dynamics of gaseous and particle pollutants and their sources in residences, and could support the development of refined procedures for residential air quality monitoring and inhalation exposure assessment.

Real-time detection of mercury ions based on vertically grown ReS2 film

Mercury (Hg2+), one of the most dangerous toxins in water, is a heavy metal that causes organ damage from both short-term and chronic exposure. Conventional methods for detecting mercury such as atomic absorption spectrometry or Raman spectroscopy require bulky equipment with complicated procedures. In this work, we fabricated a highly sensitive, real-time thin-film sensor based on vertically aligned rhenium disulfide (ReS2). Its outstanding large surface area and the unique electronic appearance of its layered architecture make a ReS2 nanosheet a strong contender for such an application. The sensor exhibited a fast response speed (< 2 s) to Hg2+ and an ultralow detection limit of 4 nM, which is significantly less than that of the U.S. Environmental Protection Agency's (U.S. EPA) allowed utmost contamination limit for Hg2+ in drinking water (10 nM). It also exhibited strong selectivity for Hg2+ against other metal ions such as Na+, Zn2+, Fe3+, Cu2+, Ca2+, Ni2+, Ag+, Cd2+, Fe2+, and Pb2+. Because this nanosheet can be replaced with any secondary substrate and possibly patterned into a microscale size, the sensor can be integrated into multiple platforms such as portable devices or sensor nodes in a grid network.

A data-driven predictive maintenance model for hospital HVAC system with machine learning

ABSTRACT Corrective and preventive maintenance strategies are typically employed to maintain an efficient functionality of different facility systems. This entails the evaluation of current conditions and the prediction of future conditions. Such prediction is highly needed for critical building systems such as Heating, Ventilation, and Air Conditioning (HVAC) of hospitals to maintain their functionality and extend their lifetime. Current literature highlights the benefits of adopting machine-learning algorithms for predictive modelling. Literature also reveals a gap in predictive modelling based on real-time sensor data and the prediction of both short-term and long-term future conditions. This paper presents a data-driven predictive maintenance model of a hospital’s HVAC system with a focus on the Air Handling Units (AHUs). The developed model adopts machine-learning using the sensor data acquired by the BMS and the database of the hospital’s CMMS. Support Vector Machine (SVM), Decision Trees (DT), and K-Nearest Neighbours (KNN) algorithms are used for the prediction of AHU’s short-term conditions. Prophet Forecasting and Seasonal Auto-Regressive Integrated Moving Average (SARIMA) algorithms are then used to predict the AHU’s long-term future conditions. The study also highlights the benefits of adopting the proposed model in terms of reduced maintenance cost and improved operational effectiveness of hospital AHUs.

2D MOF-enhanced SPR sensing platform: Facile and ultrasensitive detection of Sulfamethazine via supramolecular probe

Sulfamethazine (SMZ) is widely present in the environment and can cause severe allergic reactions and cancer in humans. Accurate and facile monitoring of SMZ is crucial for maintaining environmental safety, ecological balance, and human health. In this work, a real-time and label-free surface plasmon resonance (SPR) sensor was devised using a two-dimensional metal-organic framework with superior photoelectric performance as an SPR sensitizer. The supramolecular probe was incorporated at the sensing interface, allowing for the specific capture of SMZ from other analogous antibiotics through host-guest recognition. The intrinsic mechanism of the specific interaction of the supramolecular probe-SMZ was elucidated through the SPR selectivity test in combination with analysis by density functional theory, including p-π conjugation, size effect, electrostatic interaction, π-π stacking, and hydrophobic interaction. This method facilitates a facile and ultrasensitive detection of SMZ with a limit of detection of 75.54 pM. The accurate detection of SMZ in six environmental samples demonstrates the potential practical application of the sensor. Leveraging the specific recognition of supramolecular probes, this direct and simple approach offers a novel pathway for the development of novel SPR biosensors with outstanding sensitivity.

A novel real-time wireless sensor integration for enhancing positive pressure system operation in single limb passive vented circuit

Non-invasive ventilation has become increasingly popular with recent technological advancements, and with the adaption of turbine-based respiratory support, NIV devices are becoming more accessible. Non-Invasive ventilation consists of a single limb or a dual limb circuit, with the dual limb being the most sophisticated. However, single limb vented circuit being the most widely adopted. The use of proximal to mask flow, pressure, and external physiological sensors are still not widely adapted to the single limb vented circuit, restricting the quality of ventilation support, primarily in-home care scenarios. We present a novel study that implements a real-time wireless pressure and flow sensor control system to a bi-level ventilator for a modular approach towards non-invasive ventilatory support. Tidal Volume, Inspiratory trigger delay, time to minimum pressure, and flow and pressure curves examined present a proximal to patient wireless system that can enhance the performance of a single limb passive vented circuit.

Real-Time Immunosensor for Small-Molecule Monitoring in Industrial Food Processes.

Industrial food processes are monitored to ensure that food is being produced with good quality, yield, and productivity. For developing innovative real-time monitoring and control strategies, real-time sensors are needed that can continuously report chemical and biochemical data of the manufacturing process. Here, we describe a generalizable methodology to develop affinity-based biosensors for the continuous monitoring of small molecules in industrial food processes. Phage-display antibody fragments were developed for the measurement of small molecules, as exemplified with the measurement of glycoalkaloids (GAs) in potato fruit juice. The recombinant antibodies were selected for use in a competition-based biosensor with single-molecule resolution, called biosensing by particle motion, using assay architectures with free particles as well as tethered particles. The resulting sensor measures GAs in the micromolar range, is reversible, has a measurement response time below 5 min, and enables continuous monitoring of GAs in protein-rich solutions for more than 20 h with concentration measurement errors below 15%. The demonstrated biosensor gives the perspective to enable a variety of monitoring and control strategies based on continuous measurement of small molecules in industrial food processes.

112 Comparing Various Size Fractions Collected Using Dusttrak And Gravimetric Samples

Abstract Occupational exposure concentrations are routinely collected using gravimetric samples. However, gravimetric samples typically require a long sampling time, especially when there are low exposure concentrations. While long sampling time may be appropriate to verify compliance with 8-h occupational limit values (OELs), they cannot address exposure's temporal and spatial heterogeneity or provide information on peak exposures present in the work environment. Real-time light scattering sensors enable the possibility of collecting high-resolution, low-cost measurements, quickly identifying peak and short-term concentrations, linking exposure to emission sources, and better understanding within-day variation. These technologies have become increasingly popular also for occupational monitoring purposes. In this presentation, the results of the dust samples collected for a randomized double-blinded controlled human exposure chamber study are presented. In six groups, 24 healthy volunteers were exposed to two common stone minerals, quartz diorite, and rhomb porphyry, in 4-h exposure sessions. To control the exposure in the chamber, personal gravimetric samples of respirable dust and stationary gravimetric samples of total dust, PM2.5, and respiratory dust were collected. Additionally, stationary real-time samples of total dust, respirable dust, PM10, PM2.5, and PM1 were collected continuously using DustTrak TSI model DRX 8533. Despite the many advantages of the real-time sensors, evaluating compliance with OELs should be done with caution. Except for the PM2.5 fraction, a significant difference was observed between the gravimetric and the DustTrak samples. While the DustTrak overestimated the PM2.5 concentrations, the total dust concatenations were underestimated by a factor of almost two compared to the gravimetric samples.

Label free impedance based acetylcholinesterase enzymatic biosensors for the detection of acetylcholine

Realtime monitoring of neurotransmitters is of great interest for understanding their fundamental role in a wide range of biological processes in the central and peripheral nervous system, as well as their role, in several degenerative brain diseases. The measurement of acetylcholine in the brain is particularly challenging due to the complex environment of the brain and the low concentration and short lifetime of acetylcholine. In this paper, we demonstrated a novel, label-free biosensor for the detection of Ach using a single enzyme, acetylcholinesterase (ACHE), and electrochemical impedance spectroscopy (EIS). Acetylcholinesterase was covalently immobilized onto the surface of gold microelectrodes through an amine-reactive crosslinker dithiobis(succinimidyl propionate) (DSP). Passivation of the gold electrode with SuperBlock eliminated or reduced any non-specific response to other major interfering neurotransmitter molecules such as dopamine (DA), norepinephrine (NE) and epinephrine (EH). The sensors were able to detect acetylcholine over a wide concentration range (5.5–550 μM) in sample volumes as small as 300 μL by applying a 10 mV AC voltage at a frequency of 500 Hz. The sensors showed a linear relationship between Ach concentration and ΔZmod(R2 = 0.99) in PBS. The sensor responded to acetylcholine not only when evaluated in a simple buffer (PBS buffer) but in several more complex environments such as rat brain slurry and rat whole blood. The sensor remained responsive to acetylcholine after being implanted ex vivo in rat brain tissue. These results bode well for the future application of these novel sensors for real time in vivo monitoring of acetylcholine.

Real-Time Intent Sensing for Assistive Devices with Implications for Minimising Maintenance

Intent sensing is a growing field within medical device control, with major potential applications for assistive devices, such as prosthetics. As many sensors as possible should be utilised to maximise accuracy. The availability of sensors may change over time due to changing surroundings or activities, sensors failing, and electrode contact being lost. The sensor network should be dynamic and modular in nature, continuing to function even when some sensors are unavailable. The management of sensor unavailability may help to reduce the need for device maintenance, particularly in developing nations with limited availability of these services. An algorithm is proposed to classify intent using networked sensors in real time. Data are gathered using human participants wearing four surface electromyography sensors and performing a pseudo-random sequence of grasps. The relationship between time offset and prediction accuracy is investigated, with the algorithm predicting future intent actions up to half a second in advance. Sensor dropout is simulated by randomly replacing sensor readings with recorded noise. The new algorithm is compared to existing algorithms and shown to be more accurate in situations of sensor dropout, with the difference increasing as more sensors become unavailable. This suggests that when reductions in sensing capabilities are likely to occur over time, the modular method is more appropriate for control.