Exhaled breath analysis presents a promising approach for drug monitoring. While the range of drugs known to undergo pulmonary exhalation remains limited, innovative experimental models are needed to explore this field. This study aimed to develop anex-vivoplatform as a general experimental setup for studying volatile and semi-volatile compounds in exhaled breath, using propofol as a pragmatic validation compound because it can be reliably detected both in exhaled air and in blood under our experimental conditions. A porcine lung model was created using lungs from a commercial slaughter. Each experiment was performed with a single isolated lung in an individual setup. The lungs were perfused and ventilated. Propofol exhalation was validated under various conditions (boluses, infusion rates, blood flow, and ventilation) using a propofol calibrated multi-capillary column-ion mobility spectrometer (MCC-IMS). Blood gas analysis and plasma concentration samples were collected every 20 min to continuously monitor propofol plasma levels, pulmonary respiration, and metabolism. We established a functionalex-vivoplatform using nine commercially slaughtered porcine lungs, enabling extended measurements for up to 13 h. Hematocrit was set to 35% and hemoglobin to approximately 12 g dl-1, with glucose supplementation of 921 mg h-1. Lactate increased by 308% over the perfusion period. The lungs were ventilated in volume-controlled mode (Vt700 ml, RR 14 min-1, PEEP 8 mbar, I:E 1:1) and perfused at a standard blood flow of 1.0 l min-1, and mean pH was 7.31 over the perfusion period. Exhaled Propofol was detected on average 19 min after the first administration using MCC-IMS. Changes in blood flow and minute volume were accompanied by corresponding changes in the time course and magnitude of exhaled propofol. Thisex-vivomodel of perfused and ventilated porcine lungs provides a controlled setting to study the appearance of intravenously administered drugs in exhaled air under defined ventilation and perfusion conditions. The platform enabled prolonged measurements and detection of exhaled propofol signals and may support future screening of candidate drugs for breath-based drug monitoring pending validation across additional compounds and conditions.

Technological solutions might be of great importance for reducing food waste. In the scope of this article, gas sensor systems for assessing the edibility of food have been studied, which can help to avoid food losses by suggesting consumption before spoilage or by separating infected fruits from fresh ones. Several series of measurements with various foodstuffs were conducted to develop methods that enable the identification of possible use cases in which gas sensors could be used to assess food condition as well as methods to calibrate such sensor systems. This paper presents results for oranges as an important target for grocery stores. The fruit headspace was measured by gas sensors, reference data were acquired using human assessment (appearance, odor, edibility) and gas chromatography-mass spectrometry (GC-MS) analysis. Data evaluation shows correlations between the performance of individual sensors for a technical assessment of fruit condition with marker substances identified by GC-MS, e.g., limonene for damaged oranges. Models were derived that are, in general, able to quantify the edibility or to classify defects/mold, but limitations in the applicability/transferability, e.g., between orange varieties, were also identified. With the knowledge gained, important steps could be taken towards an application-oriented setup, and recommendations regarding the sensors used, food trained, and calibration methods applied are derived.

This study presents a novel approach for model-based calibration transfer and drift compensation for metal oxide semiconductor (MOS) gas sensors. The sensors are lab-calibrated and different calibration models for each of the eight volatiles contained in the calibration are trained to allow for an interpretable quantification of individual volatiles in complex mixtures. Calibration transfer and drift compensation are used to compensate for domain shifts that particularly affect the model accuracy. Here, several domain shifts are considered, e.g., sensor-to-sensor variation among different production batches (calibration transfer) or time-related changes in sensor response like poisoning and aging (drift compensation). Such domain shifts can lead to a substantial performance degradation and are critical for reliable field deployment. Since interpretable and robust machine learning algorithms based on feature extraction, feature selection, and regression (FESR) are not inherently capable of model-based calibration transfer and drift compensation, recalibration typically requires time-consuming and labor-intensive laboratory calibration procedures. To address this challenge, a novel approach represents the interpretable FESR machine learning models as a deep neural network (IDNNRep), enabling the application of transfer learning techniques from the field of deep neural networks (DNNs). This allows the reuse of knowledge gained in an initial calibration domain and facilitates model transfer using only a small amount of new calibration data, thereby reducing calibration effort and time. The proposed method is evaluated across multiple gases, including acetone and toluene, for four domain-shift scenarios and compared with FESR models retrained exclusively on data from the new domain and orthogonal signal correction (OSC). The results demonstrate that the proposed approach reduces the root mean square error (RMSE) compared to the initial model, achieving values of 18.0–28.0 ppb (normalized RMSE: 6.3–9.3%) for both gases with only 0.1 of the calibration data, resulting in a reduction of up to 93% compared to the initial calibration model and 89% compared to the OSC. Furthermore, due to the interpretable nature of the underlying FESR structure, the calibration transfer enables additional sensor- and gas-specific insights.

Temperature Cycled Operation (TCO) has been proven to increase both selectivity and sensitivity for metal oxide semiconductor (MOS) gas sensors. However, TCO requires a stable gas atmosphere during the temperature cycle, which has a typical length of a few seconds to a few minutes. In this work a typical temperature cycle with a duration of 144 s with twelve temperature steps is shortened to 12 s by distributing each temperature step to an individual sensor. Combining the data of all sensors recovers the full original cycle. Hence, this method is named Virtual Temperature Cycle (VTC). In this study, a commercially available digital MOS sensor is used and lab calibrated in a gas mixing apparatus with complex gas mixtures. Data obtained from the original temperature cycle as well as from the VTC are evaluated using the Feature Extraction, Selection and Regression (FESR) method. Different generated test scenarios are evaluated by comparing the model performance determined by the root mean squared error (RMSE) of prediction. The results show that distributing a long temperature over several sensors leads to comparable results in RMSE, i.e. 57 ppb to 55 ppb for acetone. Thus, stable same model quality (e.g. acetone with a test RMSE of under 60 ppb) is achieved while reducing the cycle length by more than 90%.

In gas measurements, there is inherent competition between sensors and analytical systems, often made difficult by a lack of understanding of approaches and constraints from both sides. This leads to controversies, sometimes simple to avoid or defuse, but sometimes leading to harsh discussions and resentment of "wrong" concepts from the respective other side. Meanwhile, collaboration between disciplines is often a key to scientific and, in the long term, economic and societal success. This is especially true today, when digitization is the pervasive topic in our society and where "sensors act as a bridge between biology and digitization." This article addresses this conflict for the field of high-performance gas measurement systems, where the two worlds of low-cost sensor systems on the one hand and high-performance analytical instruments on the other are still often at odds.

Food waste is often driven by consumer uncertainty about the spoilage of stored food, especially for cooked meal leftovers where microbial growth is the main concern. We analyzed whether metal oxide semiconductor (MOS) gas sensors placed inside ordinary food containers can monitor the edibility of leftovers, specifically cooked meatballs. Sensors were operated using temperature cycling to enhance selectivity, and cycle-aligned features were extracted. A prior calibration campaign produced information used to map cycle-aligned features into estimated gas concentrations for relevant VOCs. Total viable counts, which represent the growth of total number of spoilage microorganisms, were analyzed on days 0, 5 and 7 to determine the food’s freshness. Both the raw sensor features and the calibration-derived gas concentration estimates were analyzed with principal component analysis (PCA) and evaluated with a leave-one-sensor-out (LOSO) binary classifier for multiple food containers. PCA on the calibrated gas estimates revealed a dominant axis that consistently tracks food degradation over time across various containers. LOSO classification accuracy improved from 81.7% using raw sensor features to 87.8% using calibrated gas concentration estimates. These findings represent a proof of principle that calibrated MOS sensor systems can robustly support in situ edibility assessment for cooked food.

Metal oxide semiconductor gas sensors are used in a wide range of applications, including indoor air quality monitoring, breath analysis, and the control of industrial emissions. To optimize their performance, especially in terms of sensitivity and gas discrimination, these sensors are operated in temperature cycled operations. To achieve reliable results, sensor systems require complex calibration and the development of robust, data-driven models. Especially interpretable machine learning algorithms, with physically meaningful feature selection, allow to build robust models to predict gas concentrations. This paper utilizes a novel approach that represents interpretable machine learning algorithms as deep neural networks to enable model-based calibration transfer through transfer learning. This reduces the calibration effort while maintaining the model's interpretability. In this study, an interpretable model that combines feature extraction, feature selection, and regression to estimate the acetone concentration (3-500 ppb), was transformed into a deep neural network. The novel approach reduces the calibration time to 7.5% of the initial duration, reduces the prediction RMSE by 37.3 ppb (47.9%) relative to a global model and reaches a R-squared value of 92.7%, while enabling interpretation of the calibration transfer.

Exhaled breath analysis presents a promising approach for drug monitoring. While the range of drugs known to undergo pulmonary exhalation remains limited, innovative experimental models are needed to explore this field. This study aimed to develop an ex-vivo platform as a general experimental setup for studying volatile and semi-volatile compounds in exhaled breath, using propofol as a pragmatic validation compound because it can be reliably detected both in exhaled air and in blood under our experimental conditions. A porcine lung model was created using lungs from a commercial slaughter. Each experiment was performed with a single isolated lung in an individual setup. The lungs were perfused and ventilated. Propofol exhalation was validated under various conditions (boluses, infusion rates, blood flow, and ventilation) using a propofol calibrated multi-capillary column-ion mobility spectrometer (MCC-IMS). Blood gas analysis and plasma concentration samples were collected every 20 min to continuously monitor propofol plasma levels, pulmonary respiration, and metabolism. We established a functional ex-vivo platform using nine commercially slaughtered porcine lungs, enabling extended measurements for up to 13 h. Hematocrit was set to 35% and hemoglobin to approximately 12 g dl-1, with glucose supplementation of 921 mg h-1. Lactate increased by 308% over the perfusion period. The lungs were ventilated in volume-controlled mode (Vt 700 ml, RR 14 min-1, PEEP 8 mbar, I:E 1:1) and perfused at a standard blood flow of 1.0 l min-1, and mean pH was 7.31 over the perfusion period. Exhaled Propofol was detected on average 19 min after the first administration using MCC-IMS. Changes in blood flow and minute volume were accompanied by corresponding changes in the time course and magnitude of exhaled propofol. This ex-vivo model of perfused and ventilated porcine lungs provides a controlled setting to study the appearance of intravenously administered drugs in exhaled air under defined ventilation and perfusion conditions. The platform enabled prolonged measurements and detection of exhaled propofol signals and may support future screening of candidate drugs for breath-based drug monitoring pending validation across additional compounds and conditions.

This paper presents the development and characterization of a gas sensor characterization platform for gas chromatography (GC) systems. With fluidic modeling we can accurately calculate the flow ratio between two detectors at different outlet pressures - in the presented case a mass spectrometer (MS) at high vacuum and metal oxide semiconductor (MOS) detector at ambient pressure. We further validate the fluidic model through experimental testing with commercially available restrictions and a separation column. A volatile organic calibration mix (EPA 502/524) is used to characterize a commercially available MOS gas sensor as GC detector. The temperature-dependent behavior of MOS sensors is studied, and the overall performance is compared with a laboratory MS. The results show the potential for these sensors in various applications, highlighting the importance of characterizing and verifying system components to ensure reliable results. The study concludes by demonstrating that MOS sensors can be effectively used as GC detectors, with comparable sensitivity and retention times to the total ion current of a traditional MS, paving the way for more accessible and portable GC solutions.

The Viking Ship Museum Oslo started in 1957 using saturated salt solutions (here calcium nitrate) to stabilise the relative humidity (RH) in display cases. Three decades later, excellent results with low effort and cost were reported. However, many museums have since replaced these solutions with preconditioned silica gels. Their much lower water capacity causing more frequent need for maintenance, shifting equilibrium RH, unreliability as pollutant absorbers, and possible contamination were not considered or unknown at the time. The absorption capacity of saturated magnesium nitrate solutions for corrosive pollutant traces was studied using metal-oxide-semiconductor (MOS) gas sensors. Corrosive carbonyl pollutants (formic and acetic acid, formaldehyde, and acetaldehyde), often emitted from wood, are indeed highly absorbed. Some absorption was also demonstrated for hydrogen sulfide and sulfur dioxide. Practical application tests worldwide in museums have confirmed the advantages of this sustainable and failsafe passive method.

This paper presents a Bayesian inference approach that enables the evaluation of the aleatoric and epistemic uncertainties of metal oxide semiconductor (MOS) gas sensors in temperature-cycled operation (TCO) based on a convolutional neural network and a Markov-Chain-Monte-Carlo method (MCMC). A convolutional neural network (TCOCNN) is first trained on preprocessed sensor signals to predict gas concentrations. Then, MCMC sampling is applied starting from the trained network weights to estimate a posterior distribution over model parameters. This allows uncertainty in predictions to be decomposed into epistemic components, reflecting uncertainty due to limited training data and aleatoric components, arising from measurement. The approach is applied to data from the H2 sensing task of a public dataset. Results show that epistemic uncertainty increases in regions with sparse data, while aleatoric uncertainty dominates overall, likely due to sensor noise or deficiencies in the temperature cycle.

Metal oxide semiconductor gas sensors are used in several applications, such as indoor air quality assessment, breath analysis, and industrial emission monitoring. In order to reach the full potential regarding the sensitivity and selectivity of those sensors for any application, temperature cycled operation, as well as careful calibration in the form of building a data-driven model is necessary. However, the calibration can be costly in terms of time and money as every sensor needs individual calibration. This work demonstrates that deep-learning based calibration transfer can be utilised to reuse the calibration model across different environmental scenarios for indoor air quality monitoring. Specifically, a sensor can be calibrated for new background conditions within six hours with the model achieving almost the same performance (RMSE: 16.4 ppb) as calibrated for 80 hours (RMSE: 14.8 ppb). This shows that it is possible to transfer a calibration between datasets with different gases and concentrations. The great reduction of the calibration time by 90 % or more would also allow in field calibration using reference analytical methods.

Commercially available low-cost metal oxide semiconductor gas sensors run in temperature cycled operation are studied for online monitoring of propofol in an ex vivo perfusion lung model. A lung from a slaughter pig was connected to a heart lung machine and mechanically ventilated. Propofol as an intravenous anesthetic was added into the blood stream. The sensors were able to quantitatively detect exhaled propofol in the very low ppb-range with a strong correlation of up to 0.95 to the blood plasma concentration and a correlation up to 0.92 to the ion mobility spectrometer used as a reference. This demonstrates the potential of gas sensor systems for online drug monitoring.

In the light of sustainability, the nearly forgotten control of relative humidity (RH) in display cases with saturated salt solutions is re-evaluated. Spilling and creeping of salts can be avoided. By choosing magnesium nitrate or potassium carbonate, the temperature dependence of the RH and the emission of harmful gases are negligible. Measurements with metal oxide semiconductor (MOS) gas sensors showed such solutions to be excellent absorbers for corrosive pollutants such as acids and aldehydes, especially the alkaline potassium carbonate (hitherto not used in museums). The practical application of such solutions in display cases is currently being tested in a community science project in a number of museums. This will result in improved guidelines for conservators and it is hoped, in a revival of this passive, failsafe, low-carbon footprint technique. Desde el punto de vista de la sostenibilidad, se reeval & uacute;a el casi olvidado control de la humedad relativa (HR) en las vitrinas con soluciones salinas saturadas. Se pueden evitar derrames y filtraciones de sales. Al elegir nitrato de magnesio o carbonato de potasio, la dependencia de la temperatura relativa de la HR y la emisi & oacute;n de gases nocivos son insignificantes. Las mediciones con sensores de gas semiconductores de & oacute;xido met & aacute;lico (MOS) mostraron que estas soluciones son excelentes absorbentes de contaminantes corrosivos como & aacute;cidos y aldeh & iacute;dos, especialmente el carbonato de potasio alcalino (hasta ahora no utilizado en museos). La aplicaci & oacute;n pr & aacute;ctica de estas soluciones en vitrinas se est & aacute; probando actualmente en un proyecto cient & iacute;fico comunitario en varios museos. Esto dar & aacute; como resultado pautas mejoradas para los conservadores y se espera que resucite esta t & eacute;cnica pasiva, a prueba de fallas y de baja huella de carbono.

This paper presents a novel application of Gas Chromatography-Metal Oxide Semiconductor (GC-MOS) gas sensor systems for the quantitative measurement of propofol in exhaled air. By integrating advanced MOS sensor technology and analytical techniques, the presented GC-MOS-detector demonstrates quantification abilities similar to the GC-MS system. The results promise great potential for miniaturization and shortening of measuring cycles for drug monitoring in e.g. anesthetic applications with propofol. The proposed setup demonstrates great selectivity in a low-cost solution for propofol measurement.