Electronic noses (e-noses) are devices based on combining different gas sensors’ responses to a given sample for identifying specific odor fingerprints. In recent years, this technology has been considered a promising novel tool in several fields of application, but several issues still hamper its widespread use. This review paper describes how some physical confounding factors, such as temperature, humidity, and gas flow, in terms of flow direction and flow rate, can drastically influence gas sensors’ responses and, consequently, e-nose results. Among the software and hardware approaches adopted to address such issues, different hardware compensation strategies proposed in the literature were critically analyzed. Solutions related to e-nose sensors’ modification, design and readout, sampling system and/or chamber geometry design were investigated. A trade-off between the loss of volatile compounds of interest, the decrease of sensors’ sensitivity, and the lack of fast responses need to be pointed out. The existing body of knowledge suggests that the e-nose design needs to be highly tailored to the target application to exploit the technology potentialities fully and highlights the need for further studies comparing the several solutions proposed as a starting point for the application-driven design of e-nose-based systems.

Background: Non-invasive, bedside diagnostic tools are extremely important for tailo ring the management of respiratory failure patients. The use of electronic noses (ENs) for exhaled breath analysis has the potential to provide useful information for phenotyping different respiratory disorders and improving diagnosis, but their application in respiratory failure patients remains a challenge. We developed a novel measurement apparatus for analysing exhaled breath in such patients. Methods: The breath sampling apparatus uses hospital medical air and oxygen pipeline systems to control the fraction of inspired oxygen and prevent contamination of exhaled gas from ambient Volatile Organic Compounds (VOCs) It is designed to minimise the dead space and respiratory load imposed on patients. Breath odour fingerprints were assessed using a commercial EN with custom MOX sensors. We carried out a feasibility study on 33 SARS-CoV-2 patients (25 with respiratory failure and 8 asymptomatic) and 22 controls to gather data on tolerability and for a preliminary assessment of sensitivity and specificity. The most significant features for the discrimination between breath-odour fingerprints from respiratory failure patients and controls were identified using the Boruta algorithm and then implemented in the development of a support vector machine (SVM) classification model. Results: The novel sampling system was well-tolerated by all patients. The SVM differentiated between respiratory failure patients and controls with an accuracy of 0.81 (area under the ROC curve) and a sensitivity and specificity of 0.920 and 0.682, respectively. The selected features were significantly different in SARS-CoV-2 patients with respiratory failure versus controls and asymptomatic SARS-CoV-2 patients (p < 0.001 and 0.046, respectively). Conclusions: the developed system is suitable for the collection of exhaled breath samples from respiratory failure patients. Our preliminary results suggest that breath-odour fingerprints may be sensitive markers of lung disease severity and aetiology.

Electronic noses (E-Noses) equipped with metal-oxide semiconductor (MOS) sensors are promising tools for non-invasive medical diagnostics. Their adoption in clinical practice, however, is limited—among others—by sensor variability across devices, which makes individual calibration necessary. This study presents an approach for the development of a calibration transfer (CT) methodology for urine headspace analysis, involving the design and realization of a dedicated experimental setup and protocol. Partial least squares-discriminant analysis (PLS-DA) models were trained on human urine samples enriched with selected biomarkers to simulate pathological states. Models from a reference (“master”) device were transferred to other (“slave”) units in multiple master–slave configurations using Direct Standardization (DS). To overcome the variability of human urine, synthetic urine recipes were formulated to mimic sensor responses and serve as reproducible transfer samples. Several strategies for selecting transfer samples were evaluated, including the Kennard–Stone algorithm, a DBSCAN-based approach, and random selection. Without CT, classification accuracy on slave devices decreased markedly (37–55%) compared to the master’s performance (79%), whereas applying DS with synthetic standards restored accuracy to 75–80%. These results demonstrate that combining reproducible synthetic standards with DS enables effective model transfer across E-Noses, reducing calibration requirements and supporting their broader applicability in medical diagnostics.

Tracheal occlusion (TO) is a fetal treatment option for congenital diaphragmatic hernia. It aims to reduce lung hypoplasia and hypertension by occluding the trachea with a subsequent increase in pulmonary pressure. However, how TO affects lung pressure is unknown, and no devices are available for its measurement. We developed an implantable device for real-time wireless monitoring and long-term logging of lung pressure in ovine fetuses. The device comprised a microcontroller unit, two pressure sensors, an inertial measurement unit (IMU), a radio frequency unit, and an antenna. In vitro and in vivo tests were performed to evaluate device performances. A diaphragmatic hernia was created in ovine fetuses via hysterotomy. Nine devices were implanted at 100 days of gestation and remained until delivery (~142 d). One pressure sensor measured pressure in the amniotic sac and the other in the fetal trachea. In vitro pressure accuracy was median (interquartile range) 0.31% (0.1%). In vitro device lifetime was ~60 days. The in vivo device recording time was 26.0 (3.9) days and it was well tolerated by the fetus and the ewe. Data telemetry operated throughout the device lifetime. At delivery, amniotic fluid, device encapsulation, and lamb skin did not present evidence of damages. Pressure data in four lambs were analyzed. Fast pressure changes due to the spontaneous fetal activity and long-term mean pressure changes resulting from TO were detected. The IMU allowed for the assessment of changes in fetal position. This device allows long-term monitoring of mean lung pressure, spontaneous breathing activity, and movements in physiological and pathological ovine fetal models.

Sepsis is a severe systemic condition due to an extreme response of the body to an infection. It is responsible for a significant number of deaths worldwide, and is still difficult to diagnose early. In this study, a system was developed for exhaled breath sampling in mechanically ventilated patients at the intensive care unit (ICU), together with a custom-made electronic nose (e-Nose) device for detecting sepsis in exhaled breath. The diagnostic performance of this system was evaluated in an animal sepsis model. Ten pigs (LPS group) were administered lipopolysaccharide (LPS) to induce a systemic inflammatory response. Nine other pigs received a placebo solution (control group). Exhaled breath samples were collected in Nalophan^TM bags and stored for temperature and humidity equilibration before e-Nose analysis. Measurements were corrected for the effects of different fractions of inspired oxygen (FiO₂) on e-Nose sensors. Two classification models using e-Nose and physiological measurements were developed and compared. One hour after LPS administration, the e-Nose data model with FiO₂ correction showed a higher accuracy (76.2% (95% confidence interval (CI) [58.0, 94.2])) than the physiological data model (59.0% (95% CI [39.5, 79.5])), indicating the potential of the early detection of sepsis with an e-Nose.

This study assesses the agreement in terms of linearity errors between simulated and experimental data from MOX sensors operated in dielectric excitation mode. Both simulated and experimental reactance spectra exhibit high linearity with respect to gas concentration in the high-frequency shoulder of the relaxation peak. The results demonstrate strong agreement between simulated and experimental 95% CI of absolute linearity errors as a function of frequency. As expected, the empirical errors are slightly bigger compared to the simulation prediction, since the latter only considers the linearity errors. The good correspondence between simulations and empirical results supports the use of simulation to optimize the tuning capacitor and the selection of the optimal operating frequency.