This paper presents the development of a novel miniature electrical impedance tomography (EIT) system made out of glass, along with the training, validation, and testing of an accompanying open-source machine learning image reconstruction model. Our 1-dimensional convolutional neural network (1D-CNN) models were uniquely benchmarked, both qualitatively and quantitatively, using synthetic and experimental data, against well-established image reconstruction methods: the one-step Gauss-Newton method and the total variation reconstruction method. Image reconstruction results obtained using our 1D-CNN show significant advantages over these traditional methods, achieving an up to 5-fold reduction in mean square error on synthetic data. These results were replicated for two common excitation/measurement modes and extended to objects with varying conductivity and quantities. The superior EIT reconstruction capabilities of our 1D-CNN were further validated experimentally across a similarly broad range of parameters, achieving an average positional accuracy of 147 μm and an average dimensional resolution of 70 μm. To demonstrate potential applications in in vitro monitoring, we used our platform to observe zebrafish development through three distinct phases, from embryo to larvae, showcasing our platform's compatibility with biological imaging.

Wearable bifunctional sensors demonstrate considerable potential for implementation in early warning systems targeting febrile convulsions. However, significant challenges persist in simultaneously achieving real-time and precise detection of dual sensing modalities with accurate signal discrimination for clinical diagnosis. Here, we have fabricated hollow-microstructure Ti3C2Tx/poly(benzodifurandione) composite fibers for synergistic optimization of humidity-pressure sensing performance through rational microstructure engineering of the sensitive layer coupled with the dynamic reconfiguration of conductive pathways. The bifunctional sensor provides high sensitivity (168.72 kPa-1), a low detection limit (0.67 Pa) for pressure sensing, rapid response/recovery time (3 s/20 s), and low humidity hysteresis (2.18% RH) for humidity sensing. Integrated with a control circuit, a febrile convulsion early alarm system is established for real-time monitoring of muscle activities and breathing rate by sending an early alarm to the mobile terminal when in abnormal conditions. This work develops a new avenue for prompt treatment of intelligent healthcare.

Thiamphenicol (TAP) is primarily used to treat bacterial infections in animals, but its residues lead to antimicrobial resistance and public health risks. This study constructed an aggregation-induced electrochemiluminescence sensor (AIECLS) by embedding iridium nanoflowers into graphene aerogel (GA) through electrostatic interactions, combined with bifunctional monomers molecularly imprinted polymers (MIPs), achieving sensitive and targeted detection of TAP. Specifically, hydrophilic iridium nanoflowers were synthesized by surfactant-assisted reprecipitation, where the cationic surfactant cetyltrimethylammonium bromide (CTAB) directed self-assembly of iridium complexes and endowed them with hydrophilicity. The long alkyl chains of CTAB facilitated the formation of iridium nanoflower aggregates, restricting movement of luminophores and reducing nonradiative energy loss. The 3D porous structure and excellent conductivity of GA provided an ideal carrier for AIE molecules and a multifunctional microreactor. Its porous structure effectively confined and enriched luminophores, while the 3D conductive network of GA accelerated electron transfer kinetics. Moreover, the AIECL enhancement effect of CTAB-Ir caused by the electrostatic interaction synergistically amplified the electrochemical luminescence intensity. The synthesized MIPs enhanced both binding capacity and affinity toward TAP through synergistic interactions of multiple complementary functional groups, achieving specific detection of TAP. The linear range of the established sensor was 1.00 × 10-9-5.00 × 10-5 mol L-1 with a limit of detection of 1.85 × 10-10 mol L-1, and TAP recoveries in spiked experiments were 87.02-98.06%, which were consistent with high-performance liquid chromatography results. This study provides new insights into the design of AIECLS using iridium-based materials and confinement enhancement strategies while establishing a new method for monitoring hazardous substances in food.

The development of next-generation wearable electronic nose (e-nose) systems for real-time environmental monitoring requires miniaturized gas sensor arrays with high sensitivity and low-power operation. Current limitations persist in the incompatibility between conventional sensing material deposition methods and MEMS microheater architectures. Here, we present an intelligent wristwatch-formatted e-nose system, integrating a printable quantum dot (QD) sensor array fabricated using an optimized colloidal quantum dot (CQD) ink formulation and a precision inkjet printing strategy. We engineered metal cation-surrounded quantum dots (MCSQDs) via liquid-phase ligand exchange with transition metal chlorides (FeCl3, CoCl2, NiCl2, CuCl2), achieving tailored surface functionalities and enhanced gas discrimination capabilities. The engineered MCSQD inks demonstrated exceptional colloidal stability and seamless MEMS microheater integration, enabling gas sensors with parts-per-billion-level detection limits (4 ppb ethanol). A 16-unit sensor array was embedded into a wearable platform incorporating cloud-based neural network processing. System validation achieved 100% classification accuracy in indoor odor recognition tests using a fully connected neural network (FCNN), while field tests at a transportation hub demonstrated reliable monitoring of Total Volatile Organic Compounds (TVOC), NO2, SO2, and CO with <15% deviation from the reference sensors. This work establishes a viable manufacturing framework bridging quantum-confined material engineering to IoT-enabled artificial olfaction, paving the way for scalable production of QD gas sensor array-based e-noses.

High-sensitivity viral diagnostics typically use PCR to detect and amplify viral nucleic acids which requires fluorescence reporters, enzymatic amplification, specialized equipment and can be time-consuming. In this work, we describe fuel-free (FF) Rolosense, a diagnostic approach that leverages mechanical force sensing as a fundamental transduction mechanism. We use the Brownian motion of aptamer-coated microparticles on an aptamer-modified surface for viral detection. The microparticles function as both the sensing and transduction elements, reporting specific molecular interactions where the presence of viral particles stalls their motion by cross-linking them to the surface. FF-Rolosense harnesses biased motion and thermal fluctuations to achieve rapid, sensitive, and specific detection of intact virions─the active agents of infection. This approach represents a fundamental shift from conventional diagnostic methods and demonstrates a limit of detection as low as 103 copies/mL for SARS-CoV-2 variants, including BA.1 and BA.5, and effectively differentiates SARS-CoV-2 from other viral pathogens such as Influenza A, HCoV OC43, and 229E. We also show that FF-Rolosense readout is amenable to deep learning analysis revealing single particle viral binding events. Finally, we demonstrate potential for point-of-care and home-based applications by using a 3D-printed brightfield microscope, Roloscope, for FF-Rolosense readout. Taken together, this work shows a complementary strategy for viral diagnostics that employs a mechanical mechanism of transduction.

Surface plasmon resonance microscopy (SPRM) is widely used for label-free imaging of biomolecular interactions at cell membranes. In cell-based assays, SPRM often produces distinctive patterns, such as edge-dominated signals and spatially heterogeneous binding kinetics, yet the physical origins of these features remain incompletely understood and cannot be resolved by SPRM alone. Here, we combine SPRM with depth- and time-resolved confocal fluorescence imaging to address this question. Using lectin binding on fixed cells as a model system, we show that SPRM signals primarily originate from ligand binding within the portion of the membrane exposed to the evanescent field, with additional contributions from binding-induced membrane deformation. We also find that local membrane features strongly correlate with kinetic heterogeneity. These results provide direct experimental evidence clarifying SPRM signal formation in cell-based measurements and offer practical guidance for accurate quantification and interpretation in label-free plasmonic imaging of cellular interfaces.

Surface-enhanced Raman scattering (SERS)-based pH sensors have been widely applied; However, the used 4-mercaptobenzoic acid (4-MBA) probes exhibit small pH-sensitive peak changes (carboxyl group) and strong susceptibility to interference, leading to inaccurate measurements. To address these limitations, we developed a SERS pH sensor using 2,5-dimercaptoterephthalic acid (2,5-DMTA) as the probe, which contains dual carboxyl groups. These carboxyl groups undergo reversible protonation-deprotonation, producing pronounced and reproducible spectral responses that enhance detection accuracy. The developed sensor enabled reliable detection across the acidic pH range of 0-7, showing good linearity (R2 = 0.9786) and compensating for the weak acidic response of 4-MBA. Importantly, the 2,5-DMTA-based pH sensor demonstrated much better detection accuracy (detection relative standard deviation RSD less than 5%) than the 4-MBA-based SERS pH sensor (detection RSD ≈ 20%). To further improve measurement accuracy in complex matrices, the sensor was embedded in hot agarose to form an AuNP@hydrogel substrate, effectively suppressing interference from small molecules. Moreover, the developed sensor also shows satisfactory online pH monitoring features, including good reversibility (≥6 cycles), high stability of continuous measurements (30 min), and long-term storage stability (30 days). Integrated with a microfluidic 3D-printed flow cell, the system enabled rapid response (∼120 s) and online pH monitoring, and was successfully applied to continuous testing in lake water. Overall, this SERS platform provides a robust and accurate solution for SERS-based pH detection under acidic and complex environmental conditions.

Frequency-dependent impedance spectroscopy in combination with machine learning offers a powerful strategy for discriminating among gas species using mutually interacting semiconductor metal oxide (SMO) gas sensors. In this study, 0.3 at% platinum-loaded SnO2 sensing materials were employed to breath-based disease detection, with a focus on machine learning-assisted discrimination of mixtures of acetone (0.5-2.5 ppm) and ethanol (0.5-2.5 ppm) under both dry and humid environments (80% relative humidity). Data features derived from the real, imaginary, and magnitude components of complex impedance obtained at the frequency range from 105 to 104 Hz were used to enhance gas discrimination performance through supervised deep learning neural networks (DNNs). Even with a single sensor designed through structural and compositional modifications, frequency-dependent impedance features enabled accurate identification of acetone concentrations in acetone-ethanol mixtures under humid conditions, achieving 99% accuracy using single-frequency impedance data (i.e., 105 Hz), compared to 66% with DC-based (voltage) signals. This innovative strategy offers an effective and scalable solution for detecting not only breath acetone but also gas mixtures composed of chemically similar gas species.

Ratiometric electrochemical immunosensors offer improved reproducibility and accuracy compared with conventional single-signal platforms, yet their use in label-free mycotoxin detection remains limited due to dependence on signal-molecule labels. Herein, we report a label-free ratiometric electrochemical immunosensor for zearalenone (ZEN), chosen as a representative mycotoxin, constructed on a porous screen-printed carbon-carbon nanotube electrode (porous CNT SPE) functionalized with a nickel metal-organic framework/carbon nanotube composite (Ni-MOF/CNT). The MOF was engineered by partially substituting 2-aminoterephthalic acid (NH2-BDC) with 2,3-diaminoterephthalic acid ((NH2)2-BDC), thereby increasing amino group density and enabling efficient antibody immobilization. Ratiometric detection was achieved by confining a solid-state redox probe within the porous electrode as an internal reference, while a solution-phase probe served as the analytical signal. The MOF layer not only enabled stable biomolecule conjugation but also effectively minimized redox probe leaching. With this concept, the proposed configuration significantly reduced signal fluctuation and improved measurement consistency compared with the non-ratiometric approach. The sensor exhibited high selectivity against common interferents and was validated using real food samples. Fabricated through scalable screen-printing technology, this integrated platform demonstrates strong potential as a universal, low-cost, and practical approach for on-site electrochemical immunosensing of mycotoxins.

Acetone has been identified as a volatile organic compound (VOC) biomarker, whose concentration in exhaled breath shows a strong correlation with diabetic status. Consequently, reliable detection of trace acetone gas is of paramount importance for diabetes diagnosis. However, acetone gas sensors still encounter specific technical challenges, such as an excessively high detection limit and insufficient sensitivity. To develop high-performance acetone sensing materials, constructing highly reactive composites has emerged as an effective approach to modulate the sensing properties of gas-sensitive materials. Herein, Co1-xZnxFe2O4 spinel bulk heterojunction (BHJ) composites with mixed spinel structure were efficiently synthesized via a simple one-step solvothermal process. The optimized Co0.5Zn0.5Fe2O4 sensor exhibited superior acetone-sensing performance, featuring a high response (S = 4.36 to 1 ppm acetone) and a low detection limit (S = 2.24 to 0.2 ppm acetone) at 200 °C. Significantly, the sensor could distinguish between exhaled breath samples from healthy individuals and those from simulated diabetic patients, indicating that Co1-xZnxFe2O4 composites represent a promising tool for noninvasive diabetes diagnosis via breath analysis.