Finding Hidden Signals in Chemical Sensors Using Deep Learning

Unlocking Predictive Capability and Enhancing Sensing Performances of Plasmonic Hydrogen Sensors via Phase Space Reconstruction and Convolutional Neural Networks

This study innovates plasmonic hydrogen sensors (PHSs) by applying phase space reconstruction (PSR) and convolutional neural networks (CNNs), overcoming previous predictive and sensing limitations. Utilizing a low-cost and efficient colloidal lithography technique, palladium nanocap arrays are created and their spectral signals are transformed into images using PSR and then trained using CNNs for predicting the hydrogen level. The model achieves accurate predictions with average accuracies of 0.95 for pure hydrogen and 0.97 for mixed gases. Performance improvements observed are a reduction in response time by up to 3.7 times (average 2.1 times) across pressures, SNR increased by up to 9.3 times (average 3.9 times) across pressures, and LOD decreased from 16 Pa to an extrapolated 3 Pa, a 5.3-fold improvement. A practical application of remote hydrogen sensing without electronics in hydrogen environments is actualized and achieves a 0.98 average test accuracy. This methodology reimagines PHS capabilities, facilitating advancements in hydrogen monitoring technologies and intelligent spectrum-based sensing.

AI-Driven Sensing Technology: Review

Machine learning and deep learning technologies are rapidly advancing the capabilities of sensing technologies, bringing about significant improvements in accuracy, sensitivity, and adaptability. These advancements are making a notable impact across a broad spectrum of fields, including industrial automation, robotics, biomedical engineering, and civil infrastructure monitoring. The core of this transformative shift lies in the integration of artificial intelligence (AI) with sensor technology, focusing on the development of efficient algorithms that drive both device performance enhancements and novel applications in various biomedical and engineering fields. This review delves into the fusion of ML/DL algorithms with sensor technologies, shedding light on their profound impact on sensor design, calibration and compensation, object recognition, and behavior prediction. Through a series of exemplary applications, the review showcases the potential of AI algorithms to significantly upgrade sensor functionalities and widen their application range. Moreover, it addresses the challenges encountered in exploiting these technologies for sensing applications and offers insights into future trends and potential advancements.

Toward Nano- and Microplastic Sensors: Identification of Nano- and Microplastic Particles via Artificial Intelligence Combined with a Plasmonic Probe Functionalized with an Estrogen Receptor

Nano- and microplastic particles are a global and emerging environmental issue that might pose potential threats to human health. The present work exploits artificial intelligence (AI) to identify nano- and microplastics in water by monitoring the interaction of the sample with a sensitive surface. An estrogen receptor (ER) grafted onto a gold surface, realized on a nonexpensive and easy-to-produce plastic optical fiber (POF) platform in order to excite a surface plasmon resonance (SPR) phenomenon, has been developed in order to carry out a "smart" sensitive interface (ER-SPR-POF interface). The ER-SPR-POF interface offers output data useful for exploiting a machine learning-based approach to achieve nano- and microplastic particle sensors. This work developed a proof-of-concept sensor through a training phase carried out by different particles, in terms of materials and size. The experimental results have demonstrated that the proposed "smart" ER-SPR-POF interface combined with AI can be used to identify the kind of particles in terms of the materials (polystyrene; poly(methyl methacrylate)) and size (20 μm; 100 nm) with an accuracy of 90.3%.

Improving biosensor accuracy and speed using dynamic signal change and theory-guided deep learning

False results and time delay are longstanding challenges in biosensing. While classification models and deep learning may provide new opportunities for improving biosensor performance, such as measurement confidence and speed, it remains a challenge to ensure that predictions are explainable and consistent with domain knowledge. Here, we show that consistency of deep learning classification model predictions with domain knowledge in biosensing can be achieved by cost function supervision and enables rapid and accurate biosensing using the biosensor dynamic response. The impact and utility of the methodology were validated by rapid and accurate quantification of microRNA (let-7a) across the nanomolar (nM) to femtomolar (fM) concentration range using the dynamic response of cantilever biosensors. Data augmentation and cost function supervision based on the consistency of model predictions and experimental observations with the theory of surface-based biosensors improved the F1 score, precision, and recall of a recurrent neural network (RNN) classifier by an average of 13.8%. The theory-guided RNN (TGRNN) classifier enabled quantification of target analyte concentration and false results with an average prediction accuracy, precision, and recall of 98.5% using the initial transient or entire dynamic response, which is indicative of high prediction accuracy and low probability of false-negative and false-positive results. Classification scores were used to establish new relationships among biosensor performance characteristics (e.g., measurement confidence) and design parameters (e.g., inputs and hyperparameters of classification models and data acquisition parameters) that may be used for characterizing biosensor performance.

Rational design of hybrid sensor arrays combined synergistically with machine learning for rapid response to a hazardous gas leak environment in chemical plants

Combinations of semiconductor metal oxide (SMO) sensors, electrochemical (EC) sensors, and photoionization detection (PID) sensors were used to discriminate chemical hazards on the basis of machine learning. Sensing data inputs were exploited in the form of either numerical or image data formats, and the classification of chemical hazards with high accuracy was achieved in both cases. Even a small amount of gas sensing or purging data (input for ∼30 s) input can be exploited in machine-learning-based gas discrimination. SMO sensors exhibit high performance even in a single-sensor mode, presumably because of the intrinsic cross-sensitivity of metal oxides, which is otherwise considered a major disadvantage of SMO sensors. EC sensors were enhanced through synergistic integration of sensor combinations with machine learning. For precision detection of multiple target analytes, a minimum number of sensors can be proposed for gas detection/discrimination by combining sensors with dissimilar operating principles. The Type I hybrid sensor combines one SMO sensor, one EC sensor, and one PID sensor and is used to identify NH3 gas mixed with sulfur compounds in simulations of NH3 gas leak accidents in chemical plants. The portable remote sensing module made with a Type I hybrid sensor and LTE module can identify mixed NH3 gas with a detection time of 60 s, demonstrating the potential of the proposed system to quickly respond to hazardous gas leak accidents and prevent additional damage to the environment.

A Study on E-Nose System in Terms of the Learning Efficiency and Accuracy of Boosting Approaches

With the development of the field of e-nose research, the potential for application is increasing in various fields, such as leak measurement, environmental monitoring, and virtual reality. In this study, we characterize electronic nose data as structured data and investigate and analyze the learning efficiency and accuracy of deep learning models that use unstructured data. For this purpose, we use the MOX sensor dataset collected in a wind tunnel, which is one of the most popular public datasets in electronic nose research. Additionally, a gas detection platform was constructed using commercial sensors and embedded boards, and experimental data were collected in a hood environment such as used in chemical experiments. We investigated the accuracy and learning efficiency of deep learning models such as deep learning networks, convolutional neural networks, and long short-term memory, as well as boosting models, which are robust models on structured data, using both public and specially collected datasets. The results showed that the boosting models had a faster and more robust performance than deep learning series models.

A deep learning-based simulator for comprehensive two-dimensional GC applications

Among the main approaches for predicting the spatial positions of eluates in comprehensive two-dimensional gas chromatography, the still under-explored computational models based on deep learning algorithms emerge as robust and reliable options due to their high adaptability to the structure and complexity of the data. In this work, an open-source program based on deep neural networks was developed to optimize chromatographic methods and simulate operating conditions outside the laboratory. The deep neural networks models were fit to convenient experimental predictors, resulting in scaled losses (mean squared error) equivalent to 0.006 (relative average deviation = 8.56%, R2  = 0.9202) and 0.014 (relative average deviation = 1.67%, R2  = 0.8009) in the prediction of the first- and second-dimension retention times, respectively. Good compliance was observed for the main chemical classes, such as environmental contaminants: volatile, semivolatile organic compounds, and pesticides; biochemistry molecules: amino acids and lipids; pharmaceutical industry and personal care products and residues: drugs and metabolites; among others. On the other hand, there is a need for continuous database updates to predict retention times of less common compounds accurately. Thus, forming a collaborative database is proposed, gathering voluntary findings from other users.

Skin-Interfaced Wearable Sweat Sensors for Precision Medicine

Wearable sensors hold great potential in empowering personalized health monitoring, predictive analytics, and timely intervention toward personalized healthcare. Advances in flexible electronics, materials science, and electrochemistry have spurred the development of wearable sweat sensors that enable the continuous and noninvasive screening of analytes indicative of health status. Existing major challenges in wearable sensors include: improving the sweat extraction and sweat sensing capabilities, improving the form factor of the wearable device for minimal discomfort and reliable measurements when worn, and understanding the clinical value of sweat analytes toward biomarker discovery. This review provides a comprehensive review of wearable sweat sensors and outlines state-of-the-art technologies and research that strive to bridge these gaps. The physiology of sweat, materials, biosensing mechanisms and advances, and approaches for sweat induction and sampling are introduced. Additionally, design considerations for the system-level development of wearable sweat sensing devices, spanning from strategies for prolonged sweat extraction to efficient powering of wearables, are discussed. Furthermore, the applications, data analytics, commercialization efforts, challenges, and prospects of wearable sweat sensors for precision medicine are discussed.

Machine learning toward high-performance electrochemical sensors

The so-coined fourth paradigm in science has reached the sensing area, with the use of machine learning (ML) toward data-driven improvements in sensitivity, reproducibility, and accuracy, along with the determination of multiple targets from a single measurement using multi-output regression models. Particularly, the use of supervised ML models trained on large data sets produced by electrical and electrochemical bio/sensors has emerged as an impacting trend in the literature by allowing accurate analyses even in the presence of usual issues such as electrode fouling, poor signal-to-noise ratio, chemical interferences, and matrix effects. In this trend article, apart from an outlook for the coming years, we present examples from the literature that demonstrate how helpful ML algorithms can be for dispensing the adoption of experimental methods to address the aforesaid interfering issues, ultimately contributing to translate testing technologies into on-site, practical, and daily applications.

Machine learning for yield prediction for chemical reactions using in situ sensors

Machine learning models were developed to predict product formation from time-series reaction data for ten Buchwald-Hartwig coupling reactions. The data was provided by DeepMatter and was collected in their DigitalGlassware cloud platform. The reaction probe has 12 sensors to measure properties of interest, including temperature, pressure, and colour. Colour was a good predictor of product formation for this reaction and machine learning models were able to learn which of the properties were important. Predictions for the current product formation (in terms of % yield) had a mean absolute error of 1.2%. For predicting 30, 60 and 120 min ahead the error rose to 3.4, 4.1 and 4.6%, respectively. The work here presents an example into the insight that can be obtained from applying machine learning methods to sensor data in synthetic chemistry.

Multilevel Self-Assembly of Block Copolymers and Polymer Colloids for a Transparent and Sensitive Gas Sensor Platform

The recent emerging significance of the Internet of Things (IoT) demands sensor devices to be integrated with many different functional structures and devices while conserving their original functionalities. To this end, optical transparency and mechanical flexibility of sensor devices are critical requirements for optimal integration as well as high sensitivity. In this work, a transparent, flexible, and sensitive gas sensor building platform is introduced by using multilevel self-assembly of block copolymers (BCPs) and polystyrene (PS) colloids. For the demonstration of an H₂ gas sensor, a hierarchically porous Pd metal mesh structure is obtained by overlaying the two different patterned template structures with synergistic, distinctive characteristic length scales. The hierarchical Pd mesh shows not only high transparency over 90% but also superior sensing performance in terms of response and recovery time owing to enhanced Pd-to-hydride ratio and short H₂ diffusion lengths from the enlarged active surface areas. The hierarchical morphology also endows high mechanical flexibility while securing reliable sensing performance even under severe mechanical deformation cycles. Our scalable self-assembly based multiscale nanopatterning offers an intriguing generalized platform for many different multifunctional devices requiring hidden in situ monitoring of environmental signals.

Machine learning, artificial intelligence, and chemistry: how smart algorithms are reshaping simulation and the laboratory

Machine learning and artificial intelligence are increasingly gaining in prominence through image analysis, language processing, and automation, to name a few applications. Machine learning is also making profound changes in chemistry. From revisiting decades-old analytical techniques for the purpose of creating better calibration curves, to assisting and accelerating traditional in silico simulations, to automating entire scientific workflows, to being used as an approach to deduce underlying physics of unexplained chemical phenomena, machine learning and artificial intelligence are reshaping chemistry, accelerating scientific discovery, and yielding new insights. This review provides an overview of machine learning and artificial intelligence from a chemist’s perspective and focuses on a number of examples of the use of these approaches in computational chemistry and in the laboratory.

Deep Learning-Based Sensor Array: 3D Fluorescence Spectra of Gold Nanoclusters for Qualitative and Quantitative Analysis of Vitamin B6 Derivatives

Vitamin B₆ derivatives (VB6Ds) are of great importance for all living organisms to complete their physiological processes. However, their excess in the body can cause serious problems. What is more, the qualitative and quantitative analysis of different VB6Ds may present significant challenges due to the high similarity of their chemical structures. Also, the transfer of deep learning model from one task to a similar task needs to be present more in the fluorescence-based biosensor. Therefore, to address these problems, two deep learning models based on the intrinsic fingerprint of 3D fluorescence spectra have been developed to identify five VB6Ds. The accuracy ranges of a deep neural network (DNN) and a convolutional neural network (CNN) were 94.44-97.77% and 97.77-100%, respectively. After that, the developed models were transferred for quantitative analysis of the selected VB6Ds at a broad concentration range (1-100 μM). The determination coefficient (R²) values of the test set for DNN and CNN were 93.28 and 97.01%, respectively, which also represents the outperformance of CNN over DNN. Therefore, our approach opens new avenues for qualitative and quantitative sensing of small molecules, which will enrich fields related to deep learning, analytical chemistry, and especially sensor array chemistry.

The case for data science in experimental chemistry: examples and recommendations

The physical sciences community is increasingly taking advantage of the possibilities offered by modern data science to solve problems in experimental chemistry and potentially to change the way we design, conduct and understand results from experiments. Successfully exploiting these opportunities involves considerable challenges. In this Expert Recommendation, we focus on experimental co-design and its importance to experimental chemistry. We provide examples of how data science is changing the way we conduct experiments, and we outline opportunities for further integration of data science and experimental chemistry to advance these fields. Our recommendations include establishing stronger links between chemists and data scientists; developing chemistry-specific data science methods; integrating algorithms, software and hardware to 'co-design' chemistry experiments from inception; and combining diverse and disparate data sources into a data network for chemistry research.

Machine learning and chemometrics for electrochemical sensors: moving forward to the future of analytical chemistry

Electrochemical sensors and biosensors have been successfully used in a wide range of applications, but systematic optimization and nonlinear relationships have been compromised for electrode fabrication and data analysis. Machine learning and experimental designs are chemometric tools that have been proved to be useful in method development and data analysis. This minireview summarizes recent applications of machine learning and experimental designs in electroanalytical chemistry. First, experimental designs, e.g., full factorial, central composite, and Box-Behnken are discussed as systematic approaches to optimize electrode fabrication to consider the effects from individual variables and their interactions. Then, the principles of machine learning algorithms, including linear and logistic regressions, neural network, and support vector machine, are introduced. These machine learning models have been implemented to extract complex relationships between chemical structures and their electrochemical properties and to analyze complicated electrochemical data to improve calibration and analyte classification, such as in electronic tongues. Lastly, the future of machine learning and experimental designs in electrochemical sensors is outlined. These chemometric strategies will accelerate the development and enhance the performance of electrochemical devices for point-of-care diagnostics and commercialization.

Weak Reaction Scatterometry of Plasmonic Resonance Light Scattering with Machine Learning

Weak reactions are usually overlooked due to weak detectable features and susceptibility to interference from noise signals. Strategies for detecting weak reactions are essential for exploring reaction mechanisms and exploiting potential applications. Machine learning has recently been successfully used to identify patterns and trends in the data. Here, it is demonstrated that machine learning-based techniques can offer accurate local surface plasmon resonance (LSPR) scatterometry by improving the precision of the plasmonic scattering imaging in weak chemical reactions. Dark-field microscopy (DFM) imaging technique is the most effective method for high-sensitivity plasmonic nanoparticles LSPR scatterometry. Unfortunately, deviations caused by the instrument and operating errors are inevitable, and it is difficult to effectively detect the presence of weak reactions. Thus, introducing a machine learning calibration model to automatically calibrate the scattering signal of the nanoprobe in the reaction process can greatly improve the confidence of LSPR scatterometry under DFM imaging and allow DFM imaging to effectively monitor unobvious or weak reactions. By this approach, the weak oxidation of silver nanoparticles (AgNPs) in water by dissolved oxygen was successfully monitored. Moreover, a trivial reaction between AgNPs and mercury ions was detected in a dilute mercury solution with a concentration greater than 1.0 × 10^-10 mol/L. These results suggest the great potential of the combination of LSPR scatterometry and machine learning as a method for imaging analysis and intelligent sensing.