Abstract: Organic small molecules are proven to be capable of passivating the bulk/interfacial defects in inorganic perovskite solar cells. Considering the burdensome situation to screen the functional small molecules, we employ a modified machine learning (ML) strategy to guide screening suitable small molecules toward efficient solar cells through three modified ML algorithms to construct the prediction model: (i) random forest algorithm (RF), (ii) support vector machine algorithm (SVR), and (iii) XGBoost. Among them, the XGBoost algorithm displays a better overall predictive performance, whereby the <i>R</i>² index reaches 0.939. Accordingly, eight small molecules are selected to modify the interface of perovskite films, and both the theoretical and experimental results certify that the difluoro­benzylamine with additional fluorine atoms has a better interface modification effect among the small molecules containing functional groups, e.g., the benzene ring and amino group. The high accuracy of the modified machine learning model enables us to simplify the small-molecule screening process and form an important step for ongoing developments in perovskite solar cells and other optoelectronic devices.

Abstract: The amalgamation of surface-enhanced Raman spectroscopy (SERS) and machine learning (ML) presents a discerning capability to differentiate among a diverse spectrum of bacterial strains. However, addressing the challenge of achieving expeditious and robust bacterial detection remains a prominent focal point. This study delineates a comprehensive bacterial classification and identification methodology grounded in the amoxicillin response, which effectively categorizes bacteria utilizing SERS and ML within a time frame of less than 20 min. The bacterial specimens are subjected to pharmacological stimulation, inducing the release of purine molecules that are integral to metabolic processes. Capitalizing on the preferential entry of these molecules into SERS hot spots over the bacteria themselves facilitates the consistent acquisition of stable SERS signals. Experimental evidence demonstrates that the interaction of S. aureus, E. coli, S. epidermidis, C. albicans, and K. pneumoniae with amoxicillin contributes to an enhancement in the stability and signal intensity of bacterial SERS. Utilizing a random forest (RF) model on pure bacterial samples yields an exemplary classification accuracy of 99%. Furthermore, the application of three distinct models, support vector machine (SVM), RF, and CNN-LSTM-Attention (CLA) in the analysis of clinical samples culminates in final classification accuracies of 92%, 87%, and 96%, respectively. This approach establishes a rapid, straightforward, and stable classification methodology for SERS-based bacterial detection, demonstrating significant potential for clinical diagnostic applications.

Abstract: Understanding the phase stability of gas hydrates under confinement is fundamental to the geological stability evolutions of gas hydrate systems on Earth. Herein, the phase stability of CH₄ and CO₂ hydrates under confinement is predicted by machine learning. Three machine learning models, including support vector machine, random forest, and gradient boosting decision tree, are constructed to predict the phase stability of CH₄ and CO₂ hydrates under confinement. Our machine learning results show that the prediction accuracy of the support vector machine model is highest, yet the prediction accuracy of the random forest model is lowest among those machine learning models in determining the phase stability of confined gas hydrates. Based on their performance in predicting the phase stability of confined gas hydrates, the support vector machine model with a training set fraction of 0.7 is finally chosen to deal with the unknown phase stability of confined gas hydrates. Importantly, the average accuracy of the support vector machine model can reach more than 90% in predicting the unknown phase stability of both CH₄ and CO₂ hydrates. The trained machine learning models can help us to quickly and accurately determine the phase stability of CH₄ and CO₂ hydrates under confinement in future applications.

Abstract: Over the past few years, new psychoactive substances (NPS) have become a global health and social problem because of their wide variety, constant structural renewal, vague legal definitions, and rapid adaptation to legal restrictions. The rapid structural modifications of NPS have posed significant challenges for the screening and identification of these new substances using traditional mass spectrometric techniques based on reference substances or a mass spectral database. Here, we propose supervised machine learning (ML) classification models such as k-nearest neighbors, support vector machine, random forest, and multigrained cascade forest for the rapid screening of NPS using mass spectrometric data. This approach utilizes ML methods to learn the statistical probability distributions of mass spectral data for NPS and non-NPS. Four classification ML models were generated and evaluated using a data set comprising 567 LC-MS and 732 GC-MS spectra. Through cross validation, we achieved an F1 score of 0.35–0.97. These algorithms were applied in conjunction with mass spectrometry techniques for the detection of six seizures including electronic cigarette oil and suspected powdered substances netted in drug trafficking cases. The models provided warning signals for synthetic cannabinoids, synthetic cathinones, and fentanyl. Thus, an early warning system was successfully established, which provided a useful method for reliable and effective identifications of unknown NPS.