Deep learning framework for sensor array precision and accuracy enhancement

Pd-Modified Microneedle Array Sensor Integration with Deep Learning for Predicting Silica Aerogel Properties in Real Time

The continuous global effort to predict material properties through artificial intelligence has predominantly focused on utilizing material stoichiometry or structures in deep learning models. This study aims to predict material properties using electrochemical impedance data, along with frequency and time parameters, that can be obtained during processing stages. The target material, silica aerogel, is widely recognized for its lightweight structure and excellent insulating properties, which are attributed to its large surface area and pore size. However, production is often delayed due to the prolonged aging process. Real-time prediction of material properties during processing can significantly enhance process optimization and monitoring. In this study, we developed a system to predict the physical properties of silica aerogel, specifically pore diameter, pore volume, and surface area. This system integrates a 3 × 3 array Pd/Au sensor, which exhibits high sensitivity to varying pH levels during aerogel synthesis and is capable of acquiring a large data set (impedance, frequency, time) in real-time. The collected data is then processed through a deep neural network algorithm. Because the system is trained with data obtained during the processing stage, it enables real-time predictions of the critical properties of silica aerogel, thus facilitating process optimization and monitoring. The final performance evaluation demonstrated an optimal alignment between true and predicted values for silica aerogel properties, with a mean absolute percentage error of approximately 0.9%. This approach holds great promise for significantly improving the efficiency and effectiveness of silica aerogel production by providing accurate real-time predictions.

Single-Atom Iridium-doped Carbon Dots Nanozyme with High Peroxidase-Like Activity as Colorimetric Sensors for Multimodal Detection of Mercury Ions

Nanozyme-based colorimetric sensors are promising approaches for environmental monitoring, food safety, and medical diagnostics. However, developing novel nanozymes that exhibit high catalytic activity, good dispersion in aqueous solution, high sensitivity, selectivity, and stability is challenging. In this study, for the first time, single-atom iridium-doped carbon dot nanozymes (SA Ir-CDs) are synthesized via a simple in situ pyrolysis process. Doping carbon dots with iridium in the form of single atoms to achieve maximum atomic utilization not only enhances peroxidase (POD)-like activity to 178.81 U mg-1 but also improves the dispersibility of single-atom nanozymes in aqueous solutions over 30 days. Hence, the SA Ir-CD colorimetric platform is developed for mercury ions (Hg2+) detection and exhibited a good linear relationship from 0.01 to 10 µm and a detection limit of 4.4 nm. Notably, the changes in color can be observed not only through the naked eye but also via a smartphone, enabling convenient field and onsite monitoring without the need for sophisticated analytical equipment. In this study, an approach for fabricating single-atom metal-based carbon dot nanozymes with high POD-like activity is developed, and a new effective and easy-to-use colorimetric sensor for Hg2+ detection is constructed.

Feasibility of Using a Cheap Colour Sensor to Detect Blends of Vegetable Oils in Avocado Oil

This proof-of-concept study explored the use of an RGB colour sensor to identify different blends of vegetable oils in avocado oil. The main aim of this work was to distinguish avocado oil from its blends with canola, sunflower, corn, olive, and soybean oils. The study involved RGB measurements conducted using two different light sources: UV (395 nm) and white light. Classification methods, such as Linear Discriminant Analysis (LDA) and Least Squares Support Vector Machine (LS-SVM), were employed for detecting the blends. The LS-SVM model exhibited superior classification performance under white light, with an accuracy exceeding 90%, thus demonstrating a robust prediction capability without evidence of random adjustments. A quantitative approach was followed as well, employing Multiple Linear Regression (MLR) and LS-SVM, for the quantification of each vegetable oil in the blends. The LS-SVM model consistently achieved good performance (R2 > 0.9) in all examined cases, both for internal and external validation. Additionally, under white light, LS-SVM models yielded root mean square errors (RMSE) between 1.17-3.07%, indicating a high accuracy in blend prediction. The method proved to be rapid and cost-effective, without the necessity of any sample pretreatment. These findings highlight the feasibility of a cost-effective colour sensor in identifying avocado oil blended with other oils, such as canola, sunflower, corn, olive, and soybean oils, suggesting its potential as a low-cost and efficient alternative for on-site oil analysis.