High Accuracy Of Model Research Articles

A Transition-Matrix Growth Model and Equilibrium Curve for Natural Forests in Jiangxi Province

Researchers build growth models to predict the growth of forest stands and propose management measures to improve the overall quality of these stands. In this study, data collected from 91 sample plots from the eighth (2010) and ninth (2015) Chinese National Forest Inventories in Jiangxi Province were used to establish a transition-matrix growth model. Then, 12 potential equilibrium curves were set to guide forest management, and a transition-matrix growth model was used to predict stand growth in Jiangxi Province. In each 10-year management period, trees with diameters that exceeded the equilibrium curve were cut down. The results show that species diversity (H1), size diversity (H2), and basal area (B) have statistically significant influences on growth, mortality, and recruitment. Moreover, the high accuracy of the transition-matrix growth model is demonstrated. According to the simulation results, B = 35 m2/ha, the maximum diameter of retained trees Dmax = 45 cm and the adjacent diameter ratio q = 1.7 constitute the optimal equilibrium curve to guide forest management. The diameter distribution guided by the equilibrium curve is reverse J-shaped and is associated with significant increases in the hardwood stock volume and current annual growth. Under the guidance of the equilibrium curve, the forests in Jiangxi Province can be reasonably managed, produce more high-economic-value timber, and achieve a more stable species composition. This study will help maximize the ecological and economic benefits of forests and provide a reference for the realization of the sustainable development of forestry. Furthermore, the results can be used to improve the facility and accuracy of natural forest harvesting.

Experimental study on mix proportion optimization of anti-calcium dissolution shotcrete for tunnels based on response surface methodology

Aiming at the issue of crystallization and blockage of drainage system due to the massive calcium loss from the tunnel shotcrete, a self-designed tunnel seepage crystallization modelling system was developed. This system was produced in conjunction with the initial tunnel support shotcrete construction and drainage pipe installation, and is capable of simulating both the seepage process of groundwater in the shotcrete and the process of crystallization in the drainage pipe. Based on three different mechanisms of anti-crystallization, which include absorbing free calcium, reducing the porosity and increasing hydrophobicity, antialkali agent, nano-calcium carbonate, and silane were selected to test, respectively. Firstly, the suitable dosing ranges of these three external admixtures for resisting calcium loss in shotcrete were determined by single factor tests, which were 7%–11%, 4%–8%, and 0.3%–0.5%, respectively. Thereafter, the response surface method was employed to evaluate the interaction of antialkali agent, nano-calcium carbonate and silane on calcium loss in shotcrete, and to establish the relationship between them, and thus to determine the admixture ratio that can effectively reduce calcium loss crystallization in shotcrete, with the optimal admixture amounts of antialkali agent being 9.242%, nano-calcium carbonate 4.889% and silane 0.366%. Lastly, the reliability of the model test results was verified by the microscopic analysis, and the results showed that the total amount of calcium dissolution in the optimized group could be reduced by 75% compared with the blank control group, and was basically consistent with that derived from the response surface regression model, validating the high accuracy of the buildup response surface regression model. The present study can provide some ideas and references for reducing the seepage crystallization behavior of groundwater in the initial tunnel support shotcrete.

High resolution and accuracy model fitting interrogation method for optical fiber Polarization interferometric sensor

Interrogation methods largely determine the performance of the entire sensing system. However, interrogation methods alone are unlikely to provide very good results. An accurate model for the optical fiber polarization interferometric sensor (PIS) has been meticulously formulated within the frequency domain, incorporating the mode birefringence dispersion term of polarization maintaining fibers (PMFs). A novel model-fitting interrogation approach for the PIS is initially proposed and validated through experimental demonstration, which is bifurcated into parameter initialisation and model fitting interrogation process during the operation of the PIS. Subsequently, the parameters are determined at varying temperatures using this method. The interrogation capabilities of the proposed method are then experimentally contrasted with the valley wavelength tracking method. Results corroborate that the measurement range of model-fitting approach is no longer restricted by the spectrum bandwidth. The resolution and accuracy of our approach are 89% and 8 times higher than the valley wavelength tracking method, respectively.

Experimental Investigations and Modeling of Interference Fringe Geometry in Line-Shaped Gaussian Beam Intersections for Laser Doppler Sensors

Line-shaped beam-based Doppler distance sensors enable 3D shape measurements of rotating objects, for instance for working lathe monitoring with a simultaneous, multipoint velocity, and distance measurement. The velocity and distance estimations are fundamentally referred to the interference fringe spacing of the sensors. In this case, the fringe spacing variation-caused measurement error is significant; however, a complete, accurate model of the fringe geometry distribution for the line-shaped fringe volume is still missing and required to identify and minimize the error. Therefore, this work presents a high-accuracy 3D model for the fringe spacing evaluation of the sensors with experimental investigations. The model is derived from the phase expression of Gaussian beams introducing extension ratio, and can be universally applied to describe fringe geometry distribution throughout the intersection volume of spherical and line-shaped beams. With an experimental setup of a laser Doppler sensor, a full-field fringe spacing estimation using a high-resolution matrix camera is performed with dual-wavelength beams. The numerical modeling and experimental results show an average relative difference below 0.6%.

Prediction of arsenic adsorption onto metal organic frameworks and adsorption mechanisms interpretation by machine learning

Metal–organic frameworks (MOFs) are promising adsorbents for the removal of arsenic (As) from wastewater. The As removal efficiency is influenced by several factors, such as the textural properties of MOFs, adsorption conditions, and As species. Examining all of the relevant factors through traditional experiments is challenging. To predict the As adsorption capacities of MOFs toward organic, inorganic, and total As and reveal the adsorption mechanisms, four machine learning-based models were developed, with the adsorption conditions, MOF properties, and characteristics of different As species as inputs. The results demonstrated that the extreme gradient boosting (XGBoost) model exhibited the best predictive performance (test R2 = 0.93–0.96). The validation experiments demonstrated the high accuracy of the inorganic As-based XGBoost model. The feature importance analysis showed that the concentration of As, the surface area of MOFs, and the pH of the solution were the three key factors governing inorganic-As adsorption, while those governing organic-As adsorption were the concentration of As, the pHpzc value of MOFs, and the oxidation state of the metal clusters. The formation of coordination complexes between As and MOFs is possibly the major adsorption mechanism for both inorganic and organic As. However, electrostatic interaction may have a greater effect on organic-As adsorption than on inorganic-As adsorption. Overall, this study provides a new strategy for evaluating As adsorption on MOFs and discovering the underlying decisive factors and adsorption mechanisms, thereby facilitating the investigation of As wastewater treatment.

AN EMOTION RECOGNITION MODEL BASED ON LONG SHORT-TERM MEMORY NETWORKS AND EEG SIGNALS AND ITS APPLICATION IN PARAMETRIC DESIGN

One of the design objectives of a product is to create a positive emotional user experience. Through careful design, the product can evoke emotional resonance in users and stimulate their pleasure and satisfaction. Therefore, emotion recognition is crucial for parameterized product design. Considering that emotion recognition based on electroencephalogram (EEG) signals is more objective and accurate compared to methods such as text and surveys, this paper proposes an emotion analysis model based on long short-term memory (LSTM) and EEG and applies it to parameterized design. The main contributions of this paper are as follows. (1) Constructing a high-accuracy emotion recognition model. First, EEG data reflecting the characteristic patterns of brain activities in different emotional states are collected through EEG electrodes. Then, the EEG data are input into the LSTM network for training, enabling it to learn and capture the features associated with emotional states. During the training process, the model learns to extract crucial emotional features from the EEG data for emotion state recognition. This model can automatically learn emotional features, handle long-term dependencies and provide a more accurate and reliable solution for emotion recognition tasks. (2) Creating an EEG dataset specifically for evaluating emotions related to a product and using the trained emotion recognition model to classify this dataset, obtaining emotion classification results. The emotion classification results can be used to determine which parameter designs in product development need to be retained or discarded. These parameter designs can involve aspects such as user experience, functionality, aesthetics, usability and user-friendliness. Decisions can be made based on the emotion classification results to improve the quality and user satisfaction of the product.

Artificial Intelligence and its Application in the Prediction and Diagnosis of Animal Diseases: A Review

The role of artificial intelligence (AI) in veterinary science is becoming increasingly important as the technology advances. AI applications have the potential to revolutionize the prediction and diagnosis of animal diseases, thereby improving animal health by enhancing disease management. Given the complexity and subjectivity associated with evaluating medical images, as well as the difficulties inherent in handling animals, automating the analysis process using deep learning and artificial intelligence techniques would be highly beneficial. The main focus of the research study is to examine the applications of artificial intelligence in the prediction and diagnosis of Animal Diseases by the comprehensive process of literature review. It was found that many researchers have been using these techniques for diagnosis from the images, creating software that can help doctors, veterinarians and radiologists in their day-to-day work. The study includes machine learning and deep learning algorithms like PLS-DA, ANN and CNN models for the prediction and characterization of lesions. The results were analyzed and compared to obtain a high-accuracy model.

Sim-GAIL: A generative adversarial imitation learning approach of student modelling for intelligent tutoring systems

The continuous application of artificial intelligence (AI) technologies in online education has led to significant progress, especially in the field of Intelligent Tutoring Systems (ITS), online courses and learning management systems (LMS). An important research direction of the field is to provide students with customised learning trajectories via student modelling. Previous studies have shown that customisation of learning trajectories could effectively improve students’ learning experiences and outcomes. However, training an ITS that can customise students’ learning trajectories suffers from cold-start, time-consumption, human labour-intensity, and cost problems. One feasible approach is to simulate real students’ behaviour trajectories through algorithms, to generate data that could be used to train the ITS. Nonetheless, implementing high-accuracy student modelling methods that effectively address these issues remains an ongoing challenge. Traditional simulation methods, in particular, encounter difficulties in ensuring the quality and diversity of the generated data, thereby limiting their capacity to provide intelligent tutoring systems (ITS) with high-fidelity and diverse training data. We thus propose Sim-GAIL, a novel student modelling method based on generative adversarial imitation learning (GAIL). To the best of our knowledge, it is the first method using GAIL to address the challenge of lacking training data, resulting from the issues mentioned above. We analyse and compare the performance of Sim-GAIL with two traditional Reinforcement Learning-based and Imitation Learning-based methods using action distribution evaluation, cumulative reward evaluation, and offline-policy evaluation. The experiments demonstrate that our method outperforms traditional ones on most metrics. Moreover, we apply our method to a domain plagued by the cold-start problem, knowledge tracing (KT), and the results show that our novel method could effectively improve the KT model’s prediction accuracy in a cold-start scenario.

Comparison of Sentiment Analysis Methods on Topic Haram of Music In Youtube

Sentiment analysis on video lectures on YouTube that discuss the haram of music is an exciting topic to find out public opinion. This study aims to find what factors affect the model's accuracy in sentiment analysis, especially on video lecture content on YouTube. The data used is comment data on three video lectures that discuss the haram of music, which has been labelled into two categories: positive and negative. The data is divided into two categories, namely primary data, as many as 2099 data that have not been normalized, while secondary data has 1001 data that have been normalized. The experiment shows that the validity of the data, labelling the data, the amount of data, and preprocessing are essential points in forming a good sentiment analysis classification model because, from the test results, it was found that imbalance techniques such as SMOTE, word embedding word2Vec and FastText, and SVM and KNN classification algorithms do not provide maximum accuracy if the data used primary data. However, the data imbalance process, such as oversampling and SVM and KNN classification algorithms, will provide better model accuracy if used with secondary data. Based on the trial results, it is found that when using the SVM algorithm, primary data produces the highest accuracy at 58.35%, while secondary data is 72.23%. If using KNN, the primary data provides the highest model accuracy at 53.54%, while the secondary data has the highest accuracy at 72.81%. Based on these results, it was found that the validity of the data or data must be appropriate and related to the case raised and labelling the data must be done carefully because the most crucial is the inappropriate data in preprocessing the data must be done correctly, if data preprocessing is done in an inappropriate way then data imbalance techniques such as oversampling do not have enough influence on increasing accuracy, but if on the contrary then accuracy will increase. The selection of the right word embedding also affects accuracy. It is necessary to do many experiments to select the correct algorithm and follow the data owned because selecting the correct algorithm will provide maximum accuracy model results

Machine learning predictive model for motor moise by expanding dataset using electromagnetic force characteristics

To understand the variations of electric vehicle motor noise, the electromagnetic force change due to the tilting and the eccentricity between the rotor and the stator on the noise should be studied. However, it takes enormous computation time to consider all the various conditions, so machine learning (ML) model that can reduce the analysis time of electromagnetic force and noise was developed. The developed ML model showed very high accuracy prediction performance with an R2 value higher than 0.99 even for data that was not used at all for training. The speed increase was more than 100 times of FE analysis. The authors doubt that the high accuracy of the developed model implies the potential for electromagnetic force data structure to be much simpler than it appears. Therefore, various analyzes were conducted on the force and uneven magnetic pull. The change in the temporal-spatial main frequency component was not large, and only the change in the side order was meaningful. It was confirmed that to create a predictive model using a smaller dataset for not only the UMP but also all the individual teeth force is possible. This new way of creating a training dataset for an ML model helps to increase the training efficiency. Based on the generated prediction model, it can be used to identify changes in noise characteristics in major rotational orders and derive suitable improvements.

Investigation on solid–liquid equilibrium behavior of 4-cyanobenzoic acid in fourteen mono-solvents: Determination, correlation, molecular simulation and thermodynamic analysis

4-Cyanobenzoic acid (4-CyBZA) is an important pharmaceutical intermediate, and the research on solubility behavior of 4-CyBZA in organic solvents is very necessary for its crystallization and separation processes. Therefore, this study aimed at exploring dissolution behavior of 4-CyBZA in several pure solvents via experiment, calculation, and simulation process. In this study, mole-fraction solubility values of 4-CyBZA in fourteen common solvents including four low alcohols (methanol (MeOH), ethanol (EtOH), n-propanol (NPOH), n-butanol (NBOH)), four acetates (methyl acetate (MeAC), ethyl acetate (EtAC), n-propyl acetate (NPAC), n-butyl acetate (NBAC)), two alcohol ethers (2-ethoxyethanol (EGEE), 2-methoxyethanol (EGME)) as well as four common heteroatom/heterocyclic solvents (acetonitrile (MeCN), N,N-dimethylformamide (DMF), N-methylpyrrolidone (NMP), 1,4-dioxane (Diox)) were critically measured by using laser monitoring method from 278.15/288.15 K to 323.15 K under 100 kPa. Interestingly, experimental results indicated that an increase in temperature facilitated the dissolution of 4-CyBZA in selected 14 solvents. At the same time, the solubility curve underwent a crossover phenomenon. At 298.15 K, the common order of 4-CyBZA solubility was: DMF > EGME > EGEE > NMP > Diox > MeOH > MeAC > EtOH > EtAC > NPAC > NPOH > NBOH > NBAC > MeCN. The Hansen solubility parameter (HSP) was introduced to analyze solubility behavior via molecular similarity between 4-CyBZA and chosen solvents. Molecular modeling including Hirshfeld surface (HS) analysis and molecular electrostatic potential surface (MEPS) were employed to understand internal interactions within 4-CyBZA crystals. Moreover, solubility was fitted by Margules model, NRTL model, NRTL-SAC model, UNIQUAC model as well as a combined model (Jouyban model). Results showed that NRTL and NRTL-SAC model were more suitable for correlating 4-CyBZA solubility with absolute temperature. Furthermore, given the high accuracy of the NRTL model, dissolution thermodynamic characteristics of 4-CyBZA in chosen solvents were analyzed by NRTL model, which revealed that the dissolving process and mixing process were spontaneous and entropy-driven.

An Improved Data-Driven Method for Steering Feedback Torque of Driving Simulator

Traditional modeling methods of steering feedback torque (SFT) widely used in driving simulator are mainly physics-based, but suffers from high model complexity and low model accuracy due to the inaccurate or unknown structure, dynamics, and related parameters of the steering system, such as backlash or dry friction. For these problems, data-driven methods were considered to have advantageous in modeling such a complex yet dynamic system. However, due to the limitation of the data collected for model training from real vehicle test, the commonly used artificial neural network (ANN) cannot fully reflect the steering dynamics, and does not work well in many cases where collected data do not cover, and thus lead to poor generalization performance. In this article, a long short-term memory (LSTM) model is proposed and proved to simulate well on the dynamics of SFT, and works well in most driving conditions. A driving simulator is developed to verify the performance of different models in terms of generalization of working conditions. Based on the driving simulator data, we found that ANN failure generally arises in the transition between conventional and unconventional operating conditions, which is recognized by the proposed driving pattern recognition method. The LSTM-based modeling method can effectively solve this problem. The comparative analysis reveals that the improved LSTM model can simulate the SFT with higher accuracy and broader adaptability to driving simulator conditions.

VFLR: An Efficient and Privacy-Preserving Vertical Federated Framework for Logistic Regression

With the explosive growth of data volume and computing capability, federated learning, which involves constructing global models over multiple data islands, has demonstrated its advantages and vast prospects in the field of machine learning. However, due to commonly vertically partitioned data, coupled with privacy concerns about data leakage, there are still some challenging issues in traditional federated learning. To tackle these challenges, in this paper, we propose an efficient and privacy-preserving vertical federated learning framework for logistic regression, named VFLR, where multiple participants can collaboratively perform global model training and query over their vertically partitioned data. Specifically, we first design a data aggregation matrix construction algorithm, with which the vertically partitioned data can be aggregated for high-accuracy global model training. Then, by utilizing a novel symmetric homomorphic encryption, our framework can ensure that the whole training and query processes do not leak any private information. Moreover, based on the data aggregation matrix, multi-round interactions are not required in VFLR, improving training efficiency significantly. Detailed security analysis shows that VFLR can well protect data and model information from inference attacks. In addition, extensive experiments demonstrate that VFLR has high training and query accuracy and low computation and communication overhead.

SAS-HRM: Secure Authentication System for Human Resource Management

To guarantee data confidentiality and information sensitivity, human resource management requires secure systems. In the field of authorization and dependability in recognizing and identifying persons, facial recognition has grown in importance. In this research, a secure authentication system is proposed based on biometric aspects of the user's face and identifying it using the CNN classification model is provided to give access to human resource management and update data. The system is divided into four major stages: First, set up the system environment, beginning with smart cards, card readers, Arduino, and so on. Second, after undergoing pre-treatment steps, the facial characteristics are extracted using LDA. Third, create a high-accuracy CNN model to recognize and classify the user's face among the system's users. Finally, the user is allowed to enter the system and update his information. When compared to the accuracy of classification using machine learning techniques with a CNN proposed model, the accuracy of the model with LDA was up to 100%. K-NN has 91%, while TD has 94%.

Application of thermal imaging combined with machine learning for detecting the deterioration of the cassava root

Freshness is an important parameter that is indexed in the quality assessment of commercial cassava tubers. Cassava tubers that are not fresh have reduced starch content. Therefore, in this study, we aimed to develop a new approach to detect cassava root deterioration levels using thermal imaging with machine learning (ML). An underlying assumption was that nonfresh cassava roots may have fermentation inside that causes a difference in the inner temperature of the tuber. This creates the opportunity for the deterioration level to be measured using thermal imaging. The features (pixel intensity and temperature) that were extracted from the region of interest (ROI) in the form of tuber thermal images were analyzed with ML. Linear discriminant analysis (LDA), k-nearest neighbor (kNN), support vector machine (SVM), decision tree, and ensemble classifiers were applied to establish the optimal classification modeling algorithms. The highest accuracy model was developed from thermal images of cassava roots captured in a darkroom under a control temperature of 25 °C in the measurement chamber. The LDA, SVM, and ensemble classifiers gave the best overall performance for the discrimination of cassava root deterioration levels, with an accuracy of 86.7%. Interestingly, under uncontrolled environmental conditions, the combination of thermal imaging plus ML gave results that were of lower accuracy but still acceptable. Thus, our work revealed that thermal imaging coupled with ML was a promising method for the nondestructive evaluation of cassava root deterioration levels.

A mechanism-data hybrid-driven framework for identifying dynamic characteristic of actual chemical processes

As a prerequisite of accurate process analysis, prediction and optimization, the precise identification of process dynamic characteristics is of great significance. Traditional identification methods and data-driven technologies suffer from low accuracy and noise interference when facing the identification of actual chemical process. In this contribution, a mechanism-data hybrid-driven framework is developed to simultaneously capture the delay time and dynamic response of process variables. Combining non-parametric identification method, theoretical model and big data analysis technology, the proposed framework is able to derive more accurate and reliable identification results from the distributed control system (DCS) data directly, avoiding drawbacks brought by single method. The presented framework successfully identifies the flowrate dynamic properties of an actual deethanization process, and outperforms traditional approaches including finite impulse response model and Welch's averaged periodogram method. The contribution is significant for dynamic characteristic identification of actual chemical process, facilitating the developments of high-accuracy models for process prediction, control and intelligent optimization.

Groundwater pollution source identification using Metropolis-Hasting algorithm combined with Kalman filter algorithm

Increasing the precision of groundwater pollution source identification (GPSI) is crucial for groundwater pollution control and risk management. Bayesian theory based on the Markov Chain Monte Carlo (MCMC) method is a useful strategy of solving the GPSI problem. However, because of the nonlinear and uncertainty characteristics of GPSI, the Metropolis-Hasting (MH) algorithm, one of the most well-known MCMC algorithms, has the disadvantage of relatively low precision and is time-consuming. To address this problem, the Kalman filter (KF) algorithm was combined with the MH algorithm and referred to as the Kalman filter Metropolis-Hasting (KF–MH) algorithm. The algorithm generates a new initial distribution that is close to the true value through a prior distribution, and the new initial distribution is used to perform subsequent iterations of the calculation. The viability and superiority of the proposed KF-MH algorithm were assessed in three hypothetical GPSI cases under different conditions. In the inversion process, a surrogate model was constructed using a temporal convolutional network (TCN) to reduce the computational pressure imposed by the numerical simulation model. The results of the TCN surrogate model in the cases illustrate the high accuracy of the TCN surrogate model in fitting the groundwater numerical model. In the three cases, the normalized errors between the identification results and the true values of the source features obtained with the KF–MH algorithm were significantly lower than those of the MH algorithm. The results indicate that the proposed KF–MH algorithm has higher inversion accuracy than the MH algorithm.

Cloud Modelling of Property-Level Flood Exposure in Megacities

Surface water flood risk is projected to increase worldwide due to the growth of cities as well as the frequency of extreme rainfall events. Flood risk modelling at high resolution in megacities is now feasible due to the advent of high spatial resolution terrain data, fast and accurate hydrodynamic models, and the power of cloud computing platforms. Analysing the flood exposure of urban features in these cities during multiple storm events is essential to understanding flood risk for insurance and planning and ultimately for designing resilient solutions. This study focuses on London, UK, a sprawling megacity that has experienced damaging floods in the last few years. The analysis highlights the key role of accurate digital terrain models (DTMs) in hydrodynamic models. Flood exposure at individual building level is evaluated using the outputs from the CityCAT model driven by a range of design storms of different magnitudes, including validation with observations of a real storm event that hit London on the 12 July 2021. Overall, a novel demonstration is presented of how cloud-based flood modelling can be used to inform exposure insurance and flood resilience in cities of any size worldwide, and a specification is presented of what datasets are needed to achieve this aim.

A totally coupled multi time-scale framework containing full parameters online identification and SOC real-time estimation of lithium-ion battery based on a fractional order model

Fractional order model (FOM) is one of the most promising choices for lithium-ion battery simulation and internal states estimation due to its high modeling accuracy. However, online identifying the order and parameters of FOM is extremely tough, which also makes it difficult to be applied in complex and variable operating conditions. To address this issue, this paper proposes a totally coupled multi time-scale framework that contains full parameters online identification of FOM and state-of-charge (SOC) real-time estimation of lithium-ion battery. Firstly, a novel global optimization algorithm named improved particle swarm optimization is designed to update the order of FOM on macro-time scale with low computational burden. On the basis of the determined order, the rest parameters of FOM are further identified by forgetting factor recursive least square algorithm on micro-time scale. With those online identified parameters, adaptive extended Kalman filter is finally adopted to track battery SOC in every instant. The estimation accuracy, adaptability to different operating temperatures and battery aging, and robustness ability against uncertain initialization of the developed method is verified under Federal Urban Driving Schedule tests. The corresponding results demonstrate that the proposed method can realize accurate full parameters online identification of FOM and precise SOC real-time estimation against different temperatures, battery aging states and initialization. Using the developed method, the mean absolute error (MAE) and root mean square error (RMSE) of SOC estimation of the fresh battery under temperature ranging from 0 °C to 50 °C can be limited below 0.9 % and 1.1 %, respectively. Even if the SOC-OCV relationship is not timely updated, those two indexes of the aged battery achieved by the proposed method can still be controlled within 3 % under temperature ranging from 0 °C to 50 °C. Moreover, the comparison with other typical FOM-based estimation methods is also conducted, whose results indicate that the developed method has apparently superior comprehensive performance on estimation accuracy and adaptation.

Machine Learning Integrated Workflow for Predicting Schwann Cell Viability on Conductive MXene Biointerfaces.

Severe injuries to the peripheral nervous system (PNS) require Schwann cells to aid in neuronal regeneration. Low-frequency electrical stimulation is known to induce the cogrowth of neurons and Schwann cells in an injured PNS. However, the correlations between electrical stimulation and Schwann cell viability are complex and not well understood. In this work, we develop a machine learning (ML)-integrated workflow that uses conductive hydrogel biointerfaces to evaluate the impacts of fabrication parameters and electrical stimulation on the Schwann cell viability. First, a hydrogel array with varying MXene and peptide loadings is fabricated, which serves as conductive biointerfaces to incubate Schwann cells and introduce various electrical stimulation (at different voltages and frequencies). Upon specific fabrication parameters and stimulation, the cell viability is evaluated and input into an artificial neural network model to train the model. Additionally, a data augmentation method is applied to synthesize 1000-fold virtual data points, enabling the construction of a high-accuracy prediction model (with a testing mean absolute error ≤11%). By harnessing the model's predictive power, we can accurately predict Schwann cell viability based on a given set of fabrication/stimulation parameters. Finally, the SHapley Additive exPlanations model interpretation provides several data-scientific insights that are validated by microscopic cellular observations. Our hybrid approach, involving conductive biointerface fabrication, ML algorithms, and data analysis, offers an unconventional platform to construct a preclinical prediction model at the cellular level.