Back-propagation Artificial Neural Network Model Research Articles

Formation and prediction of heterocyclic amines and N-nitrosamines in smoked sausages using back propagation artificial neural network.

Heterocyclic amines (HAs) and N-nitrosamines (NAs) are formed easily during the thermal processing of food, and epidemiological studies have demonstrated that consuming HAs and NAs increases the risk of cancer. However, there are few studies on the application of back propagation artificial neural network (BP-ANN) models to simultaneously predict the content of HAs and NAs in sausages. This study aimed to investigate the effects of cooking time and temperature, smoking time and temperature, and fat-to-lean ratio on the formation of HAs and NAs in smoked sausages, and to predict their total content based on the BP-ANN model. With an increase in processing time, processing temperature and fat ratio, the content of HAs and NAs in smoked sausages increased significantly, while the content of HA precursors and nitrite residues decreased significantly. The optimal network topology of the BP-ANN model was 5-11-2, the correlation coefficient values for training, validation, testing and all datasets were 0.99228, 0.99785, 0.99520 and 0.99369, respectively, and the mean squared error value of the best validation performance was 0.11326. The bias factor and the accuracy factor were within acceptable limits, and the predicted values approximated the true values, indicating that the model has good predictive performance. The contents of HAs and NAs in smoked sausages were significantly influenced by the cooking conditions, smoking conditions and fat ratio. The BP-ANN model has high application value in predicting the contents of HAs and NAs in sausages, which provides a theoretical basis for the suppression of carcinogen formation. © 2024 Society of Chemical Industry.

Predictive Quality Analytics of Surface Roughness in Turning Operation Using Polynomial and Artificial Neural Network Models

The variability of the material properties of steel from different suppliers causes problems in achieving the required surface quality after turning. Therefore, the manufacturer needs to estimate the resulting quality before starting production, especially if it is an expensive, small-batch production from stainless steel. Predictive models will make it possible to estimate the surface roughness from the mechanical properties of steel and thus support decision making about supplier selection or acceptance of a material supply. This research presents a step-by-step decision-making procedure, which enables the trained staff to make quick decisions based on commonly available information in the Mill Test Certificate (MTC). A new multivariate second-order polynomial model and feedforward backpropagation artificial neural network (ANN) models have been developed using input variables from the MTC: Tensile Strength, Yield Strength, Elongation, and Hardness. Models were used to enhance the methodological robustness in formulating the decision if the predicted surface roughness is outside the required range, even before accepting the delivery. Both models can accurately predict surface roughness, while the ANN model is more accurate than the polynomial model; however, the predictive model is sensitive to the accuracy of the input data, and the model’s prediction is valid only under precisely defined conditions.

Assessing the impact of occlusal plane rotation on facial aesthetics in orthodontic treatment: a machine learning approach

BackgroundAdequate occlusal plane (OP) rotation through orthodontic therapy enables satisfying profile improvements for patients who are disturbed by their maxillomandibular imbalance but reluctant to surgery. The study aims to quantify profile improvements that OP rotation could produce in orthodontic treatment and whether the efficacy differs among skeletal types via machine learning.Materials and methodsCephalometric radiographs of 903 patients were marked and analyzed by trained orthodontists with assistance of Uceph, a commercial software which use artificial intelligence to perform the cephalometrics analysis. Back-propagation artificial neural network (BP-ANN) models were then trained based on collected samples to fit the relationship among maxillomandibular structural indicators, SN-OP and P-A Face Height ratio (FHR), Facial Angle (FA). After corroborating the precision and reliability of the models by T-test and Bland-Altman analysis, simulation strategy and matrix computation were combined to predict the consequent changes of FHR, FA to OP rotation. Linear regression and statistical approaches were then applied for coefficient calculation and differences comparison.ResultsThe regression scores calculating the similarity between predicted and true values reached 0.916 and 0.908 in FHR, FA models respectively, and almost all pairs were in 95% CI of Bland-Altman analysis, confirming the effectiveness of our models. Matrix simulation was used to ascertain the efficacy of OP control in aesthetic improvements. Intriguingly, though FHR change rate appeared to be constant across groups, in FA models, hypodivergent group displayed more sensitive changes to SN-OP than normodivergent, hypodivergent group, and Class III group significantly showed larger changes than Class I and II.ConclusionsRotation of OP could yield differently to facial aesthetic improvements as more efficient in hypodivergent groups vertically and Class III groups sagittally.

Intelligence detection of oil absorption in French fries by surface profiles

Surface profiles are important evaluation indices for oil absorption behavior of fried foods. This research established two intelligent models of partial least-squares regression (PLSR) and back propagation artificial neural network (BP-ANN) for monitoring the oil absorption behavior of French fries based on the surface characteristics. Surface morphology and texture of French fries by rapeseed oil (RO) and high-oleic peanut oil (HOPO) at different temperatures were investigated. Results showed that oil content of samples increased with frying temperature, accounting for 37.7% and 41.4% of samples fried by RO and HOPO respectively. The increase of crust ratio, roughness and texture parameters (Fm, Nwr, fwr, Wc) and the decrease of uniformity were observed with the frying temperature. Coefficients of prediction set of PLSR and BP-ANN models were more than 0.93, which indicated that surface features combined with chemometrics were rapid and precise methods for determining the oil content of French fries.

Bioenergy potential from agricultural by-product in 2030: An AI-based spatial analysis and climate change scenarios in a Chinese region

Addressing the pressing need for sustainable bioenergy sources amid the growing climate crisis, this study investigates potential crop bioenergy output in the Jianghan Plain, Hubei, China. The research leverages Backpropagation Artificial Neural Network models to predict the accumulated net primary productivity (NPP) of various crops, i.e., single rice, double rice, maize, and wheat, in Jianghan Plain. The models highlight a strong predictive performance, with the SE model achieving an R2 of 0.83 and wheat model an R2 of 0.90, implying these models can be utilized reliably for future NPP forecasting. Our evaluation of past and future NPP values reveals a negative correlation between the increase in NPP values and observed temperature trends. This emphasizes the significant impact of temperature on crop bioenergy potential. For future NPP predictions, scenario (S3a) under Temperature Control strategies at medium-effort mitigation yields a favourable outcome for all crops, particularly SE. Additionally, the results suggest that aggressive decarbonization and global warming mitigation strategies are vital for enhancing NPP values. However, the study also highlights potential trade-offs and stresses the necessity for region-specific climate reconciliation strategies; for instance, midland counties benefit from temperature control, whereas the sideline favours CO2 mitigation. The study further investigates the spatial distribution of potential bioenergy, with the southwestern part of the plain holding more bioenergy potential, providing an essential reference for future energy planning. By 2030, in Jianghan Plain, when fully integrated with reasonable climate responses, there could be substantial bioenergy potential from crop residues reaching (1.2–1.9) × 109 MJ, which is 11% less than the current.

Food safety testing by negentropy-sorted kernel independent component analysis based on infrared spectroscopy

In the field of food safety testing, variety, brand, origin, and adulteration are four important factors. In this study, a novel food safety testing method based on infrared spectroscopy is proposed to investigate these factors. Fourier transform infrared spectroscopy data are analyzed using negentropy-sorted kernel independent component analysis (NS-kICA) as the feature optimization method. To rank the components, negentropy is performed to measure the non-Gaussian independent components. In our experiment, the proposed method was run on four datasets to comprehensively investigate the variety, brand, origin, and adulteration of agricultural products. The experimental results show that NS-kICA outperforms conventional feature selection methods. The support vector machine model outperforms the backpropagation artificial neural network and partial least squares models. The combination of NS-kICA and support vector machine (SVM) is the best method for achieving high, stable, and efficient recognition performance. These findings are of great importance for food safety testing.

Predicting the medium-temperature thermal stability of impact-strengthened medium-thick plate aluminum alloy using a back propagation artificial neural network

A normalized medium-thick plate of aluminum alloy (4038) was impact-strengthened using a free-fall method at room temperature (approximately 20 °C). Specimens were then aged at 450 °C, 550 °C and 650 °C for 10, 20, 30 and 40 min respectively. Micro-hardness of each sample was tested. Micro-structure of samples annealed at 650 °C for different durations was characterized. A three-layer back propagation artificial neural network (BPANN) was trained using actual state parameters of the prepared samples. Results reveal that medium-temperature thermal stability of the prepared plate can be predicted through the BPANN model. Deviation of predicted values from the experimental ones is within 6 %, with a prediction accuracy exceeding 94 %. Variation trend of the predicted and the experimental thermal stability is consistent, but the predicted values are all higher than the measurements. Prediction accuracy of BPANN can be improved by increasing convergence rate of the error function. By adding relevant parameters of the micro-structure from samples aged at 650 °C to the input layer, BPANN model further improve its output and approach the real state of samples. The findings of this study can help researchers reduce the number and cost of experiments. The aim of this work was to predict the medium-temperature thermal stability of impact-strengthened normalized medium-thick plate of aluminum alloy annealed at different temperatures, and it also can be used as reference for other similar experiments.

Comparative QSAR Modeling for Predicting Anticancer Potency of Imidazo[4,5-b]Pyridine Derivatives Using GA-MLR and BP-ANN Techniques

Background: Prediction of toxicity of imidazo[4,5-b]pyridine derivatives is carried out using GA-MLR and BPANN methods. Objective: A quantitative structure-property relationship (QSPR) was determined based on methods, including genetic algorithm-multiple linear regression (GA-MLR) and backpropagation artificial neural network (BP-ANN). These methods were employed for modeling and predicting the anticancer potency of imidazo[4,5-b]pyridine derivatives. Materials and Methods: A dataset of imidazo[4,5-b]pyridine derivatives was randomly divided into two groups, training and test sets consisting of 75% and 25% of data points, respectively. The optimized conformation of compounds was obtained using the DFT-B3LYP method and 6-31G* basis sets level with Gaussian 09 software. A large number of molecular descriptors were calculated using Dragon software. The QSAR models were optimized using multiple linear regressions (MLR). Results: The most relevant molecular descriptors were obtained using the genetic algorithm (GA) and backward stepwise regression. The predictive powers of the GA-MLR models were studied using leaveone- out (LOO) cross-validation and an external test set. Conclusion: The obtained results of statistical parameters showed the BP-ANN model to have better performance compared to the GA-MLR model. To assess the predictive ability of QSAR models, many statistical terms, such as correlation coefficient (R2), leave-one-out cross-validation (LOOCV), root mean squared error (RMSE), and external and internal validation were used. The results of validation methods demonstrate the QSAR model to be robust and with high predictivity.

Back propagation artificial neural network (BP-ANN) for prediction of the quality of gamma-irradiated smoked bacon

This study investigated the effect of gamma irradiation on smoked bacon quality during storage and developed a multi-quality prediction model based on gamma irradiation. Gamma irradiation reduced moisture content and improved the microbial safety of smoked bacon. It also accelerated protein and lipid oxidation and altered free amino acids and fatty acids composition. It was effective in slowing down quality deterioration and sensory quality decline during storage. The backpropagation artificial neural network (BP-ANN) model was constructed by using physical and chemical indicators, irradiation dose, and storage time as input variables, and the total number of colonies and sensory scores as output layers. The transfer functions of the input-hidden layer and hidden-output layer were ReLu and Sigmoid, respectively. There were 13 neurons in the hidden layer. Results showed that BP-ANN based on physical and chemical indicators, irradiation dose, and storage time had great potential in predicting the multiple quality of smoked bacon.

Infilling of missing data in groundwater pollution prediction models using statistical methods

ABSTRACT Missing data is ubiquitous in hydrology. This phenomenon poses difficulty in the development of data-driven models. Events of missing data in groundwater pollution monitoring networks may occur due to failure of recording devices, malfunctioning of sensors, etc. Handling such missing data implies filling the missing portions of the data structure. Though several studies are available for dealing with missing data in the field of hydrology, literature dealing with such scenarios in groundwater pollution prediction is scarce. This paper assesses four imputation techniques – viz. linear, cubic spline, piece-wise cubic Hermite and modified Akima with cubic Hermite interpolation methods – for developing groundwater pollution prediction models using artificial neural network (ANN). The study employs the development of cascade-forward back-propagation ANN models using missing data ranging from 5% to 75% and evaluating their performance. Results show that imputation techniques can be effective in such circumstances.

A Multi-Factor Combination Model for Medium to Long-Term Runoff Prediction Based on Improved BP Neural Network

Taking a certain coastal area of Jiangsu province as the research background, this study scientifically predicts the runoff on the medium and long-term time scale according to the changes of various climate factors such as atmospheric circulation, sea surface temperature, and solar activity in the first half of the year. A lag correlation is established between various related climate factors and the monthly runoff process in the research area for the previous 1–6 months. Selecting advantageous factors and constructing a significant factor set. Using the improved BP (Back-Propagation) artificial neural network model and combining it with the sensitivity analysis method, a specific number of 8-factor combinations are selected from the set of significant factors for medium and long-term runoff prediction. After that, the prediction results are compared with the forecasting effects of two multi-factor combination runoff simulation schemes formed by stepwise regression and Spearman rank correlation methods. The study concluded that the multi-factor combination simulation effect formed through sensitivity analysis was the best. The 20% standard forecast qualification rate of the three schemes is not significantly different. The Mean Absolute Relative Error of the multi-factor combination training and validation periods simulated through sensitivity analysis is the smallest among the three schemes, which are 36.61% and 38.01%, respectively. The Nash Efficiency Coefficient in the validation period is 0.45, which is far better than other schemes and has better generalization ability. The Standard Deviation of Relative Error in the training and validation periods is much smaller than other schemes, and the dispersion of relative errors is the smallest.

Prediction of biological nutrients removal in full-scale wastewater treatment plants using H2O automated machine learning and back propagation artificial neural network model: Optimization and comparison

The effective control of total nitrogen (ETN) and total phosphorus (ETP) in effluent is challenging for wastewater treatment plants (WWTPs). In this work, automated machine learning (AutoML) (mean square error = 0.4200 ∼ 3.8245, R2 = 0.5699 ∼ 0.6219) and back propagation artificial neural network (BPANN) model (mean square error = 0.0012 ∼ 6.9067, R2 = 0.4326 ∼ 0.8908) were used to predict and analyze biological nutrients removal in full-scale WWTPs. Interestingly, BPANN model presented high prediction performance and general applicability for WWTPs with different biological treatment units. However, the AutoML candidate models were more interpretable, and the results showed that electricity carbon emission dominated the prediction. Meanwhile, increasing data volume and types of WWTP hardly affected the interpretable results, demonstrating its wide applicability. This study demonstrated the validity and the specific advantages of predicting ETN and ETP using H2O AutoML and BPANN model, which provided guidance on the prediction and improvement of biological nutrients removal in WWTPs.

Prediction of membrane purification by membrane fouling based on mathematic and machine learning models combined with image processing technology

Apples have been widely cultivated due to their multiple functional nutrients such as polyphenols. In this study, polyphenol of apple skins were extracted and purified by ultrasound treatment and membrane filtration, respectively. The effects of feed liquid concentration, transmembrane pressure, and rotational speed on the purification of polyphenols by membrane filtration were mainly investigated. The efficiency of polyphenol purification was analyzed based on the particle deposition effect, steric rejection effect, and pectin sieving effect. Single and multi-stage Hermia models were applied to corroborate the aforementioned theory. The degree of membrane fouling was quantitatively analyzed using a self-written image processing program. The study compared the quantitative prediction models between the degree of membrane fouling and polyphenol purification using mathematical models and two machine learning (ML) models, namely Support Vector Machine Regression (SVR) and Back Propagation Artificial neural networks (BPNN). The degree of membrane fouling and the degree of polyphenol purification were used as input parameters and output parameters of the ML prediction model, respectively. The ratio of the training set (98 data points) and the test set (42 data points) was 7:3. The obtained results showed that the prediction efficiency of BPNN model (R2 =0.99968) outperformed the SVR model (R2 =0.97697), which was better than the mathematical model in terms of prediction performance due to the high generalization ability of the ML model and the high robustness of BPNN. The study provided an effective and accurate method for predicting membrane fouling and filtration purification. Thus, the use of ML as an artificial intelligence technique has a promising future in membrane separation and purification.

Geo-spatial intelligence for searching and rescuing man overboard incidents using an artificial neural network: An empirical study of the Royal Thai Navy in the Gulf of Thailand

Man overboard (MOB) incidents happened continuously and they are the main type of suffering incident at the sea. From the MOB situation, searching and rescuing (SAR) operations are occurring in global coastal countries. According to the literature reviews on SAR operations, most studies presented the methodology to search the MOB in better performance indicators after the MOB occurred however, no evidence has been found on using Geo-spatial intelligence with Artificial Neural Networks (ANN) to predict MOB incidents under various weather conditions before the MOB occurred. The objective of this research, therefore, is to propose an intelligent methodology to reduce the searching and rescuing operation time of MOB incidents. A model of Artificial Neural Network (ANN) is applied due to its ability to accurately predict location and practicality of the Royal Thai Navy (RTN). The MOB incidents from 2019 to 2022 are collected and generated for the methodology implementation. This research utilizes the advantage of the Back-Propagation Artificial Neural Network (BP-ANN) to apply in prediction model in which the input variables are various weather condition information including the location of MOB from Geo-spatial, and the output variables are MOB location in the form of Latitude and Longitude. The appropriate BP-ANN model is created under the trial and error method, and the results show that the BP-ANN can predict the location of the MOB accurately. The benefit of this research is to reduce the time consumed by SAR operations in the Gulf of Thailand. As a result, when the traditional SAR operation time is compared with the SAR operation time by using BP-ANN predicting model, it was found that the SAR operation time by using BP-ANN predicting model is more accurate and can reduce the operation time by 5 hrs.

The Analysis and Rapid Non-Destructive Evaluation of Yongchuan Xiuya Quality Based on NIRS Combined with Machine Learning Methods

This paper attempts to analyze and assess Yongchuan Xiuya tea quality quickly, accurately, and digitally. The sensory evaluation method was first used to assess Yongchuan Xiuya tea quality, and then near infrared spectroscopy (NIRS) was obtained, and standard methods were applied to the testing of the chemical components. Next, principal component analysis (PCA) and the correlation coefficient method were used to comprehensively screen out the representative components. Finally, NIRS combined with partial least squares regression (PLSR) and back propagation artificial neural network (BP-ANN) methods were applied to build quality evaluation models for Yongchuan Xiuya tea, respectively, and external samples were employed to examine the practical application results of the best model. The cumulative variance contribution rate of the first three principal components of the ingredients in tea was 97.73%. Seven components closely related to tea quality were screened out, namely, amino acids, total catechin, epigallocatechin gallate (EGCG), tea polyphenols, water extracts, epicatechin gallate (ECG), and epigallocatechin (EGC) (p < 0.01). Between the two models established to predict the tea quality, the model built by the PLS method had the better results, whose coefficient of determination of prediction (Rp2) and root mean square error of prediction (RMSEP) were 0.7955 and 1.2263, respectively, and the best results were obtained by the nonlinear BP-ANN model, whose Rp2 and RMSEP were 0.9315 and 0.6787, respectively. The 10 external Yongchuan Xiuya samples were employed to test the best BP-ANN model, and the results of R2 and RMSEP were 0.9579 and 0.6086, respectively, meaning that the model has good robustness. Therefore, the model established by NIRS combined with the BP-ANN method can be used to assess Yongchuan Xiuya tea quality rapidly, accurately, and digitally, and it can also provide new ideas and methods for evaluating the quality of other teas.

Optimization of effluents using a neural network in the treatment of industrial wastewater

The growth of the planet's population leads to an increase in the problem of access to fresh water. The main sources of water on Earth are brackish and sea water. In connection with the water crisis, water purification becomes an extremely important process, and its achievement is carried out through desalination and various methods of water treatment. In this context, research into the possibility of using neural networks to improve the operation of sewage treatment plants is necessary. The purpose of the research was to optimize and analyze the efficiency of the work of treatment facilities in the treatment of industrial wastewater. Soft computing methods were used to optimize the proposed models. In this study, the exact results of the application of the neural network were determined using analytical and comparative approaches. Treatment of all wastewater and waste generated in the treatment industry involves a number of processes including air flotation, chemical coagulation, settling and biological treatment using fully mixed activated sludge. Various learning functions have been considered, including forward-propagation artificial neural networks (ANNs) such as multilayer perceptron (MLP), cascaded forward-propagation ANNs, and support vector regression (SVR) models. The learning process includes the use of Levenberg-Marquardt optimization algorithms and sequential minimum. The article also provides graphical images illustrating the different types of pollutants, the costs associated with treatment plants, and the color changes in wastewater observed after the treatment process. The obtained results show a high degree of similarity between the predicted and experimental data, which emphasizes the effectiveness of the backpropagation ANN model for accurate predictions. In addition, the integration of machine learning into the production of detergents can be extremely effective in promoting the efficient and sustainable use of water resources. Overall, the paper provides valuable insights into the use of machine learning to address freshwater scarcity

Predicting the Temperature-Dependent Long-Term Creep Mechanical Response of Silica Sand-Textured Geomembrane Interfaces Based on Physical Tests and Machine Learning Techniques.

Preciously assessing the creep mechanical response of sand-geomembrane interfaces is vital for the design of relevant engineering applications, which is inevitable to be influenced by temperature and stress statuses. In this paper, based on the self-developed temperature-controlled large interface shear apparatus, a series of long-term creep shear tests on textured geomembrane-silica sand interfaces in different temperatures, normal pressure, and creep shear pressure were conducted, and a database compiled from the physical creep shear test results is constructed. By adopting the database, three disparate machine learning algorithms of the Back Propagation Artificial Neural Network (BPANN), the Support Vector Machine (SVM) and the Extreme Learning Machine (ELM) were adopted to assess the long-term creep mechanical properties of sand-geomembrane interfaces while also considering the influence of temperature. Then, the forecasting results of the different algorithms was compared and analyzed. Furthermore, by using the optimal machine learning model, sensitivity analysis was carried out. The research indicated that the BPANN model has the best forecasting performance according to the statistics criteria of the Root-Mean-Square Error, the Correlation Coefficient, Wilmot's Index of Agreement, and the Mean Absolute Percentage Error among the developed models. Temperature is the most important influence factor on the creep interface mechanical properties, followed with time. The research findings can support the operating safety of the related engineering facilities installed with the geomembrane.

Statistical and Artificial Neural Network Coupled Technique for Prediction of Tribo-Performance in Amine-Cured Bio-Based Epoxy/MMT Nanocomposites

This study explores the effects of four independent variables—the nanoclay weight percentage, sliding velocity, load, and sliding distance—on the wear rate and frictional force of nanoclay-filled FormuLITETM amine-cured bio-based epoxy composites. An experimental design based on the Taguchi method revealed diverging optimal conditions for minimizing the wear and frictional force. These observations were further validated using a Back-propagation Artificial Neural Network (BPANN) model, demonstrating its proficiency in predicting complex system behavior. Material characterization, conducted through Scanning Electron Microscopy (SEM) and Energy-dispersive X-ray Spectroscopy (EDS), illustrated the homogeneous distribution of the nanoclay within the FormuliteTM matrix, which is crucial for enhancing the load transfer and stress distribution. Atomic Force Microscopy (AFM) analysis indicated that the incorporation of nanoclay increases the surface roughness and peak height, which are important determinants of the material performance. However, an increase in the nanoclay percentage decreased these attributes, suggesting an interaction saturation point. Due to their augmented mechanical properties, the present study underscores the potential of amine-cured bio-based epoxy systems in diverse applications, such as automotive, aerospace, and biomedical engineering.

Estimating Forest Aboveground Biomass Combining Pléiades Satellite Imagery and Field Inventory Data in the Peak–Cluster Karst Region of Southwestern China

The mountainous region of southwest China has the largest karst geomorphology in China and in the world. Quantifying the forest aboveground biomass in this karst region is of great significance for the investigation of carbon storage and carbon cycling in terrestrial ecosystems. In this study, the actual measured aboveground biomass was calculated based on the allometric functions of 106 quadrats from 2012 to 2015. A backpropagation artificial neural network (BPANN) inversion model was constructed by combining very high-resolution satellite imagery, field inventory data, and land use/land cover data to estimate the forest aboveground biomass in the Banzhai watershed, a typical peak–cluster karst basin in southern Guizhou Province. We used 70% of the actual measured aboveground biomass for training the BPANN model, 20% for accuracy verification, and 10% to prevent overtraining. The results show that the absolute root mean square error of the BPANN model was 11.80 t/ha, which accounted for 9.92% of the mean value of aboveground biomass. Based on the BPANN inversion model, the average value of the forests’ aboveground biomass was 135.63 t/ha. The results showed that our study presented a quick, easy, and relatively high-precision method for estimating forest aboveground biomass in the Banzhai watershed. This indicates that the Pléiades image-based BPANN model displayed satisfactory results for estimating the forests’ aboveground biomass in a typical peak–cluster karst basin. This method can be applied to the estimation of forest AGB in the karst mountainous areas of southwest China.

Artificial Neural Network Analysis of Determinants of Tacrolimus Pharmacokinetics in Liver Transplant Recipients.

The efficacy and toxicity of tacrolimus are closely related to its trough blood concentrations. Identifying the influencing factors of pharmacokinetics of tacrolimus in the early postoperative period is conducive to the optimization of the individualized tacrolimus administration protocol and to help liver transplant (LT) recipients achieve the target blood concentrations. This study aimed to develop an artificial neural network (ANN) for predicting the blood concentration of tacrolimus soon after liver transplantation and for identifying determinants of the concentration based on Shapley additive explanation (SHAP). In this retrospective study, we enrolled 31 recipients who were first treated with liver transplantation from the Department of Liver Transplantation and Hepatic Surgery, the First Affiliated Hospital of Shandong First Medical University (Shandong Provincial Qianfoshan Hospital) from November 2020 to May 2021. The basic information, biochemical indexes, use of concomitant drugs, and genetic factors of organ donors and recipients were used for the ANN model inputs, and the output was the steady-state trough concentration (C0) of tacrolimus after oral administration in LT recipients. The ANN model was established to predict C0 of tacrolimus, SHAP was applied to the trained model, and the SHAP value of each input was calculated to analyze quantitatively the influencing factors for the output C0. A back-propagation ANN model with 3 hidden layers was established using deep learning. The mean prediction error was 0.27 ± 0.75 ng/mL; mean absolute error, 0.60 ± 0.52 ng/mL; correlation coefficient between predicted and actual C0 values, 0.9677; and absolute prediction error of all blood concentrations obtained by the ANN model, ≤3.0 ng/mL. The results indicated that the following factors had the most significant effect on C0: age, daily drug dose, genotype at CYP3A5 polymorphism rs776746 in both recipient and donor, and concomitant use of caspofungin. The predicted C0 value of tacrolimus in LT recipients increased in a dose-dependent manner when the daily dose exceeded 3 mg, whereas it decreased with age when LT recipients were older than 48 years. The predicted C0 was higher when recipients and donors had the genotype CYP3A5*3*3 than when they had the genotype CYP3A5*1. The predicted C0 value also increased with the use of caspofungin or Wuzhi capsule. The established ANN model can be used to predict the C0 value of tacrolimus in LT recipients with high accuracy and good predictive ability, serving as a reference for personalized treatment in the early stage after liver transplantation.