Fast occupational upper-limb radiation dose prediction using machine learning and Monte Carlo simulation

Improving carbon flux estimation in tea plantation ecosystems: A machine learning ensemble approach

Accurate estimation of reference evapotranspiration (ETo) plays a vital role in irrigation and water resource planning. The Penman–Monteith method recommended by the Food and Agriculture Organization (FAO PM56) is widely used and considered a standard to calculate ETo. However, FAO PM56 cannot be used with limited meteorological variables, so it is compulsory to choose an alternative model for ETo estimation, which requires fewer variables. This study built ten machine learning (ML) models based on multi-function, neural network, and tree-based structure against the FAO PM56 method. For this purpose, time series temperature data on a monthly scale are only used to train ML models. The developed ML models were applied to estimate ETo at different test stations and the obtained results were compared with the FAO PM56 method to verify and validate their performance in ETo estimation for the selected stations. In addition, multiple statistical indicators, including root-mean-square error (RMSE), coefficient of determination (R2), mean absolute error (MAE), Nash–Sutcliffe efficiency (NSE), and correlation coefficient (r) were calculated to compare the performance of each ML model on ETo estimation. Among the applied ML models, the ETo tree boost (TB) ML model outperformed the other ML models in estimating ETo in diverse climatic conditions based on statistical indicators (R2, NSE, r, RMSE, and MAE). Moreover, the observed R2, NSE, and r were the highest for the TB ML model, while RMSE and MAE were found to be the lowest at the study sites compared to other applied ML models. Lastly, ETo point data yielded from the TB ML model was used in an interpolation process to create monthly and annual ETo maps. Based on the ETo maps, this study suggests mainly a focus on areas with high ETo values and proper irrigation scheduling of crops to ensure water sustainability.

The permeability coefficient of soils is an essential measure for designing geotechnical construction. The aim of this paper was to select a highest performance and reliable machine learning (ML) model to predict the permeability coefficient of soil and quantify the feature importance on the predicted value of the soil permeability coefficient with aided machine learning‐based SHapley Additive exPlanations (SHAP) and Partial Dependence Plot 1D (PDP 1D). To acquire this purpose, five single ML algorithms including K‐nearest neighbors (KNN), support vector machine (SVM), light gradient boosting machine (LightGBM), random forest (RF), and gradient boosting (GB) are used to build ML models for predicting the permeability coefficient of soils. Performance criteria for ML models include the coefficient of correlation R 2 , root mean square error (RMSE), mean absolute percentage error (MAPE), and mean absolute error (MAE). The best performance and reliable single ML model for predicting the permeability coefficient of soil for the testing dataset is the gradient boosting (GB) model, which has R 2 = 0.971, RMSE = 0.199 × 10 −11 m/s, MAE = 0.161 × 10 −11 m/s, and MAPE = 0.185%. To identify and quantify the feature importance on the permeability coefficient of soil, sensitivity studies using permutation importance, SHapley Additive exPlanations (SHAP), and Partial Dependence Plot 1D (PDP 1D) are performed with the aided best performance and reliable ML model GB. Plasticity index, density > water content, liquid limit, and plastic limit > clay content > void ratio are the order effects on the predicted value of the permeability coefficient. The plasticity index and density of soil are the first priority soil properties to measure when assessing the permeability coefficient of soil.

The sea surface temperature (SST) is an environmental indicator closely related to climate, weather, and atmospheric events worldwide. Its forecasting is essential for supporting the decision of governments and environmental organizations. Literature has shown that single machine learning (ML) models are generally more accurate than traditional statistical models for SST time series modeling. However, the parameters tuning of these ML models is a challenging task, mainly when complex phenomena, such as SST forecasting, are addressed. Issues related to misspecification, overfitting, or underfitting of the ML models can lead to underperforming forecasts. This work proposes using hybrid systems (HS) that combine (ML) models using residual forecasting as an alternative to enhance the performance of SST forecasting. In this context, two types of combinations are evaluated using two ML models: support vector regression (SVR) and long short-term memory (LSTM). The experimental evaluation was performed on three datasets from different regions of the Atlantic Ocean using three well-known measures: mean square error (MSE), mean absolute percentage error (MAPE), and mean absolute error (MAE). The best HS based on SVR improved the MSE value for each analyzed series by 82.26%, 98.93%, and 65.03% compared to its respective single model. The HS employing the LSTM improved 92.15%, 98.69%, and 32.41% concerning the single LSTM model. Compared to literature approaches, at least one version of HS attained higher accuracy than statistical and ML models in all study cases. In particular, the nonlinear combination of the ML models obtained the best performance among the proposed HS versions.

Prediction of deflections of reinforced concrete (RC) flexural structures is vital to evaluate the workability and safety of structures during its life cycle. Empirical methods are limited to predict a long-term deflection of RC structures because they are difficult to consider all influencing factors. This study presents data-driven machine learning (ML) models to early predict the long-term deflections in RC structures. An experimental dataset was used to build and evaluate single and ensemble ML models. The models were trained and tested using the stratified 10-fold cross-validation algorithm. Analytical results revealed that the ML model is effective in predicting the deflection of RC structures with good accuracy of 0.972 in correlation coefficient (R), 8.190 mm in root mean square error (RMSE), 4.597 mm in mean absolute error (MAE), and 16.749% in mean absolute percentage error (MAPE). In performance comparison against with empirical methods, the prediction accuracy of the ML model improved significantly up to 66.41% in the RMSE and up to 82.04% in the MAE. As a contribution, this study proposed the effective ML model to facilitate designers in early forecasting long-term deflections in RC structures and evaluating their long-term serviceability and safety.

A deep learning (DL) model has been developed to estimate patient-lens dose in real-time for given exposure and geometric conditions during fluoroscopically-guided neuro-interventional procedures. Parameters input into the DL model for dose prediction include the patient head shift from isocenter and cephalometric landmark locations as a surrogate for head size. Machine learning (ML) models were investigated to automatically detect these parameters from the in-procedure fluoroscopic image. Fluoroscopic images of a Kyoto Kagaku anthropomorphic head phantom were taken at various known X (transverse) and Y (longitudinal) shifts, as well as different magnification modes, to create an image database. For each image, anatomical landmark coordinate locations were obtained manually using ImageJ and are used as ground-truth labels for training. This database was then used to train the two separate ML models. One ML model predicts the patient head shift in both the X and Y directions and the other model predicts the coordinates of the anatomical landmarks. From the coordinates, the distance between these anatomical landmarks is calculated, and input into the DL dose-prediction model. Model performance was evaluated using mean absolute error (MAE) and mean absolute percentage error (MAPE) for the head-shift and landmark-coordinate models, respectively. The goal is to implement these two separate models into the Dose Tracking System (DTS) developed by our group. This would allow the DTS to automatically detect the patient head size and position for eye-lens dose prediction and eliminate the need for manual input by the clinical staff.

Simple SummaryThis work describes the development of a fast and accurate machine learning (ML) 3D U-Net dose engine, trained with Monte Carlo (MC) radiation transport simulations, to calculate the dose in rat patients treated in Microbeam Radiation Therapy (MRT) preclinical studies at the Imaging and Medical Beamline at the Australian Synchrotron. Digital phantoms are created based on CT scans of sixteen rats and are augmented to obtain enough anatomical data. Augmented variations of the digital phantoms are then used to simulate with Geant4 the energy depositions of an MRT beam inside the phantoms with 15% (high-noise) and 2% (low-noise) statistical uncertainty. The high-noise MC simulations are used for ML model training and validation, while the low-noise ones for testing. The results show that the ML dose engine provides a satisfactory dose description in the tumor target and generates the dose maps in less than one second.Microbeam radiation therapy (MRT) utilizes coplanar synchrotron radiation beamlets and is a proposed treatment approach for several tumor diagnoses that currently have poor clinical treatment outcomes, such as gliosarcomas. Monte Carlo (MC) simulations are one of the most used methods at the Imaging and Medical Beamline, Australian Synchrotron to calculate the dose in MRT preclinical studies. The steep dose gradients associated with the 50m-wide coplanar beamlets present a significant challenge for precise MC simulation of the dose deposition of an MRT irradiation treatment field in a short time frame. The long computation times inhibit the ability to perform dose optimization in treatment planning or apply online image-adaptive radiotherapy techniques to MRT. Much research has been conducted on fast dose estimation methods for clinically available treatments. However, such methods, including GPU Monte Carlo implementations and machine learning (ML) models, are unavailable for novel and emerging cancer radiotherapy options such as MRT. In this work, the successful application of a fast and accurate ML dose prediction model for a preclinical MRT rodent study is presented for the first time. The ML model predicts the peak doses in the path of the microbeams and the valley doses between them, delivered to the tumor target in rat patients. A CT imaging dataset is used to generate digital phantoms for each patient. Augmented variations of the digital phantoms are used to simulate with Geant4 the energy depositions of an MRT beam inside the phantoms with 15% (high-noise) and 2% (low-noise) statistical uncertainty. The high-noise MC simulation data are used to train the ML model to predict the energy depositions in the digital phantoms. The low-noise MC simulations data are used to test the predictive power of the ML model. The predictions of the ML model show an agreement within 3% with low-noise MC simulations for at least 77.6% of all predicted voxels (at least 95.9% of voxels containing tumor) in the case of the valley dose prediction and for at least 93.9% of all predicted voxels (100.0% of voxels containing tumor) in the case of the peak dose prediction. The successful use of high-noise MC simulations for the training, which are much faster to produce, accelerates the production of the training data of the ML model and encourages transfer of the ML model to different treatment modalities for other future applications in novel radiation cancer therapies.

In this study, machine learning (ML) models were developed to predict permeability (k), porosity (φ) and water saturation (Sw) using 1241 datasets obtained from well-logs data in the Niger Delta. The datasets were screened to remove incomplete sets and outliers and make them suitable for adequate training using the maximum-minimum normalization approach. Three multiple-input multiple-output (MIMO) machine learning methods, namely artificial neural network (ANN), decision tree (DT) and random forest (RF), were used to train the datasets. Five performance metrics, coefficient of determination (R2), correlation coefficient (R), mean absolute error (MAE), average absolute relative error (AARE), and average relative error (ARE), were used to evaluate the performance of the developed models. The results indicate that the MIMO neural-based model had overall MSE and R values of 1.9801×10-3 and 0.9866, while the DT model had 2.2540×10-3 and 0.98281, and the RF model had 5.1490×10-3 and 0.95989. The ANN model predicted k resulted in R2, R, MAE, ARE, and AARE of 0.95740, 0.97847, 2.0677, -0.0011, and 0.0343, respectively, while the predicted φ had R2 of 0.96336, R of 0.98151, MAE of 0.0055, ARE of -0.0006, and AARE of 0.0185. The predicted Sw had an R2 of 0.98430, R of 0.99212, MAE of 0.0265, ARE of -0.0045, and AARE of 0.0521. Also, the developed DT model predicted k resulted in R2, R, MAE, ARE and AARE of 0.95250, 0.97596, 0.0277, 5.6981 and 0.0382, respectively, while the predicted φ had R2 of 0.9380, R of 0.9685, MAE of 0.0276, ARE of -0.5796 and AARE of 5.8199. The predicted Sw had R2 of 0.99039, R of 0.9518, MAE of 0.0182, ARE of -0.49969 and AARE of 5.0452. Furthermore, the developed RF model predicted k resulted in R2, R, MAE, ARE, and AARE of 0.88438, 0.94041, 0.0552, -6.8754 and 15.8391, respectively, while the predicted φ had R2 of 0.90377, R of 0.95067, MAE of 0.0504, ARE of -5.3429 and AARE of 12.8260. The predicted Sw had R2 of 0.95495, R of 0.97722, MAE of 0.0469, ARE of -25.1422 and AARE of 32.6698. The relative importance of the ML input parameters on the predicted outputs is RES>D>GR>VSh>RHOB>NPHI>CALI. Based on the statistical indicators obtained, the predictions of the developed ML-based models were close to the actual field datasets. Thus, the ML-based models should be used as tools for predicting k, φ and Sw in the Niger Delta.

Predicting operative time is essential for scheduling surgery and managing the operating room. This study aimed to develop machine learning (ML) models to predict the operative time for metabolic and bariatric surgery (MBS) and to compare each model. The authors used the Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program database between 2016 and 2020 to develop ML models, including linear regression, random forest, support vector machine, gradient-boosted tree, and XGBoost model. Patient characteristics and surgical features were included as variables in the model. The authors used the mean absolute error, root mean square error, and R 2 score to evaluate model performance. The authors identified the 10 most important variables in the best-performing model using the Shapley Additive exPlanations algorithm. In total, 668723 patients were included in the study. The XGBoost model outperformed the other ML models, with the lowest root mean square error and highest R 2 score. Random forest performed better than linear regression. The relative performance of the ML algorithms remained consistent across the models, regardless of the surgery type. The surgery type and surgical approach were the most important features to predict the operative time; specifically, sleeve gastrectomy (vs. Roux-en-Y gastric bypass) and the laparoscopic approach (vs. robotic-assisted approach) were associated with a shorter operative time. The XGBoost model best predicted the operative time for MBS among the ML models examined. Our findings can be useful in managing the operating room scheduling and in developing software tools to predict the operative times of MBS in clinical settings.

This study developed an ML (machine learning) model that predicts the vegetation distribution of the following year from the current year's conditions by applying the ML model in multiple stages. The target rivers examined in this study were five Japanese large rivers, i.e., Kinugawa, Edogawa, Yahagigawa, Shonaigawa, and Ibogawa. The multi-stage ML model's explanatory and target variables were created for each river segment using DEMs (Digital Elevation Models) and river environment base maps. The multi-stage ML model consisted of three ML stages to predict the vegetation distribution of the following year from the current river vegetation distribution and topographical information. The advantage of the multi-stage ML model was that a third-stage vegetation distribution prediction model could be constructed according to the difficulty of prediction using a second-stage classification result. XGB (eXtreme Gradient Boosting) was used as the machine learning model. SHAP (SHapley Additive exPlanations) was used for factor analysis. F1 score with five-fold cross-validation was used to evaluate the model's accuracy. The result of the multi-stage ML model for the five target rivers showed that the F1 score was 0.8 or higher for all rivers except the Kinugawa River. The multi-stage ML model had an accuracy of 10% higher F1 score than a conventional single ML model. The vegetation distribution probability map indicated that the prediction had a high general accuracy but dropped near the boundary between the river's low water channel and the floodplain. SHAP analysis revealed the three prominent factors for vegetation existence: (i) the relative height near the levee and in the center of the floodplain, (ii) the distance from the river water's edge near the low water channel, and (iii) the vegetation existence history at the boundary between the low water channel and the floodplain. These results suggest that combining the prediction map with factor analysis could identify the factors that significantly influence where vegetation recruits in a river course.

The use of recycled glass in the concrete mix instead of natural coarse aggregates and supplemental cementitious material has several advantages, including the conservation of natural resources, the reduction of CO2 emissions, and cost savings. However, due to their qualities, the mechanical properties of concrete containing Ground Glass Particles (GGP) differ from those of natural aggregates concrete. As a result, assessing the compressive strength (CS) of concrete with GGP is crucial. Therefore, this paper proposes the hybrid Machine Learning (ML) model including the Gradient Boosting (GB) and Bayesian optimization (BO) algorithms for predicting the compressive strength of concrete containing GGP. The hybrid ML model is developed and validated based on the training dataset (70% of the data) and the test dataset (30% of the remaining data), respectively. The performance of hybrid ML model is evaluated by three criteria, such as the Pearson correlation coefficient (R), Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). The K-Fold Cross-Validation technique is also used to verify the reliability of the hybrid ML model). The best performance of the hybrid ML model is determined with the R = 0.9843, RMSE = 1.7256 (MPa), and MAE = 1.3154 (MPa) for training dataset and R = 0.9784, RMSE = 2.4338 (MPa) and MAE = 1.9618 (MPa) for testing dataset. Based on the best hybrid ML model, the sensitivity analysis including SHapley Additive exPlanation (SHAP) and Partial Dependence Plots (PDP) 2D are investigated to obtain an in-depth examination of each individual input variable on the predicted compressive strength of concrete contaning GGP. The sensitivity analysis shows that four factors, such as curing age, surface area, TiO2, and temperature have the most effect on the compressive strength of concrete containing GGP.

Multiple studies have attempted to generate visual field (VF) mean deviation (MD) estimates using cross-sectional optical coherence tomography (OCT) data. However, whether such models offer any value in detecting longitudinal VF progression is unclear. We address this by developing a machine learning (ML) model to convert OCT data to MD and assessing its ability to detect longitudinal worsening. Retrospective, longitudinal study. A model dataset of 70,575 paired OCT/VFs to train an ML model converting OCT to VF-MD. A separate progression dataset of 4,044 eyes with ≥ 5 paired OCT/VFs to assess the ability of OCT-derived MD to detect worsening. Progression dataset eyes had two additional unpaired VFs (≥ 7 total) to establish a "ground truth" rate of progression defined by MD slope. We trained an ML model using paired VF/OCT data to estimate MD measurements for each OCT scan (OCT-MD). We used this ML model to generate longitudinal OCT-MD estimates for progression dataset eyes. We calculated MD slopes after substituting/supplementing VF-MD with OCT-MD and measured the ability to detect progression. We labeled true progressors using a ground truth MD slope <0.5 dB/year calculated from ≥ 7 VF-MD measurements. We compared the area under the curve (AUC) of MD slopes calculated using both VF-MD (with <7 measurements) and OCT-MD. Because we found OCT-MD substitution had a statistically inferior AUC to VF-MD, we simulated the effect of reducing OCT-MD mean absolute error (MAE) on the ability to detect worsening. AUC. OCT-MD estimates had an MAE of 1.62 dB. AUC of MD slopes with partial OCT-MD substitution was significantly worse than the VF-MD slope. Supplementing VF-MD with OCT-MD also did not improve AUC, regardless of MAE. OCT-MD estimates needed an MAE ≤ 1.00 dB before AUC was statistically similar to VF-MD alone. ML models converting OCT data to VF-MD with error levels lower than published in prior work (MAE: 1.62 dB) were inferior to VF-MD data for detecting trend-based VF progression. Models converting OCT data to VF-MD must achieve better prediction errors (MAE ≤ 1 dB) to be clinically valuable at detecting VF worsening.

Developing machine learning models with multisource inputs for improved land surface soil moisture in China

PD27-01 DEVELOPMENT OF A MACHINE LEARNING (ML) MODEL TO AUTOMATICALLY AND PRECISELY IDENTIFY KIDNEY STONES FROM URETEROSCOPY VIDEO RECORDINGS

Clinically explainable machine learning models for early identification of patients at risk of hospital-acquired urinary tract infection