Random Forest Feature Selection Research Articles

A River Water Quality Prediction Method Based on Dual Signal Decomposition and Deep Learning

Traditional single prediction models struggle to address the complexity and nonlinear changes in water quality forecasting. To address this challenge, this study proposed a coupled prediction model (RF-TVSV-SCL). The model includes Random Forest (RF) feature selection, dual signal decomposition (Time-Varying Filtered Empirical Mode Decomposition, TVF-EMD, and Sparrow Search Algorithm-Optimized Variational Mode Decomposition, SSA-VMD), and a deep learning predictive model (Sparrow Search Algorithm-Convolutional Neural Network-Long Short-Term Memory, SSA-CNN-LSTM). Firstly, the RF method was used for feature selection to extract important features relevant to water quality prediction. Then, TVF-EMD was employed for preliminary decomposition of the water quality data, followed by a secondary decomposition of complex Intrinsic Mode Function (IMF) components using SSA-VMD. Finally, the SSA-CNN-LSTM model was utilized to predict the processed data. This model was evaluated for predicting total phosphorus (TP), total nitrogen (TN), ammonia nitrogen (NH3-N), dissolved oxygen (DO), permanganate index (CODMn), conductivity (EC), and turbidity (TB), across 1, 3, 5, and 7-d forecast periods. The model performed exceptionally well in short-term predictions, particularly within the 1–3 d range. For 1-, 3-, 5-, and 7-d forecasts, R2 ranged from 0.93–0.96, 0.79–0.87, 0.63–0.72, and 0.56–0.64, respectively, significantly outperforming other comparison models. The RF-TVSV-SCL model demonstrates excellent predictive capability and generalization ability, providing robust technical support for water quality forecasting and pollution prevention.

Explainable 18F-FDG PET/CT radiomics model for predicting EGFR mutation status in lung adenocarcinoma: a two-center study

PurposeTo establish an explainable 18F-FDG PET/CT-derived prediction model to identify EGFR mutation status and subtypes (EGFR wild, EGFR-E19, and EGFR-E21) in lung adenocarcinoma (LUAD).MethodsBaseline 18F-FDG PET/CT images of 478 patients with LUAD from 2 hospitals were collected. Data from hospital A (n = 390) was randomly split into a training group (n = 312) and an internal test group (n = 78), with data from hospital B (n = 88) utilized for external test. Further, a total of 4,760 handcrafted radiomics features (HRFs) were extracted from PET/CT scans. Candidates for the prediction model were constructed by cross-combinations of 11 feature selection methods and 7 classifiers. The optimal model was determined by combining the results of cross-center data validation and model visualization (Yellowbrick). The predictive performance was assessed via receiver operating characteristic curve, confusion matrix and classification report. Four explainable artificial intelligence technologies were used for optimal model interpretation.ResultsSex and SUVmax were selected as clinical risk factors, which were then combined with 8 robust PET/CT HRFs to establish the models. The optimal performance was obtained by combining a light gradient boosting machine classifier with random forest feature selection method achieving an optimal performance with a macro-average AUC of 0.75 in the internal test group and 0.81 in the external test group.ConclusionThe explainable EGFR mutation status prediction model have certain clinical practicability and good generalization performance, which may help in the timely selection of treatment options and prognosis prediction in patients with LUAD.

Analyzing Resampling Techniques for Addressing the Class Imbalance in NIDS using SVM with Random Forest Feature Selection

The purpose of Network Intrusion Detection Systems (NIDS) is to ensure and protect computer networks from harmful actions. A major concern in NIDS development is the class imbalance problem, i.e., normal traffic dominates the communication data plane more than intrusion attempts. Such a state of affairs can pose certain hazards to the effectiveness of detection algorithms, including those useful for detecting less frequent but still highly dangerous intrusions. This paper aims to utilize resampling techniques to tackle this problem of class imbalance in NIDS using a Support Vector Machine (SVM) classifier alongside utilizing features selected by Random Forest to improve the feature subset selection process. The analysis highlights the combativeness of each sampling method, offering insights into their efficiency and practicality for real-world applications. Four resampling techniques are analyzed. Such techniques include Synthetic Minority Over-sampling Technique (SMOTE), Random Under-sampling (RUS), Random Over-sampling (ROS) and SMOTE with two different combinations i.e., RUS SMOTE and RUS ROS. Feature selection was done using Random Forest, which was improved by Bayesian methods to create subsets of features with feature rankings determined by Cumulative Feature Importance Score (CFIS). The CIDDS-2017 dataset is used for the performance evaluation, and the metrics used include accuracy, precision, recall, F-measure and CPU time. The algorithm that performs best overall in the CFIS feature subsets is SMOTE, and the features that give the best result are selected at the 90% level with 25 features. This subset accomplishes a relative accuracy enhancement of 0.08% than the other approaches. The RUS+ROS technique is also fine but somehow slower than SMOTE. On the other hand, RUS+SMOTE shows relatively poor results although it consumes less time in terms of computational time compared to other methods, giving about 50% of the performance shown by the other methods. This paper's novelty is adapting the RUS method as a standalone test for screening new and potentially contaminated datasets. The standalone RUS method is more efficient in terms of computations; the algorithm returned the best result of 98.13% accuracy at 85% at the CFIS level of 34 features with a computation time of 137.812 s. It is also noted that SMOTE is considered to be proficient among all resampling techniques used for handling the problem of class imbalance in NIDS, vice 90% CFIS feature subset. Future research directions could include using these techniques in different data sets and other machine learning and deep learning methods together with ROC curve analysis to provide useful pointers to NIDS designers on how to select the right data mining tools and strategies for their projects.

Enhanced prediction model of short-term sea surface wind speed: A multiscale feature extraction and selection approach coupled with deep learning technique

Accurate prediction of short-term sea surface wind speed is essential for maritime safety and coastal management. Most conventional studies encounter challenges simply in analyzing raw wind speed sequences and extracting multiscale features directly from the original received data, which result in lower efficiency. In this paper, an enhanced hybrid model based on a novel data assemble method for original received data, a multiscale feature extraction and selection approach, and a predictive network, is proposed for accurate and efficient short-term sea surface wind speed forecasting. Firstly, the received original data including wind speed are assembled into correlation matrices in order to uncover inherent associations over varied time spans. Secondly a novel Multiscale Wind-speed Feature-Enhanced Convolutional Network (MW-FECN) is designed for efficient and selective multiscale feature extraction, which can capture comprehensive characteristics. Thirdly, a Random Forest Feature Selection (RF-FS) is employed to pinpoint crucial characteristics for enhanced prediction of wind speed with higher efficiency than the related works. Finally, the proposed hybrid model utilized a Bidirectional Long Short-Term Memory (BiLSTM) network to achieve the accurate prediction of wind speed. Experimental data are collected in Weihai sea area, and a case study consist of five benchmarks and three ablation models is conducted to assess the proposed hybrid model. Compared with the conventional methods, experiment results illustrate the effectiveness of the proposed hybrid model and demonstrate effective balancing prediction accuracy and computational time. The proposed hybrid model achieves up to a 28.45% MAE and 27.27% RMSE improvement over existing hybrid models.

A hybrid intelligent system for protection of transmission lines connected to PV farms based on linear trends

Conventional relays face challenges for transmission lines connected to inverter-based resources (IBRs). In this article, a single-ended intelligent protection of the transmission line in the zone between the grid and the PV farm is suggested. The method employs a fuzzy logic and random forest (RF)-based hybrid system to detect faults based on combined linear trend attributes of the 3-phase currents. The fault location is determined and the faulty phase is detected. RF feature selection is used to obtain the optimal linear trend feature. The performance of the methodology is examined for abnormal events such as faults, capacitor and load-switching operations simulated in PSCAD/EMTDC on IEEE 9-bus system obtained by varying various fault and switching parameters. Additionally, when validating the suggested strategy, consideration is given to the effects of conditions such as the presence of double circuit lines, PV capacity, sampling rate, data window length, noise, high impedance faults, CT saturation, compensation devices, evolving and cross-country faults, and far-end and near-end faults. The findings indicate that the suggested strategy can be used to deal with a variety of system configurations and situations while still safeguarding such complex power transmission networks.

Dense and persistent odor representations in the olfactory bulb of awake mice.

Recording and analysis of neural activity is often biased toward detecting sparse subsets of highly active neurons, masking important signals carried in low magnitude and variable responses. To investigate the contribution of seemingly noisy activity to odor encoding, we used mesoscale calcium imaging from mice of both sexes to record odor responses from the dorsal surface of bilateral olfactory bulbs (OBs). The outer layer of the mouse OB is comprised of dendrites organized into discrete "glomeruli", which are defined by odor receptor-specific sensory neuron input. We extracted activity from a large population of glomeruli and used logistic regression to classify odors from individual trials with high accuracy. We then used add-in and drop-out analyses to determine subsets of glomeruli necessary and sufficient for odor classification. Classifiers successfully predicted odor identity even after excluding sparse, highly active glomeruli, indicating that odor information is redundantly represented across a large population of glomeruli. Additionally, we found that Random Forest feature selection informed by Gini Inequality (RFGI) reliably ranked glomeruli by their contribution to overall odor classification. RFGI provided a measure of "feature importance" for each glomerulus that correlated with intuitive features like response magnitude. Finally, in agreement with previous work, we found that odor information persists in glomerular activity after odor offset. Together, our findings support a model of olfactory bulb odor coding where sparse activity is sufficient for odor identification, but information is widely, redundantly available across a large population of glomeruli, with each glomerulus representing information about more than one odor.Significance statement This study leverages meso-scale imaging and machine learning to investigate how odor information is first represented in the brain. Typically, recordings of neuronal activity focus on active individual cells, potentially overlooking broader variations in neuronal responses across populations. Our results demonstrate that a considerable amount of olfactory information is redundantly distributed across a large proportion of olfactory bulb glomeruli. Even after excluding a majority of glomeruli, odor identification remained possible. These findings indicate that, although a few glomeruli are sufficient for odor recognition, an abundance of additional information is represented across a broad population. Understanding how the brain manages redundant olfactory information will shed light on its adaptive mechanisms for navigating diverse real-world circumstances and responding to fluctuating internal states.

A novel BiGRU multi-step wind power forecasting approach based on multi-label integration random forest feature selection and neural network clustering

Accurate wind power forecasting helps to carry out effective scheduling and scientific management of wind power, and improve the security and reliability of the power grid. However, the intermittency, volatility and instability of wind energy make wind power forecasting challenging. Therefore, in order to improve the accuracy and stability of wind power forecasting, this paper proposes a bidirectional gated recurrent unit (BiGRU) multi-step wind power forecasting approach based on multi-label integration random forest (MLRF) feature selection and neural network clustering (NNClustering). The proposed MLRF method extends the applicability of random forest through multiple criteria and enables feature selection for multi-step forecasting tasks of multi-factor time series to obtain optimal input features and time steps, which reduces the computational cost and improves the generalization ability of the model. The proposed NNClustering method establishes a novel convolution-based clustering structure and adjusts the parameters by gradient descent method to obtain the optimal clustering centers, and the robust data applicability of the method is validated in multiple seasonal experiments. The WOA-BiGRU forecasting model is constructed separately for each cluster, which reduces the modeling difficulty and better extracts the characteristics. The BiGRU model extracts more efficient characteristics by processing sequences in both directions and the important parameters of BiGRU are optimized by the whale optimization algorithm (WOA) to obtain the optimal forecasting model. Experimental results over multiple seasons show that the proposed hybrid approach has good forecasting performance and robustness, which provides a novel and efficient solution for wind power forecasting.

Loss of consciousness detection model for smart triage

Falling is one of the most critical outcomes of loss of consciousness during triage in emergency department (ED). It is an important sign requires an immediate medical intervention. This paper presents a computer vision-based fall detection model in ED. In this study, we hypothesis that the proposed vision-based triage fall detection model provides accuracy equal to traditional triage system (TTS) conducted by the nursing team. Thus, to build the proposed model, we use MoveNet, a pose estimation model that can identify joints related to falls, consisting of 17 key points. To test the hypothesis, we conducted two experiments: In the deep learning (DL) model we used the complete feature consisting of 17 keypoints which was passed to the triage fall detection model and was built using Artificial Neural Network (ANN). In the second model we use dimensionality reduction Feature-Reduction for Fall model (FRF), Random Forest (RF) feature selection analysis to filter the key points triage fall classifier. We tested the performance of the two models using a dataset consisting of many images for real-world scenarios classified into two classes: Fall and Not fall. We split the dataset into 80% for training and 20% for validation. The models in these experiments were trained to obtain the results and compare them with the reference model. To test the effectiveness of the model, a t-test was performed to evaluate the null hypothesis for both experiments. The results show FRF outperforms DL model, and FRF has same accuracy of TTS.

BWO-BiLSTM & CNN composite model for prediction of atmospheric particulate matter mass concentration

Accurate prediction of the mass concentration of atmospheric particulate matter is of great significance for the rational formulation of atmospheric environment management strategies and the study of the spatial and temporal evolution of atmospheric pollutants. In order to solve the problems of low prediction accuracy and low efficiency of traditional prediction models, aiming at the nonlinear and stochastic characteristics of atmospheric particulate matter mass concentration changes, a composite prediction model based on Random Forest (RF) feature selection, Beluga Whale Optimization (BWO) algorithm, Convolutional Neural Network (CNN) and Bidirectional Long and Short-Term Storage Memory Neural Network (BiLSTM) is proposed in this paper. In this composite prediction model, the input variables are adjusted and screened through RF algorithm to reduce the network complexity. The weights and thresholds of CNN-BiLSTM are optimized using BWO to improve the prediction accuracy of the model. The publicly mass concentration data and Aerodynamic Particle Size Spectrometer (APS) measurements are used to train and compare with the predicted data. The experimental results indicate that this composite model has better prediction performance and prediction accuracy compared with the traditional single and the combined model. The fitting coefficient (R2) of PM2.5 prediction using publicly data and APS data can reach 0.8842 and 0.9762, respectively. The R2 for the prediction of PM10 using publicly data and APS data can reach 0.8635 and 0.976, respectively. It indicates that the model proposed in this paper has better generalization and robustness.

Research on Estimating Potato Fraction Vegetation Coverage (FVC) Based on the Vegetation Index Intersection Method

The acquisition of vegetation coverage information is crucial for crop field management, and utilizing visible light spectrum vegetation indices to extract vegetation coverage information is a commonly used method. However, most visible light spectrum vegetation indices do not fully consider the relationships between the red, green, and blue bands during their construction, making it difficult to ensure the accurate extraction of coverage information throughout the crop’s entire growth cycle. To rapidly and accurately obtain potato vegetation coverage information, drones were used in this study to obtain high-resolution digital orthoimages of potato growth stages. Based on the differences in the grayscale values of potato plants, soil, shadows, and drip irrigation belts, this study presents a combination index of blue and green bands (BGCI) and a combination index of red and green bands (RGCI). The vegetation index intersection method was used with 10 vegetation information indices to extract vegetation coverage, and the differences in extraction accuracy were compared with those of the maximum entropy method and bimodal histogram method. Based on the high-precision fraction vegetation coverage (FVC) extraction results, the Pearson correlation coefficient method and random forest feature selection were used to screen 10 vegetation and 24 texture features, and the top six vegetation indices most strongly correlated with the FVC were selected for potato growth stage FVC estimation and accuracy verification. A high-precision potato vegetation coverage estimation model was successfully established. This study revealed that during the potato tuber formation and expansion stages, the BGCI combined with the vegetation index intersection method achieved the highest vegetation coverage extraction accuracy, with overall accuracies of 99.61% and 98.84%, respectively. The RGCI combined with the vegetation index intersection method achieved the highest accuracy, 98.63%, during the maturation stage. For the potato vegetation coverage estimation models, the model based on the BGCI achieved the highest estimation accuracy (R2 = 0.9116, RMSE = 5.7903), and the RGCI also achieved good accuracy in terms of vegetation coverage estimation (R2 = 0.8987, RMSE = 5.8633). In the generality verification of the models, the R2 values of the FVC estimation models based on the BGCI and RGCI were both greater than 0.94. A potato vegetation coverage estimation model was constructed based on two new vegetation information indices, demonstrating good accuracy and universality.

Combined conventional factors and the radiomics signature of coronary plaque texture could improve cardiac risk prediction

ObjectivesThis study aims to investigate how radiomics analysis can help understand the association between plaque texture, epicardial adipose tissue (EAT), and cardiovascular risk. Working with a Photon-counting CT, which exhibits enhanced feature stability, offers the potential to advance radiomics analysis and enable its integration into clinical routines.MethodsCoronary plaques were manually segmented in this retrospective, single-centre study and radiomic features were extracted using pyradiomics. The study population was divided into groups according to the presence of high-risk plaques (HRP), plaques with at least 50% stenosis, plaques with at least 70% stenosis, or triple-vessel disease. A combined group with patients exhibiting at least one of these risk factors was formed. Random forest feature selection identified differentiating features for the groups. EAT thickness and density were measured and compared with feature selection results.ResultsA total number of 306 plaques from 61 patients (mean age 61 years +/− 8.85 [standard deviation], 13 female) were analysed. Plaques of patients with HRP features or relevant stenosis demonstrated a higher presence of texture heterogeneity through various radiomics features compared to patients with only an intermediate stenosis degree. While EAT thickness did not significantly differ, affected patients showed significantly higher mean densities in the 50%, HRP, and combined groups, and insignificantly higher densities in the 70% and triple-vessel groups.ConclusionThe combination of a higher EAT density and a more heterogeneous plaque texture might offer an additional tool in identifying patients with an elevated risk of cardiovascular events.Clinical relevance statementCardiovascular disease is the leading cause of mortality globally. Plaque composition and changes in the EAT are connected to cardiac risk. A better understanding of the interrelation of these risk indicators can lead to improved cardiac risk prediction.Key PointsCardiac plaque composition and changes in the EAT are connected to cardiac risk.Higher EAT density and more heterogeneous plaque texture are related to traditional risk indicators.Radiomics texture analysis conducted on PCCT scans can help identify patients with elevated cardiac risk.Graphical

Cross-Combination Analyses of Random Forest Feature Selection and Decision Tree Model for Predicting Intraoperative Hypothermia in Total Joint Arthroplasty

BackgroundIn total joint arthroplasty patients, intraoperative hypothermia (IOH) is associated with perioperative complications and an increased economic burden. Previous models have some limitations and mainly focus on regression modeling. Random forest (RF) algorithms and decision tree modeling are effective for eliminating irrelevant features and making predictions that aid in accelerating modeling and reducing application difficulty. MethodsWe conducted this prospective observational study using convenience sampling and collected data from 327 total joint arthroplasty patients in a tertiary hospital from March 4, 2023, to September 11, 2023. Of those, 229 patients were assigned to the training and 98 to the testing sets. The Chi-square, Mann–Whitney U, and t-tests were used for baseline analyses. The feature variables selection used the RF algorithms, and the decision tree model was trained on 299 examples and validated on 98. The sensitivity, specificity, recall, F1 score, and area under the curve were used to test the model’s performance. ResultsThe RF algorithms identified the preheating time, the volume of flushing fluids, the intraoperative infusion volume, the anesthesia time, the surgical time, and the core temperature after intubation as risk factors for IOH. The decision tree was grown to 5 levels with 9 terminal nodes. The overall incidence of IOH was 42.13%. The sensitivity, specificity, recall, F1 score, and area under the curve were 0.651, 0.907, 0.916, 0.761, and 0.810, respectively. The model indicated strong internal consistency and predictive ability. ConclusionsThe preheating time, the volume of flushing fluids, the intraoperative infusion volume, the anesthesia time, the surgical time, and the core temperature after intubation could accurately predict IOH in total joint arthroplasty patients. By monitoring these factors, the clinical staff could achieve early detection and intervention of IOH in total joint arthroplasty patients.

Which surrogate insulin resistance indices best predict coronary artery disease? A machine learning approach

BackgroundVarious surrogate markers of insulin resistance have been developed, capable of predicting coronary artery disease (CAD) without the need to detect serum insulin. For accurate prediction, they depend only on glucose and lipid profiles, as well as anthropometric features. However, there is still no agreement on the most suitable one for predicting CAD.MethodsWe followed a cohort of 2,000 individuals, ranging in age from 20 to 74, for a duration of 9.9 years. We utilized multivariate Cox proportional hazard models to investigate the association between TyG-index, TyG-BMI, TyG-WC, TG/HDL, plus METS-IR and the occurrence of CAD. The receiver operating curve (ROC) was employed to compare the predictive efficacy of these indices and their corresponding cutoff values for predicting CAD. We also used three distinct embedded feature selection methods: LASSO, Random Forest feature selection, and the Boruta algorithm, to evaluate and compare surrogate markers of insulin resistance in predicting CAD. In addition, we utilized the ceteris paribus profile on the Random Forest model to illustrate how the model’s predictive performance is affected by variations in individual surrogate markers, while keeping all other factors consistent in a diagram.ResultsThe TyG-index was the only surrogate marker of insulin resistance that demonstrated an association with CAD in fully adjusted model (HR: 2.54, CI: 1.34–4.81). The association was more prominent in females. Moreover, it demonstrated the highest area under the ROC curve (0.67 [0.63–0.7]) in comparison to other surrogate indices for insulin resistance. All feature selection approaches concur that the TyG-index is the most reliable surrogate insulin resistance marker for predicting CAD. Based on the Ceteris paribus profile of Random Forest the predictive ability of the TyG-index increased steadily after 9 with a positive slope, without any decline or leveling off.ConclusionDue to the simplicity of assessing the TyG-index with routine biochemical assays and given that the TyG-index was the most effective surrogate insulin resistance index for predicting CAD based on our results, it seems suitable for inclusion in future CAD prevention strategies.

Reservoir properties inversion using attention-based parallel hybrid network integrating feature selection and transfer learning

The inversion of subsurface reservoir properties is of profound significance to the oil and gas energy development and utilization. The strong heterogeneity and complex pore structure of underground reservoirs pose a challenge to efficient oil and gas energy development and utilization across the world, which increases the necessity of developing an efficient reservoir properties inversion method. However, traditional model-driven methods are confronted with the challenges of strong nonlinearity and geological heterogeneity. Moreover, previous studies rarely emphasized the importance of nonlinear feature selection and transfer learning (TL). Aiming to address the research gaps, a reservoir properties inversion method was proposed by combining random forest (RF) feature selection, bidirectional temporal convolutional network (BiTCN), bidirectional gated recurrent units (BiGRU) network, multi-head attention (MHA) mechanism and TL strategy. First, the RF was used to screen the features with significant correlation with the target reservoir properties. Thereafter, combining BiTCN and BiGRU network to leverage their complementary strengths, a parallel dual branch feature learning network was constructed to learn richer geological information from logging data. Meanwhile, MHA was introduced to fuse the output features of the dual network structure. Finally, the fused features were passed through the fully connected module to output the inversion results. TL was used to associate the correlation between reservoir properties and model inversion to improve the inversion performance. The application results with actual field data showed that the proposed method was accurate and robust in reservoir properties inversion. This study can provide a new way for reliable reservoir properties inversion and promote the application of artificial intelligence in data-driven energy science.

Stock market prediction employing ensemble methods: the Nifty50 index

<span lang="EN-US">Accurately forecasting stock fluctuations can yield high investment returns while minimizing risk. However, market volatility makes these projections unlikely. As a result, stock market data analysis is significant for research. Analysts and researchers have developed various stock price prediction systems to help investors make informed judgments. Extensive studies show that machine learning can anticipate markets by examining stock data. This article proposed and evaluated different ensemble learning techniques such as max voting, bagging, boosting, and stacking to forecast the Nifty50 index efficiently. In addition, an embedded feature selection is performed to choose an optimal set of fundamental indicators as input to the model, and extensive hyperparameter tuning is applied using grid search to each base regressor to enhance performance. Our findings suggest the bagging and stacking ensemble models with random forest (RF) feature selection offer lower error rates. The bagging and stacking regressor model 2 outperformed all other models with the lowest root mean square error (RMSE) of 0.0084 and 0.0085, respectively, showing a better fit of ensemble regressors. Finally, the findings show that machine learning algorithms can help fundamental analyses make stock investment decisions.</span>

Abstract PO2-27-05: Composite biomarkers for the prediction of progression-free survival with CDK4/6 inhibitors in metastatic HR+/HER2- breast cancer

Abstract CDK4/6 inhibitors are widely used in combination with endocrine therapy in the treatment of HR+/HER2- metastatic breast cancer. Generally the CDK4/6 inhibitors lead to a doubling of progression-free survival (PFS) over single agent endocrine therapy. In spite of a known mechanism of action, biomarkers to define the duration of response to CDK4/6 inhibitors with either aromatase inhibitors (AI) or fulvestrant (FUL) remain lacking. A cohort of over 300 patients receiving standard of care CDK4/6 inhibitor combination therapy for metastatic disease were consented for participation (NCT04526587). The duration of PFS in both AI and FUL treated sub-groups were comparable to that observed in randomized clinical trials. In this cohort progesterone receptor-status (HR 0.70, p=0.0123), visceral disease (HR 1.55, p=0.0013), prior endocrine therapy (HR 2.34, p< 0.001), and type of endocrine therapy (HR 2.16, p< 0.001) were associated with duration of PFS. To develop biomarkers, gene expression analyses was performed on 313 tumor samples representing over 200 individual patients. Focusing on the pre-treatment metastatic biopsy, established signatures associated with prognosis in HR+/HER2- breast cancer were employed to determine association with response to combination therapy in the metastatic setting. Proliferation-associated signatures were associated with shorter PFS. For example, an established RB loss signature was associated with PFS in both AI (HR 1.9, p=0.009) and FUL (HR 2.8, p=0.007) treated patients. Clinically employed signatures (e.g. OncotypeDx and Mammaprint) harbored variable associations with outcomes. Random Survival Forest feature selection was applied to individual and combined prognostic signatures. This approach resulted in a potent classifier for progression-free survival (HR 10, p< 1E-20). To enable applicability to different manifestations of disease, significant clinical pathological variables were incorporated into the model so that there was equivalent performance in AI and FUL treated patients for prediction of PFS. This predictive algorithm was also strongly associated with overall survival (HR 4.6, p=2.7E-8). Ongoing studies are evaluating the algorithms prospectively in NCT04526587. Citation Format: Agnieszka Witkiewicz, Jianxin Wang, Erik Knudsen, Ellis Levine, Tracey O'Connor. Composite biomarkers for the prediction of progression-free survival with CDK4/6 inhibitors in metastatic HR+/HER2- breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-27-05.

A maximal overlap discrete wavelet packet transform coupled with an LSTM deep learning model for improving multilevel groundwater level forecasts

Developing precise groundwater level (GWL) forecast models is essential for the optimal usage of limited groundwater resources and sustainable planning and management of water resources. In this study, an improved forecasting accuracy for up to 3 weeks ahead of GWLs in Bangladesh was achieved by employing a coupled Long Short Term Memory (LSTM) network-based deep learning algorithm and Maximal Overlap Discrete Wavelet Packet Transform (MODWPT) data preprocessing. The coupled LSTM-MODWPT model’s performance was compared with that of the LSTM model. For both standalone LSTM and LSTM-MODWPT models, the Random Forest feature selection approach was employed to select the ideal inputs from the candidate GWL lags. In the LSTM-MODWPT model, input GWL time series were decomposed using MODWPT. The ‘Fejér-Korovkin’ mother wavelet with a filter length of 18 was used to obtain a collection of scaling coefficients and wavelets for every single input time series. Model performance was assessed using five performance indices: Root Mean Squared Error; Scatter Index; Maximum Absolute Error; Median Absolute Deviation; and an a-20 index. The LSTM-MODWPT model outperformed standalone LSTM models for all time horizons in GWL forecasting. The percentage improvements in the forecasting accuracies were 36.28%, 32.97%, and 30.77%, respectively, for 1-, 2-, and 3-weeks ahead forecasts at the observation well GT3330001. Accordingly, the coupled LSTM-MODWPT model could potentially be used to enhance multiscale GWL forecasts. This research demonstrates that the coupled LSTM-MODWPT model could generate more precise GWL forecasts at the Bangladesh study site, with potential applications in other geographic locations globally.

Predictive models for flexible pavement fatigue cracking based on machine learning

Pavement performance prediction is crucial for ensuring the longevity and safety of road networks. In our extensive study, we employ a diverse array of techniques to enhance fatigue performance models in flexible pavements. The methodology begins with Random Forest feature selection, identifying the top 15 critical variables that significantly impact pavement performance. These variables form the basis for subsequent model development. Our investigation into model performance indicates the superiority of advanced machine learning methods such as Regression Trees (RT), Gaussian Process Regression (GPR), Support Vector Machines (SVM), Ensemble Trees (ET), and Artificial Neural Networks (ANN) over traditional linear regression methods. This consistent outperformance underscores their potential to reshape forecasting accuracy. Through extensive model optimization, we reveal robust performance across both complete and selected feature sets, emphasizing the importance of meticulous feature selection in enhancing forecast accuracy. The accuracy of our best optimized machine learning model is highlighted by its Performance Measurement metrics: RMSE of 22.416, MSE of 502.46, R-squared of 0.80848, and MAE of 8.9958. Additionally, comparative analysis with previous empirical models demonstrates that our best optimized machine learning model outperforms existing empirical models. This work underscores the significance of feature curation in pavement performance prediction, highlighting the potential of sophisticated modeling methodologies. Embracing cutting-edge technologies facilitates data-driven decisions, ultimately contributing to the development of more robust road networks, enhancing safety, and prolonging lifespan.

A Novel Framework for Forest Above-Ground Biomass Inversion Using Multi-Source Remote Sensing and Deep Learning

The estimation of forest above-ground biomass (AGB) can be significantly improved by leveraging remote sensing (RS) and deep learning (DL) techniques. In this process, it is crucial to obtain appropriate RS features and develop a suitable model. However, traditional methods such as random forest (RF) feature selection often fail to adequately consider the complex relationships within high-dimensional RS feature spaces. Moreover, challenges related to parameter selection and overfitting inherent in DL models may compromise the accuracy of AGB estimation. Therefore, this study proposes a novel framework based on freely available Sentinel-1 synthetic aperture radar (SAR) and Sentinel-2 optical data. Firstly, we designed new indices through the formula analogous with vegetation index calculation to integrate multidimensional spectral and structural information. Then, leveraging the simplicity of computational principles, a pigeon-inspired optimization algorithm (PIO) was introduced into a bi-directional long short-term memory neural network (PIO-BiLSTM), which achieved the set objective function through repeated iteration and validation to obtain the optimal model parameters. Finally, to verify the framework’s effect, we conducted experiments in two different tree species and compared another seven classical optimization algorithms and machine learning models. The results indicated that the new indices significantly improved the inversion accuracy of all models in both categories, and the PIO-BiLSTM model achieved the highest accuracy (Category-1: R2 = 0.8055, MAE = 8.8475 Mg·ha−1, RMSE = 12.2876 Mg·ha−1, relative RMSE = 18.1715%; Category-2: R2 = 0.7956, MAE = 1.7103 Mg·ha−1, RMSE = 2.2887 Mg·ha−1, relative RMSE = 9.3000%). Compared with existing methods, the proposed framework greatly reduced the labor costs in parameter selection, and its potential uncertainty also decreased by up to 9.0%. Furthermore, the proposed method has a strong generalization ability and is independent of tree species, indicating its great potential for future forest AGB inversion in wider regions with diverse forest types.

Value of multi-center 18F-FDG PET/CT radiomics in predicting EGFR mutation status in lung adenocarcinoma.

Accurate, noninvasive, and reliable assessment of epidermal growth factor receptor (EGFR) mutation status and EGFR molecular subtypes is essential for treatment plan selection and individualized therapy in lung adenocarcinoma (LUAD). Radiomics models based on 18F-FDG PET/CT have great potential in identifying EGFR mutation status and EGFR subtypes in patients with LUAD. The validation of multi-center data, model visualization, and interpretation are significantly important for the management, application and trust of machine learning predictive models. However, few EGFR-related research involved model visualization and interpretation, and multi-center trial. To develop explainable optimal predictive models based on handcrafted radiomics features (HRFs) extracted from multi-center 18F-FDG PET/CT to predict EGFR mutation status and molecular subtypes in LUAD. Baseline 18F-FDG PET/CT images of 383 LUAD patients from three hospitals and one public data set were collected. Further, 1808 HRFs were extracted from the primary tumor regions using Pyradiomics. Predictive models were built based on cross-combination of seven feature selection methods and seven machine learning algorithms. Yellowbrick and explainable artificial intelligence technology were used for model visualization and interpretation. Receiver operating characteristic curve, classification report and confusion matrix were used for model performance evaluation. Clinical applicability of the optimal models was assessed by decision curve analysis. STACK feature selection method combined with light gradient boosting machine (LGBM) reached optimal performance in identifying EGFR mutation status ([area under the curve] AUC=0.81 in the internal test cohort; AUC=0.62 in the external test cohort). Random forest feature selection method combined with LGBM reached optimal performance in predicting EGFR mutation molecular subtypes (AUC=0.89 in the internal test cohort; AUC=0.61 in the external test cohort). Explainable machine learning models combined with radiomics features extracted from multi-center/scanner 18F-FDG PET/CT have certain potential to identify EGFR mutation status and subtypes in LUAD, which might be helpful to the treatment of LUAD.