Forward Feature Selection Research Articles

A selective feature optimized multi-sensor based e-nose system detecting illegal drugs validated in diverse laboratory conditions

Detecting illegal drugs, such as cannabis and methamphetamine, with high accuracy and speed is a complex problem that requires an innovative solution. To address this challenge, we propose a new method that utilizes a newly developed electronic nose (e-nose) system with an unprecedented total of 56 sensors, including four different types: metal-oxide-semiconductor (MOS), electrochemical (EC), non-dispersive infrared (NDIR), and photoionization detector (PID). Previous studies on gas sensors have typically validated results in a single controlled laboratory condition. In contrast, our study evaluated the performance of our system in different environments from the original training setting. To evaluate the detection performance of our system in unfamiliar environments and its robustness, we diluted the drug gas with normal air from six different laboratory environments. In addition, we evaluated the detection accuracy of our method using forward-feature selection, which allowed us to evaluate the impact of different combinations of sensors. By selectively optimizing sensors based on their ability to capture unique features of different drugs, our proposed method reduced the number of optimal sensors to less than half of 56 (24 selected). The proposed method achieved a detection accuracy of 93.03% and reduced the error rate from 12.23% to 6.97% using 5166 datasets including cannabis, methamphetamine, and tobacco. Our research not only provides rapid and enhanced accuracy, but also has the potential to be an effective tool for detecting illegal drugs in various settings, which could greatly contribute to strengthening national and social security.

Predicting Football Team Performance with Explainable AI: Leveraging SHAP to Identify Key Team-Level Performance Metrics

Understanding the performance indicators that contribute to the final score of a football match is crucial for directing the training process towards specific goals. This paper presents a pipeline for identifying key team-level performance variables in football using explainable ML techniques. The input data includes various team-specific features such as ball possession and pass behaviors, with the target output being the average scoring performance of each team over a season. The pipeline includes data preprocessing, sequential forward feature selection, model training, prediction, and explainability using SHapley Additive exPlanations (SHAP). Results show that 14 variables have the greatest contribution to the outcome of a match, with 12 having a positive effect and 2 having a negative effect. The study also identified the importance of certain performance indicators, such as shots, chances, passing, and ball possession, to the final score. This pipeline provides valuable insights for coaches and sports analysts to understand which aspects of a team’s performance need improvement and enable targeted interventions to improve performance. The use of explainable ML techniques allows for a deeper understanding of the factors contributing to the predicted average team score performance.

PO-02-021 DIRECTION OF REPOLARIZATION IN BASELINE ECG IS PREDICTIVE OF RESPONSE TO CARDIAC RESYNCHRONIZATION THERAPY

Cardiac resynchronization therapy (CRT) is a cornerstone of treatment for advanced heart failure, but not all patients derive equal benefit from this intervention. Current approaches of using electrocardiographic and echocardiographic parameters to identify likely CRT responders demonstrate poor predictive ability. We seek to identify electrocardiographic parameters that can improve prediction of benefit with CRT This was a retrospective analysis of patients who underwent CRT-D placement at University of Pittsburgh Medical Center from 2014-2020. Patients were excluded if pre-implantation EKG or pre- and post-implantation echocardiograms were not available. Each patient’s immediate pre-implantation ECG was processed using the CineECG software (EKG Excellence) to obtain the mean temporospatial isochrone of ventricular activation. 192 features quantifying the activation and recovery pathways were computed for each patient. Response to CRT was defined as an improvement in ejection fraction (EF) of 5% on follow-up transthoracic echocardiogram 3-12 months after CRT implantation. Forward sequential feature selection was used to identify the features of highest importance based on negative logistic loss. 276 patients were included in this analysis (36% female, mean age 69 years, mean pre-procedural EF 27%). The 10 most important CineECG features were identified. When these features were used in a multivariate logistic regression, three features were independently predictive of CRT response: left bundle branch block (LBBB); leftward direction of late repolarization potentials (i.e., Tpeak - Tend rotated toward LV in the horizontal plane); and location of center-of-mass of terminal cardiac activation. Compared to CRT responders (n=169, 62%), non-responders were more likely to have abnormal direction of late repolarization potentials in both LBBB group (-0.34±0.30 vs. -0.45±0.30 vs, p=0.027) and RBBB group (0.25±0.51 vs. 0.61±0.27, p=0.031). In the latter group, direction of later repolarization potentials correlated with magnitude of change in post-procedural EF (r=0.23, p=0.016). Our findings confirm that the presence of LBBB is predictive of response to CRT. In addition, our data suggests that in patients with RBBB, a more leftward direction of late repolarization potentials is independently predictive of CRT response. Further studies into repolarization dynamics as a predictor of CRT response in RBBB are warranted.

Diagnosis of Alzheimer’s Disease using Wrapper Feature Selection Method

Alzheimer’s disease (AD) symptoms are being treated by early diagnosis, where we can only slow the symptoms and research is still undergoing. In consideration, using T1-weighted images several classification models are proposed in Machine learning to identify AD. In this paper, we consider the improvised feature selection, to reduce the complexity by using wrapping techniques and Restricted Boltzmann Machine (RBM). This present work used the subcortical and cortical features of 278 subjects from the ADNI dataset to identify AD and sMRI. Multi-class classification is used for the experiment i.e., AD, EMCI, LMCI, HC. The proposed feature selection consists of Forward feature selection, Backward feature selection, and Combined PCA & RBM. Forward and backward feature selection methods use an iterative method starting being no features in the forward feature selection and backward feature selection with all features included in the technique. PCA is used to reduce the dimensions and RBM is used to select the best feature without interpreting the features. We have compared the three models with PCA to analysis. The following experiment shows that combined PCA &RBM, and backward feature selection give the best accuracy with respective classification model RF i.e., 88.65, 88.56% respectively.

Innovative Forward Fusion Feature Selection Algorithm for Sentiment Analysis Using Supervised Classification

Sentiment analysis is considered one of the significant trends of the recent few years. Due to the high importance and increasing use of social media and electronic services, the need for reviewing and enhancing the provided services has become crucial. Revising the user services is based mainly on sentiment analysis methodologies for analyzing users’ polarities to different products and applications. Sentiment analysis for Arabic reviews is a major concern due to high morphological linguistics and complex polarity terms expressed in the reviews. In addition, the users can present their orientation towards a service or a product by using a hybrid or mix of polarity terms related to slang and standard terminologies. This paper provides a comprehensive review of recent sentiment analysis methods based on lexicon or machine learning (ML). The comparison provides a clear vision of the number of classes, the used dialect, the annotated algorithms, and their performance. The proposed methodology is based on cross-validation of Arabic data using a k-fold mechanism that splits the dataset into training and testing folds; subsequently, the data preprocessing is executed to clean sentiments from unwanted terms that can affect data analysis. A vectorization of the dataset is then applied using TF–IDF for counting word and polarity terms. Furthermore, a feature selection stage is processed using Pearson, Chi2, and Random Forest (RF) methods for mapping the compatibility between input and target features. This paper also proposed an algorithm called the forward fusion feature for sentiment analysis (FFF-SA) to provide a feature selection that applied different machine learning (ML) classification models for each chunk of k features and accumulative features on the Arabic dataset. The experimental results measured and scored all accuracies between the feature importance method and ML models. The best accuracy is recorded with the Naïve Bayes (NB) model with the RF method.

A tree growth based forward feature selection algorithm for intrusion detection system on convolutional neural network

With the rapid advancement of networking technologies, security system has become increasingly important to academics from several sectors. Intrusion detection (ID) provides a valuable protection by reducing the human resources required to keep an eye on intruders, improving the efficiency of detecting the various attacks in networks. Machine learning and deep learning are two key areas that have recently received a lot of attention, with a focus on improving the precision of detection classifiers. Using defense anvance research project agency (DARPA”98) datasets, a number of academics and research have developed intrusion detection systems. This paper discusses various approaches developed by different researchers, including scale-hybrid-IDS-AlertNet (SHIA), forward feature selection algorithm (FFSA), modified- mutual information feature selection (MMIFS), deep neural network (DNN), and the holes that remain to be filled, highlighting areas where these procedures can be improved, also are addressed and the proposed approach improved deep convolutional neural network (IDCNN) is compared with existing approach.

Feature Subset Selection With Multi-Scale Fuzzy Granulation

As a typical multigranularity data analysis model, multi-scale rough sets have attracted considerable attention in recent years. However, classical multi-scale rough sets and most of its extended models can only deal with discrete data, which limits its popularization and application. To overcome this problem, we investigate the fuzzy generalization of multi-scale rough sets as well as their application in feature selection for continuous data. To this end, a new type of decision systems, i.e., multi-scale fuzzy decision systems, is formalized to represent knowledge at different scales. Scaled fuzzy granules in terms of a family of scaled fuzzy relations are introduced, using which the granular structures of fuzzy lower and upper approximations are presented. A heuristic lattice-based optimal scale selection algorithm is then put forward from the viewpoint of maintaining the consistency of decision systems. Decision rules with strong generalization ability can be obtained by selecting appropriate scales. Finally, a forward feature selection algorithm was developed by means of the optimal scale to reduce redundant fuzzy relations. Extensive numerical experiments are further conducted to compare the proposed algorithm with some state-of-the-art algorithms. The experimental results show that our model can improve the generalization ability of fuzzy rough set, so as to be more feasible and effective.

Prediction of antioxidant peptides using a quantitative structure−activity relationship predictor (AnOxPP) based on bidirectional long short-term memory neural network and interpretable amino acid descriptors

Antioxidant peptides can protect against free radical-mediated diseases, especially food-derived antioxidant peptides are considered as potential competitors among synthetic antioxidants due to their safety, high activity and abundant sources. However, wet experimental methods can not meet the need for effectively screening and clearly elucidating the structure-activity relationship of antioxidant peptides. Therefore, it is particularly important to build a reliable prediction platform for antioxidant peptides. In this work, we developed a platform, AnOxPP, for prediction of antioxidant peptides using the bidirectional long short-term memory (BiLSTM) neural network. The sequence characteristics of peptides were converted into feature codes based on amino acid descriptors (AADs). Our results showed that the feature conversion ability of the combined-AADs optimized by the forward feature selection method was more accurate than that of the single-AADs. Especially, the model trained by the optimal descriptor SDPZ27 significantly outperformed the existing predictor on two independent test sets (Accuracy = 0.967 and 0.819, respectively). The SDPZ27-based AnOxPP learned four key structure-activity features of antioxidant peptides, with the following importance as steric properties > hydrophobic properties > electronic properties > hydrogen bond contributions. AnOxPP is a valuable tool for screening and design of peptide drugs, and the web-server is accessible at http://www.cqudfbp.net/AnOxPP/index.jsp.

Towards Optimal Variable Selection Methods for Soil Property Prediction Using a Regional Soil Vis-NIR Spectral Library

Soil visible and near-infrared (Vis-NIR, 350–2500 nm) spectroscopy has been proven as an alternative to conventional laboratory analysis due to its advantages being rapid, cost-effective, non-destructive and environmentally friendly. Different variable selection methods have been used to deal with the high redundancy, heavy computation, and model complexity of using full spectra in spectral modelling. However, most previous studies used a linear algorithm in the variable selection, and the application of a non-linear algorithm remains poorly explored. To address the current knowledge gap, based on a regional soil Vis-NIR spectral library (1430 soil samples), we evaluated seven variable selection algorithms together with three predictive algorithms in predicting seven soil properties. Our results showed that Cubist overperformed partial least squares regression (PLSR) and random forests (RF) in most soil properties (R2 > 0.75 for soil organic matter, total nitrogen and pH) when using the full spectra. Most of variable selection can greatly reduce the number of spectral bands and therefore simplified predictive models without losing accuracy. The results also showed that there was no silver bullet for the optimal variable selection algorithm among different predictive algorithms: (1) competitive adaptive reweighted sampling (CARS) always performed best for the PLSR algorithm, followed by forward recursive feature selection (FRFS); (2) recursive feature elimination (RFE) and genetic algorithm (GA) generally had better accuracy than others for the Cubist algorithm; and (3) FRFS had the best model performance for the RF algorithm. In addition, the performance was generally better when the algorithm used in the variable selection matched the predictive algorithm. The outcome of this study provides a valuable reference for predicting soil information using spectroscopic techniques together with variable selection algorithms.

Deep Learning Approach with LSTM for Daily Streamflow Prediction in a Semi-Arid Area: A Case Study of Oum Er-Rbia River Basin, Morocco

Daily hydrological modelling is among the most challenging tasks in water resource management, particularly in terms of streamflow prediction in semi-arid areas. Various methods were applied in order to deal with this complex phenomenon, but recently data-driven models have taken a better space, given their ability to solve prediction problems in time series. In this study, we have employed the Long Short-Term Memory (LSTM) network to simulate the daily streamflow over the Ait Ouchene watershed (AIO) in the Oum Er-Rbia river basin in Morocco, based on a temporal sequence of in situ and remotely sensed hydroclimatic data ranging from 2001 to 2010. The analysis adopted in this work is based on three-dimension input required by the LSTM model (1); the input samples used three splitting approaches: 70% of the dataset as training, splitting the data considering the hydrological year and the cross-validation method; (2) the sequence length; (3) and the input features using two different scenarios. The prediction results demonstrate that the LSTM performs poorly using the default data input scenario, whereas the best results during the testing were found in a sequence length of 30 days using approach 3 (R2 = 0.58). In addition, the LSTM fed with the lagged data input scenario using the Forward Feature Selection (FFS) method provides high performance accuracy using approach 2 (R2 = 0.84) in a sequence length of 20 days. Eventually, in applications related to water resources management where data are limited, the use of the deep learning technique is able to create high predictive accuracy, which can be enhanced with the right combination subset of features by using FFS.

Feature selection based on neighborhood rough sets and Gini index.

Neighborhood rough set is considered an essential approach for dealing with incomplete data and inexact knowledge representation, and it has been widely applied in feature selection. The Gini index is an indicator used to evaluate the impurity of a dataset and is also commonly employed to measure the importance of features in feature selection. This article proposes a novel feature selection methodology based on these two concepts. In this methodology, we present the neighborhood Gini index and the neighborhood class Gini index and then extensively discuss their properties and relationships with attributes. Subsequently, two forward greedy feature selection algorithms are developed using these two metrics as a foundation. Finally, to comprehensively evaluate the performance of the algorithm proposed in this article, comparative experiments were conducted on 16 UCI datasets from various domains, including industry, food, medicine, and pharmacology, against four classical neighborhood rough set-based feature selection algorithms. The experimental results indicate that the proposed algorithm improves the average classification accuracy on the 16 datasets by over 6%, with improvements exceeding 10% in five. Furthermore, statistical tests reveal no significant differences between the proposed algorithm and the four classical neighborhood rough set-based feature selection algorithms. However, the proposed algorithm demonstrates high stability, eliminating most redundant or irrelevant features effectively while enhancing classification accuracy. In summary, the algorithm proposed in this article outperforms classical neighborhood rough set-based feature selection algorithms.

Forecasting Coronary Heart Disease Risk With a 2-Step Hybrid Ensemble Learning Method and Forward Feature Selection Algorithm

Forecasting Coronary Heart Disease Risk With a 2-Step Hybrid Ensemble Learning Method and Forward Feature Selection Algorithm

A machine learning-based nano-photocatalyst module for accelerating the design of Bi2WO6/MIL-53(Al) nanocomposites with enhanced photocatalytic activity.

It is a great challenge to acquire novel Bi2WO6/MIL-53(Al) (BWO/MIL) nanocomposites with excellent catalytic activity by the trial-and-error method in the vast untapped synthetic space. The degradation rate of Rhodamine B dye (DRRhB) can be used as the main parameter to evaluate the catalytic activity of BWO/MIL nanocomposites. In this work, a machine learning-based nano-photocatalyst module was developed to speed up the design of BWO/MIL with desirable performance. Firstly, the DRRhB dataset was constructed, and four key features related to the synthetic conditions of BWO/MIL were filtered by the forward feature selection method based on support vector regression (SVR). Secondly, the SVR model with radical basis function for predicting the DRRhB of BWO/MIL was established with the key features and optimal hyperparameters. The correlation coefficients (R) between predicted and experimental DRRhB were 0.823 and 0.884 for leave-one-out cross-validation (LOOCV) and the external test, respectively. Thirdly, potential BWO/MIL nanocomposites with higher DRRhB were discovered by inverse projection, the prediction model, and virtual screening from the synthesis space. Meanwhile, an online web service (http://1.14.49.110/online_predict/BWO2) was built to share the model for predicting the DRRhB of BWO/MIL. Moreover, sensitivity analysis was brought into boosting the model's explainability and illustrating how the DRRhB of BWO/MIL changes over the four key features, respectively. The method mentioned here can provide valuable clues to develop new nanocomposites with the desired properties and accelerate the design of nano-photocatalysts.

Deteksi Kesehatan Janin Menggunakan Decision Tree dan Feature Forward Selection

Fetal health is very important, it is important for prospective mothers to know from an early age, until now a mother’s sense of ignorance abpout fetal health is very lacking and can result in fetal death. This is due to the lack of curiosity possessed by prospective moyhers and the lack of socialization and infrastructure from related parties about fetal health. Basically, the growth and development of a prospective fetus is very important so that it can be born healthy and without any obstacles at all.The pupose of this study was to detect fetal health using a decision tree classification algorithm with forward feature selection. This experiment was conducted using a public dataset of 2.126 patient data. The results showes that classification using a decision tree algorithm using a decision tree algorithm without feature selection resulted in an accuracy of 89.84%. While the use of the forward selection feature in this decision tree algorithm produces an accuracy of 91.06%.This shows that the use of the forward selection feature can increase accuracy by 1.22%.

Predicting perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods

BackgroundPerinatal mortality in Ethiopia is the highest in Africa, with 68 per 1000 pregnancies intrapartum deaths. It is mainly associated with home delivery, which contributes to more than 75% of perinatal deaths. Financial constraints significantly impact timely access to maternal health care. Financial incentives, such as health insurance, may address the demand- and supply-side factors. This study, hence, aims to predict perinatal mortality based on maternal health status and health insurance service using homogeneous ensemble machine learning methods.MethodsThe data was collected from the Ethiopian demographic health survey from 2011 to 2019 G.C. The data were pre-processed to get quality data that are suitable for the homogenous ensemble machine-learning algorithms to develop a model that predicts perinatal mortality. We have applied filter (chi-square and mutual information) and wrapper (sequential forward and sequential backward) feature selection methods. After selecting all the relevant features, we developed a predictive model using cat boost, random forest, and gradient boosting algorithms and evaluated the model using both objective (accuracy, precision, recall, F1_score, ROC) and subjective (domain expert) based evaluation techniques.ResultsPerinatal mortality prediction models were developed using random forest, gradient boosting, and cat boost algorithms with the overall accuracy of 89.95%, 90.24%, and 82%, respectively. Risk factors of perinatal mortality were identified using feature importance analysis and relevant rules were extracted using the best performing model.ConclusionsA prediction model that was developed using gradient boosting algorithms was selected for further use in the risk factor analysis, generating relevant rules, development of artifacts, and model deployment because it has registered better performance with 90.24% accuracy. The most determinant risk factors of perinatal mortality were identified using feature importance and some of them are community-based health insurance, mother's educational level, region and place of residence, age, wealth status, birth interval, preterm, smoking cigarette, anemia level, hemoglobin level, and marital status.

Correlation of image textures of a polarization feature parameter and the microstructures of liver fibrosis tissues

Mueller matrix imaging is emerging for the quantitative characterization of pathological microstructures and is especially sensitive to fibrous structures. Liver fibrosis is a characteristic of many types of chronic liver diseases. The clinical diagnosis of liver fibrosis requires time-consuming multiple staining processes that specifically target on fibrous structures. The staining proficiency of technicians and the subjective visualization of pathologists may bring inconsistency to clinical diagnosis. Mueller matrix imaging can reduce the multiple staining processes and provide quantitative diagnostic indicators to characterize liver fibrosis tissues. In this study, a fiber-sensitive polarization feature parameter (PFP) was derived through the forward sequential feature selection (SFS) and linear discriminant analysis (LDA) to target on the identification of fibrous structures. Then, the Pearson correlation coefficients and the statistical T-tests between the fiber-sensitive PFP image textures and the liver fibrosis tissues were calculated. The results show the gray level run length matrix (GLRLM)-based run entropy that measures the heterogeneity of the PFP image was most correlated to the changes of liver fibrosis tissues at four stages with a Pearson correlation of 0.6919. The results also indicate the highest Pearson correlation of 0.9996 was achieved through the linear regression predictions of the combination of the PFP image textures. This study demonstrates the potential of deriving a fiber-sensitive PFP to reduce the multiple staining process and provide textures-based quantitative diagnostic indicators for the staging of liver fibrosis.

A Proactive Attack Detection for Heating, Ventilation, and Air Conditioning (HVAC) System Using Explainable Extreme Gradient Boosting Model (XGBoost)

The advent of Industry 4.0 has revolutionized the life enormously. There is a growing trend towards the Internet of Things (IoT), which has made life easier on the one hand and improved services on the other. However, it also has vulnerabilities due to cyber security attacks. Therefore, there is a need for intelligent and reliable security systems that can proactively analyze the data generated by these devices and detect cybersecurity attacks. This study proposed a proactive interpretable prediction model using ML and explainable artificial intelligence (XAI) to detect different types of security attacks using the log data generated by heating, ventilation, and air conditioning (HVAC) attacks. Several ML algorithms were used, such as Decision Tree (DT), Random Forest (RF), Gradient Boosting (GB), Ada Boost (AB), Light Gradient Boosting (LGBM), Extreme Gradient Boosting (XGBoost), and CatBoost (CB). Furthermore, feature selection was performed using stepwise forward feature selection (FFS) technique. To alleviate the data imbalance, SMOTE and Tomeklink were used. In addition, SMOTE achieved the best results with selected features. Empirical experiments were conducted, and the results showed that the XGBoost classifier has produced the best result with 0.9999 Area Under the Curve (AUC), 0.9998, accuracy (ACC), 0.9996 Recall, 1.000 Precision and 0.9998 F1 Score got the best result. Additionally, XAI was applied to the best performing model to add the interpretability in the black-box model. Local and global explanations were generated using LIME and SHAP. The results of the proposed study have confirmed the effectiveness of ML for predicting the cyber security attacks on IoT devices and Industry 4.0.

Forest Damage by Extra-Tropical Cyclone Klaus-Modeling and Prediction

Windstorms may have negative consequences on forest ecosystems, industries, and societies. Extreme events related to extra-tropical cyclonic systems remind us that better recognition and understanding of the factors driving forest damage are needed for more efficient risk management and planning. In the present study, we statistically modelled forest damage caused by the windstorm Klaus in south-west France. This event occurred on 24 January 2009 and caused severe damage to maritime pine (Pinus pinaster) forest stands. We aimed at isolating the best potential predictors that can help to build better predictive models of forest damage. We applied the random forest (RF) technique to find the best classifiers of the forest damage binary response variable. Five-fold spatial block cross-validation, repeated five times, and forward feature selection (FFS) were applied to the control for model over-fitting. In addition, variable importance (VI) and accumulated local effect (ALE) plots were used as model performance metrics. The best RF model was used for spatial prediction and forest damage probability mapping. The ROC AUC of the best RF model was 0.895 and 0.899 for the training and test set, respectively, while the accuracy of the RF model was 0.820 for the training and 0.837 for the test set. The FFS allowed us to isolate the most important predictors, which were the distance from the windstorm trajectory, soil sand fraction content, the MODIS normalized difference vegetation index (NDVI), and the wind exposure index (WEI). In general, their influence on the forest damage probability was positive for a wide range of the observed values. The area of applicability (AOA) confirmed that the RF model can be used to construct a probability map for almost the entire study area.

Identification of risk variants related to malignant tumors in children with birth defects by whole genome sequencing

BackgroundChildren with birth defects (BD) are more likely to develop cancer and the increased risk of cancer persists into adulthood. Prior population-based assessments have demonstrated that even non-chromosomal BDs are associated with at least two-fold increase of cancer risk. Identification of variants that are associated with malignant tumor in BD patients without chromosomal anomalies may improve our understanding of the underlying molecular mechanisms and provide clues for early cancer detection in children with BD.MethodsIn this study, whole genome sequencing (WGS) data of blood-derived DNA for 1653 individuals without chromosomal anomalies were acquired from the Kids First Data Resource Center (DRC), including 541 BD probands with at least one type of malignant tumors, 767 BD probands without malignant tumor, and 345 healthy family members who are the parents or siblings of the probands. Recurrent variants exclusively seen in cancer patients were selected and mapped to their corresponding genomic regions. The targeted genes/non-coding RNAs were further reduced using random forest and forward feature selection (ffs) models.ResultsThe filtered genes/non-coding RNAs, including variants in non-coding areas, showed enrichment in cancer-related pathways. To further support the validity of these variants, blood WGS data of additional 40 independent BD probands, including 25 patients with at least one type of cancers from unrelated projects, were acquired. The counts of variants of interest identified in the Kid First data showed clear deviation in the validation dataset between BD patients with cancer and without cancer. Furthermore, a deep learning model was built to assess the predictive abilities in the 40 patients using variants of interest identified in the Kids First cohort as feature vectors. The accuracies are ~ 75%, with the noteworthy observation that variants mapped to non-coding regions provided the highest accuracy (31 out of 40 patients were labeled correctly).ConclusionWe present for the first time a panorama of genetic variants that are associated with cancers in non-chromosomal BD patients, implying that our approach may potentially serve for the early detection of malignant tumors in patients with BD.

Enhancing wind power forecast accuracy using the weather research and forecasting numerical model-based features and artificial neuronal networks

Forecasting with accuracy the quantity of energy produced by wind power plants is crucial to enabling its optimal integration into power systems and electricity markets. Despite the remarkable improvements in the wind forecasting systems in recent years, large errors can still be observed, especially for longer time horizons. This work focuses on identifying new numerical weather prediction (NWP)-based features aiming to improve the overall quality of wind power forecasts.The methodology also incorporates a sequential forward feature selection algorithm. This algorithm was designed to select iteratively the meteorological features which minimize the wind forecast errors.The methodology was applied separately to seven wind parks in Portugal with different climate characteristics. The proposed approach allowed a reduction between 13% and 37% in the root mean square errors of wind power forecasts, compared with a baseline scenario. While the meteorological features identified for each wind park showed similarities within regions with analogous wind power generation profiles, each wind park required specific meteorological parameters as input data to obtain the best performance. Thus, the results show to be crucial to select the most relevant features of a specific site to maximize the accuracy of a wind power forecast.