Relative Importance Of Features Research Articles

The Relative Importance of Psychopathy Features as Predictors of Externalizing Behaviors in Youth: A Multimethod Examination.

Research in youth psychopathy has focused heavily on the affective features (i.e., callous-unemotional [CU] traits) given robust links to severe and chronic forms of externalizing behaviors. Recently, there have been calls to expand the scope of work in this area to examine the importance of other interpersonal (i.e., antagonism) and behavioral (i.e., disinhibition) features of psychopathy. In the present study, we apply an under-utilized statistical approach (i.e., dominance analysis) to assess the relative importance of CU traits, antagonism, and disinhibition in the prediction of externalizing behaviors in youth, cross-sectionally and at 9-month follow-up. Using a multi-informant (youth- and parent-report), multi-method (questionnaire, ecological momentary assessment [EMA]) preregistered approach in a diverse sample of clinically referred youth (Mage = 12.60 years, SD = .95 years, 47% female; 61% racial/ethnic minority), we found youth- and parent-reported psychopathy features accounted for a significant proportion of variance in externalizing behavior cross-sectionally and longitudinally. However, results provided limited support for our preregistered hypotheses. While antagonism and disinhibition had larger general dominance weights relative to CU traits for both youth- and parent-report, most differences were non-significant. Thus, the interpersonal, affective, and behavioral psychopathy features could not be distinguished from one another in terms of their importance in the prediction of externalizing behavior, assessed cross-sectionally or longitudinally. Taken together, the results highlight promising avenues for future research on the relative importance of youth psychopathy features.

Optimal quantification of residence time distribution profiles from a quality assurance perspective

Residence time distribution (RTD) has been widely applied across various fields of chemical engineering, including pharmaceutical manufacturing, for applications such as material traceability, quality assurance, system health monitoring, and fault detection. Determination of a representative RTD, in principle, requires an accurate process analytical technology (PAT) procedure capturing the entire range of tracer concentrations from zero to maximum. Such a wide concentration range creates at least two problems: i) decreased accuracy of the model across the entire range of concentrations, relating to limit of quantification, and ii) ambiguity associated with the detection of the tracer for low concentration levels, relating to limit of detection (LOD). These problems affect not only the RTD profile itself, but also RTD-based applications, which can potentially lead to erroneous conclusions. This article seeks to minimize the impact of these problems by understanding the relative importance of different features of RTD on the detection of out-of-specification (OOS) products. In this work, the RTD obtained experimentally was truncated at different levels, to investigate the impact of the truncation of RTD on funnel plots for OOS detection. The main finding is that the tail of the RTD can be truncated with no loss of accuracy in the determination of exclusion intervals. This enables the manufacturing scientist to focus entirely on the peak region, maximizing the accuracy of chemometric models.

What are the Priorities of Bureaucrats? Evidence from Conjoint Experiments with Procurement Officials

A well-functioning bureaucracy is a precondition for efficient public goods provision. However, bureaucratic decision-making is still largely seen as a black box. We provide novel insights into the preferences of bureaucrats regarding their work outcomes. We focus on a major public sector activity and survey more than 900 real-life procurement officials in Finland and Germany. The questionnaire includes hypothetical choice experiments to study the relative importance of multiple features in tender outcomes. First, bureaucrats state to have substantial discretion at work but no important incentives. Second, our experimental results show that procurers are particularly worried about avoiding negative risks concerning prices and supplier reputation. Third, an avoidance of bidders with prior bad performance appears to be an extremely important factor. Fourth, procurers value a certain degree of competition, while litigation concerns and regional favoritism play only a small role. The striking lack of heterogeneous effects points towards the role of intrinsic motivation among public buyers in countries with high public sector capacity.

Random forest models for motorcycle accident prediction using naturalistic driving based big data

Motorcycle accident studies usually rely upon data collected from road accidents collected through questionnaire surveys/police reports including characteristics of motorcycle riders and contextual data such as road environment. The present study utilizes big data, in the form of vehicle trajectory patterns collected through GPS, coupled with self-reported road accident information along with motorcycle rider characteristics to predict the likelihood of involvement of a motorcyclist in an accident. Random Forest-based machine learning algorithm is employed by taking inputs based on a variety of features derived from trajectory data. These features are mobility-based features, acceleration event-based features, aggressive overtaking event-based features and motorcyclists socio-economic features. Additionally, the relative importance of features is also determined which shows that aggressive overtaking event-based features have more impact on motorcycle accidents as compared to other categories of features. The developed model is useful in identifying risky motorcyclists and implementing safety measures focused towards them.

Patient Preferences for Pharmaceutical and Device-Based Treatments for Uncontrolled Hypertension: Discrete Choice Experiment.

Discrete choice experiment is a survey method used to understand how individuals make decisions and to quantify the relative importance of features. Using discrete choice experiment methods, we quantified patient benefit-risk preferences for hypertension treatments, including pharmaceutical and interventional treatments, like renal denervation. Respondents from the United States with physician-confirmed uncontrolled hypertension selected between treatments involving a procedure or pills, using a structured survey. Treatment features included interventional, noninterventional, or no hypertension treatment; number of daily blood pressure (BP) pills; expected reduction in office systolic BP; duration of effect; and risks of drug side effects, access site pain, or vascular injury. The results of a random-parameters logit model were used to estimate the importance of each treatment attribute. Among 400 patients completing the survey between 2020 and 2021, demographics included: 52% women, mean age 59.2±13.0 years, systolic BP 155.1±12.3 mm Hg, and 1.8±0.9 prescribed antihypertensive medications. Reduction in office systolic BP was the most important treatment attribute. The remaining attributes, in decreasing order, were duration of effect, whether treatment was interventional, number of daily pills, risk of vascular injury, and risk of drug side effects. Risk of access site pain did not influence choice. In general, respondents preferred noninterventional over interventional treatments, yet only a 2.3 mm Hg reduction in office systolic BP was required to offset this preference. Small reductions in office systolic BP would offset risks of vascular injury or drug side effects. At least a 20% risk of vascular injury or drug side effects would be tolerated in exchange for improved BP. Reduction in systolic BP was identified as the most important driver of patient treatment preference, while treatment-related risks had less influence. The results indicate that respondents would accept interventional treatments in exchange for modest reductions in systolic BP compared with those observed in renal denervation trials.

Integrated metabolomics and machine learning approach to predict hypertensive disorders of pregnancy

Hypertensive disorders of pregnancy account for 3% to 10% of maternal-fetal morbidity and mortality worldwide. This condition has been considered one of the leading causes of maternal deaths in developing countries, such as India. This study aimed to discover hypertensive disorders of pregnancy-specific candidate urine metabolites as markers for hypertensive disorders of pregnancy by applying integrated metabolomics and machine learning approaches. The targeted urinary metabolomics study was conducted in 70 healthy pregnant controls and 133 pregnant patients having hypertension as cases. Hypertensive disorders of pregnancy-specific metabolites for disease prediction were further extracted using univariate and multivariate statistical analyses. For machine learning analysis, 80% of the data were used for training (79 for hypertensive disorders of pregnancy and 42 for healthy pregnancy) and validation (27 for hypertensive disorders of pregnancy and 14 for healthy pregnancy), and 20% of the data were used for test sets (27 for hypertensive disorders of pregnancy and 14 for healthy pregnancy). The statistical analysis using an unpaired t test revealed 44 differential metabolites. Pathway analysis showed mainly that purine and thiamine metabolism were altered in the group with hypertensive disorders of pregnancy compared with the healthy pregnancy group. The area under the receiver operating characteristic curves of the 5 most predominant metabolites were 0.98 (adenosine), 0.92 (adenosine monophosphate), 0.89 (deoxyadenosine), 0.81 (thiamine), and 0.81 (thiamine monophosphate). The best prediction accuracies were obtained using 2 machine learning models (95% for the gradient boost model and 98% for the decision tree) among the 5 used models. The machine learning models showed higher predictive performance for 3 metabolites (ie, thiamine monophosphate, adenosine monophosphate, and thiamine) among 5 metabolites. The combined accuracies of adenosine from all models were 98.6 in the training set and 95.6 in the test set. Moreover, the predictive performance of adenosine was higher than other metabolites. The relative feature importance of adenosine was also observed in the decision tree and the gradient boost model. Among other metabolites, adenosine and thiamine metabolites were found to differentiate participants with hypertensive disorders of pregnancy from participants with healthy pregnancies; hence, these metabolites can serve as a promising noninvasive marker for the detection of hypertensive disorders of pregnancy.

Let it go! Point of release prediction for virtual throwing

In this paper, we present a novel real-time physical interaction system that allows users to throw a virtual ball without using an intermediary device such as a controller. A dataset of throwing motions was captured from six actors, from which ground truth Point of Release (PoR) frames were calculated. We trained a PoR prediction model using motion features extracted from arm joints, and developed a detection algorithm to predict the PoR of a throwing motion in real-time. Evaluation of the system with pre-recorded throwing motion data results in detection errors of less than 50 ms. Qualitative results from six users of the real-time system in VR indicate that the task of throwing without a controller was very natural, although the system performed better for underarm than overarm throws. Finally, we report the relative importance of different joints and motion features for the PoR prediction task.

Evaluation and Prediction of Pavement Deflection Parameters Based on Machine Learning Methods

The deflection measurements made using Falling Weight Deflectometers (FWDs) are widely used in the back-calculation of pavement layer moduli. Pavement structural characteristics, changes in temperature, and other related factors exert a significant effect on the deflection measurements. Therefore, three machine learning methods—Classification and Regression Tree (CART), Random Forest (RF), and Gradient Boosting Decision Tree (GBDT)—were used to evaluate the importance of influencing factors including FWD test conditions, pavement structural parameters, climatic factors, traffic level, rehabilitation level, and service age, on the FWD measurements of deflection basin in this study. The results indicated that structural number was an important feature for all FWD measurements but its importance on lg(D0–D20) and lg(D0–D30) was smaller than other FWD measurements. The relative feature importance of the asphalt layer, base, and subbase on lg(D0–D20) and lg(D0–D30) was asphalt layer > subbase > base; their relative importance on lg(D20–D60), lg(D30–D60), and lg(D30–D90) was asphalt layer > base > subbase; and their relative importance on lg(D90−D120) and lg(D60–D120) was base > subbase > asphalt layer. Among the FWD test condition variables, drop load was the most significant factor influencing deflection measurements. The second-layer temperature was also important for lg(D0–D20), lg(D0–D30), and lg(D0–D45). The importance of precipitation was greater than the freeze index. The prediction results shown that the accuracy of GBDT was as high as 99%. Besides, GBDT outperformed RF, and RF outperformed CART. The analyses between FWD deflection parameters and influencing factors, especially the structural characteristics of the pavement, provide theoretical evidence for the evaluation of pavement layer strength on the basis of FWD data.

Implementation of boosting algorithms for prediction of punching shear strength of RC column footings

For more accurate and reliable prediction of punching shear strength of reinforced concrete (RC) column footings, machine learning (ML) was utilized. To develop ML-based strength prediction models, an experimental database was built by collecting test results (total 218 specimens without shear reinforcement and subjected to concentric loads) from the literature, and three boosting algorithms including the adaptive boosting (AdaBoost), gradient boosting (GBR), and extreme gradient boosting (XGBoost) were implemented. Input variables were carefully selected based on existing study results, and optimal hyperparameters of the ML-based models were determined based on the grid search with k-fold cross-validation. The ML-based models with the optimal hyperparameters showed highly accurate predictions without any significant and evident bias for all input variables, and outperformed current design codes and existing models. The relative feature importance of input variables showed that the effective footing depth is the most important variable for the prediction performance, and the boxplot analysis, ANOVA test, and t-test results revealed that spring or uniform load supports can be used instead of sand supports to avoid the complexity of the test setup. Finally, based on the study results, a simple design equation was proposed to widen the scope of application and to consider the effects of the shear span-to-depth ratio, steel yield strength, and test setup.

Machine learning-based prediction of harmful algal blooms in water supply reservoirs

Abstract Harmful algal blooms (HABs) pose a potential risk to human and ecosystem health. HAB occurrences are influenced by numerous environmental factors; thus, accurate predictions of HABs and explanations about the predictions are required to implement preventive water quality management. In this study, machine learning (ML) algorithms, i.e., random forest (RF) and extreme gradient boosting (XGB), were employed to predict HABs in eight water supply reservoirs in South Korea. The use of synthetic minority oversampling technique for addressing imbalanced HAB occurrences improved classification performance of the ML algorithms. Although RF and XGB resulted in marginal performance differences, XGB exhibited more stable performance in the presence of data imbalance. Furthermore, a post hoc explanation technique, Shapley additive explanation was employed to estimate relative feature importance. Among the input features, water temperature and concentrations of total nitrogen and total phosphorus appeared important in predicting HAB occurrences. The results suggest that the use of ML algorithms along with explanation methods increase the usefulness of predictive models as a decision-making tool for water quality management.

Investigation and machine learning-based prediction of parametric effects of single point incremental forming on pillow effect and wall profile of AlMn1Mg1 aluminum alloy sheets

Today the topic of incremental sheet forming (ISF) is one of the most active areas of sheet metal forming research. ISF can be an essential alternative to conventional sheet forming for prototypes or non-mass products. Single point incremental forming (SPIF) is one of the most innovative and widely used fields in ISF with the potential to form sheet products. The formed components by SPIF lack geometric accuracy, which is one of the obstacles that prevents SPIF from being adopted as a sheet forming process in the industry. Pillow effect and wall displacement are influential contributors to manufacturing defects. Thus, optimal process parameters should be selected to produce a SPIF component with sufficient quality and without defects. In this context, this study presents an insight into the effects of the different materials and shapes of forming tools, tool head diameters, tool corner radiuses, and tool surface roughness (Ra and Rz). The studied factors include the pillow effect and wall diameter of SPIF components of AlMn1Mg1 aluminum alloy blank sheets. In order to produce a well-established study of process parameters, in the scope of this paper different modeling tools were used to predict the outcomes of the process. For that purpose, actual data collected from 108 experimentally formed parts under different process conditions of SPIF were used. Neuron by Neuron (NBN), Gradient Boosting Regression (GBR), CatBoost, and two different structures of Multilayer Perceptron were used and analyzed for studying the effect of parameters on the factors under scrutiny. Different validation metrics were adopted to determine the quality of each model and to predict the impact of the pillow effect and wall diameter. For the calculation of the pillow effect and wall diameter, two equations were developed based on the research parameters. As opposed to the experimental approach, analytical equations help researchers to estimate results values relatively speedily and in a feasible way. Different partitioning weight methods have been used to determine the relative importance (RI) and individual feature importance of SPIF parameters for the expected pillow effect and wall diameter. A close relationship has been identified to exist between the actual and predicted results. For the first time in the field of incremental forming study, through the construction of Catboost models, SHapley Additive exPlanations (SHAP) was used to ascertain the impact of individual parameters on pillow effect and wall diameter predictions. CatBoost was able to predict the wall diameter with R2 values between the range of 0.9714 and 0.8947 in the case of the training and testing dataset, and between the range of 0.6062 and 0.6406 when predicting pillow effect. It was discovered that, depending on different validation metrics, the Levenberg–Marquardt training algorithm performed the most effectively in predicting the wall diameter and pillow effect with R2 values in the range of 0.9645 and 0.9082 for wall diameter and in the range of 0.7506 and 0.7129 in the case of the pillow effect. NBN has no results worthy of mentioning, and GBR yields good prediction only of the wall diameter.

Barrier islands influence the assimilation of terrestrial energy in nearshore fishes

We examined the relative importance of landscape features on estuarine fish trophic structure and dependence on terrestrial organic matter (OMterr) in four barrier island lagoon systems along the Alaskan Beaufort Sea coast. Our study compared two relatively large lagoon systems characterized by high river discharge and relatively free ocean water exchanges (central region near Prudhoe Bay, Alaska) with two highly protected lagoons characterized by low river discharge and limited exchange with ocean waters (eastern region near Kaktovik, Alaska). We hypothesized that freshwater discharge would be a strong determinant of food web structure for both resident marine and diadromous fishes if more discharge increases availability of OMterr relative to lagoons with limited or no river inputs. To consider differences in trophic characteristics in fishes between study regions, we estimated community-wide measures of trophic structure (hereafter, community metrics) and the relative use of OMterr from mixing models using stable isotope composition (δ13C and δ15N; muscle tissue) among 12 species and identified the influences of region and body size. Fish captured in lagoons well protected by barrier islands had more distinct and diverse isotopic niches relative to those in more exposed lagoons based on community metrics. The use of OMterr by nearshore fishes in both regions was substantial and was >50% for diadromous species. Between regions, OMterr use differed in 6 of the 8 species considered but was not consistently higher in one region. The relative importance of OMterr varied with fish size in 7 of 10 species considered, with more OMterr used by smaller individuals. This work highlights the importance of OMterr to Arctic fishes and fisheries, some of which are of subsistence importance, even when feeding grounds are primarily marine. We propose that landscape features, particularly barrier islands, play an important role in structuring nearshore food webs. Barrier islands may provide a previously undocumented ecosystem service of increasing food web complexity, which may promote system resilience.

Preferred green home features by potential buyers in Sibu Sarawak

Green concept was gaining attention since 1960s, with research and development initiatives been allocated to ensure its evolvement, for enhancing the minimization on the impact of global warming and greenhouse gases emissions. Enormous green home features be introduced worldwide, but such features were not collated into a single piece of information, and the preference of home buyers varies according to the culture, living environment, and personal preferences. This research aims to examine the relative importance of green home features to potential home buyers in Sibu, Sarawak. Purposive and random sampling methods were adopted as the participants need to meet the requirement of having intention to purchase a house in Sibu in the next five years. Quantitative online questionnaire survey was distributed to the potential home buyers with 118 valid responses returned (response rate 29.57%). The results indicated that air quality, energy saving lighting, and energy efficient air conditioner were the three most preferred features by the respondents. This research could contribute to the knowledge gap of the preferred green home features by the potential home buyers in Sibu, and hence, serve as a piece of supporting document to the developers, to meet buyers’ demand and further increase the supply of green home in Sibu.

Probing the Explainability of Neural Network Cloud-Top Pressure Models for LEO and GEO Imagers

Abstract Satellite low-Earth-orbiting (LEO) and geostationary (GEO) imager estimates of cloud-top pressure (CTP) have many applications in both operations and in studying long-term variations in cloud properties. Recently, machine learning (ML) approaches have shown improvement upon physically based algorithms. However, ML approaches, and especially neural networks, can suffer from a lack of interpretability, making it difficult to understand what information is most useful for accurate predictions of cloud properties. We trained several neural networks to estimate CTP from the infrared channels of the Visible Infrared Imaging Radiometer Suite (VIIRS) and the Advanced Baseline Imager (ABI). The main focus of this work is assessing the relative importance of each instrument’s infrared channels in neural networks trained to estimate CTP. We use several ML explainability methods to offer different perspectives on feature importance. These methods show many differences in the relative feature importance depending on the exact method used, but most agree on a few points. Overall, the 8.4- and 8.6-μm channels appear to be the most useful for CTP estimation on ABI and VIIRS, respectively, with other native infrared window channels and the 13.3-μm channel playing a moderate role. Furthermore, we find that the neural networks learn relationships that may account for properties of clouds such as opacity and cloud-top phase that otherwise complicate the estimation of CTP. Significance Statement Model interpretability is an important consideration for transitioning machine learning models to operations. This work applies several explainability methods in an attempt to understand what information is most important for estimating the pressure level at the top of a cloud from satellite imagers in a neural network model. We observe much disagreement between approaches, which motivates further work in this area but find agreement on the importance of channels in the infrared window region around 8.6 and 10–12 μm, informing future cloud property algorithm development. We also find some evidence suggesting that these neural networks are able to learn physically relevant variability in radiation measurements related to key cloud properties.

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Global navigation satellite system (GNSS) site coordinate time series provides essential data for geodynamic and geophysical studies, realisation of a regional or global geodetic reference frames, and crustal deformation research. The coordinate time series has been conventionally modelled by least squares (LS) fitting with harmonic functions, alongside many other analysis methods. As a key limitation, the traditional modelling approaches simply use the functions of time variable, despite good knowledge of various underlying physical mechanisms responsible for the site displacements. This paper examines the use of machine learning (ML) models to reflect the effects or residential effects of physical variables related to Sun and the Moon ephemerides, polar motion, temperature, atmospheric pressure, and hydrology on the site displacements. To form the ML problem, these variables are constructed as the input vector of each ML training sample, while the vertical displacement of a GNSS site is regarded as the output value. In the evaluation experiments, three ML approaches, namely the gradient boosting decision tree (GBDT) approach, long short-term memory (LSTM) approach, and support vector machine (SVM) approach, are introduced and evaluated with the time series datasets collected from 9 GNSS sites over the period of 13 years. The results indicate that all three approaches achieve similar fitting precision in the range of 3–5 mm in the vertical displacement component, which is an improvement in over 30% with respect to the traditional LS fitting precision in the range of 4–7 mm. The prediction of the vertical time series with the three ML approaches shows the precision in the range of 4–7 mm over the future 24- month period. The results also indicate the relative importance of different physical features causing the displacements of each site. Overall, ML approaches demonstrate better performance and effectiveness in modelling and prediction of GNSS time series, thus impacting maintenance of geodetic reference frames, geodynamics, geophysics, and crustal deformation analysis.

Interpretable stock price forecasting model using genetic algorithm-machine learning regressions and best feature subset selection

Recent stock market studies adopting machine learning and deep learning techniques have achieved remarkable performances with convenient accessibility. However, machine learning and deep learning models are notorious for their black-box structure. To build human-friendly interpretability in stock price prediction, many studies focus on the relationship between input features and the outcome by measuring the feature importance. However, the feature-importance-based interpretability methods have such drawbacks as relative feature importance, vague importance of correlated features, and impractical interpretability. Furthermore, they overlook two principal characteristics of time series stock price data: time-dependency and collective behavior of features. As a solution to catch the collective behavior of features over a whole data period, we propose the best feature subset selection. Additionally, for the solution to reflect the time-dependent characteristic of stock price data over a short data period, we propose piecewise best feature subset selection. The proposed algorithm uses two separate input feature sets: internal technical indicators and external market prices. This bilateral forecasting scheme goes through a two-stage feature selection process composed of feature set expansion, hybridized genetic algorithm-machine learning regressions to select important features, and importance score filtering to select optimal features. Finally, the best feature subset is selected for forecasting and interpretation. The proposed method achieves the best feature subset of parsimoniously fewer features for interpretability and improves average forecasting Root Mean Squared Error by 10.42% for the optimal feature set and 13.47% for the best feature subset of the internal technical indicators. For enhanced local interpretability in this study, we use Savitzky–Golay smoothing as part of piecewise optimal curve fitting to examine each potential grouping of external features. The proposed local interpretability technique using piecewise optimal curve fitting and piecewise best feature subset provides a more timely-flexible interpretation of stock price behavior using a few best features for piecewise data segments. Compared with other feature-importance interpretability techniques that only rely on either a single data point or a whole data period, the proposed interpretability technique overcomes their limitations, and reflects the main characteristics of time series data.

Fake news detection on Twitter

Purpose Owing to the increased accessibility of internet and related technologies, more and more individuals across the globe now turn to social media for their daily dose of news rather than traditional news outlets. With the global nature of social media and hardly any checks in place on posting of content, exponential increase in spread of fake news is easy. Businesses propagate fake news to improve their economic standing and influencing consumers and demand, and individuals spread fake news for personal gains like popularity and life goals. The content of fake news is diverse in terms of topics, styles and media platforms, and fake news attempts to distort truth with diverse linguistic styles while simultaneously mocking true news. All these factors together make fake news detection an arduous task. This work tried to check the spread of disinformation on Twitter. Design/methodology/approach This study carries out fake news detection using user characteristics and tweet textual content as features. For categorizing user characteristics, this study uses the XGBoost algorithm. To classify the tweet text, this study uses various natural language processing techniques to pre-process the tweets and then apply a hybrid convolutional neural network–recurrent neural network (CNN-RNN) and state-of-the-art Bidirectional Encoder Representations from Transformers (BERT) transformer. Findings This study uses a combination of machine learning and deep learning approaches for fake news detection, namely, XGBoost, hybrid CNN-RNN and BERT. The models have also been evaluated and compared with various baseline models to show that this approach effectively tackles this problem. Originality/value This study proposes a novel framework that exploits news content and social contexts to learn useful representations for predicting fake news. This model is based on a transformer architecture, which facilitates representation learning from fake news data and helps detect fake news easily. This study also carries out an investigative study on the relative importance of content and social context features for the task of detecting false news and whether absence of one of these categories of features hampers the effectiveness of the resultant system. This investigation can go a long way in aiding further research on the subject and for fake news detection in the presence of extremely noisy or unusable data.

A Novel Multidose Dry Powder Inhaler and Its Application in Patients with Asthma and COPD

Patient and physician satisfaction with the maintenance inhaler device is an important factor in medication adherence and effectiveness in respiratory disorders. We look at inhaler preferences in asthma and chronic obstructive pulmonary disease (COPD) from both the patient's and physician’s perspectives, emphasizing the relative importance of device features and patient considerations in inhaler selection. Teva Pharmaceuticals developed the multidose dry powder inhaler (M-DDPI) to treat chronic respiratory diseases. It is a metered-dose dry powder inhaler device intended to seem like a standard pressured metered-dose inhaler, but its internal geometry is somewhat distinctive. Inhalation treatment permits medication delivered directly to the airways. Inhalation devices serve a vital role in the therapy of obstructive lung illnesses such as asthma and COPD. The purpose of this research is to see how M-DDPI achieves maximum bronchial deposition of the medicine; the device must provide a high percentage of small particles, be simple to operate, and supply continuous and exact dosages of the active ingredient in COPD and asthma patients. Patients with COPD and asthma who had been diagnosed and treated at the hospital were first recruited as research participants and split into two groups. The control group got oxygen therapy as part of their standard treatment, whereas the experimental group received M-DDPI therapy. Statistical analysis techniques such as the Mann-Whitney U test and chi-squared test were used to examine the effects. The data show that the efficacies of M-DDPI therapy for COPD and asthma for obstructive lung illnesses seem to be promising right now.

Comparing lesion and feature selections to predict progression in newly diagnosed DLBCL patients with FDG PET/CT radiomics features

PurposeBiomarkers that can accurately predict outcome in DLBCL patients are urgently needed. Radiomics features extracted from baseline [18F]-FDG PET/CT scans have shown promising results. This study aims to investigate which lesion- and feature-selection approaches/methods resulted in the best prediction of progression after 2 years.MethodsA total of 296 patients were included. 485 radiomics features (n = 5 conventional PET, n = 22 morphology, n = 50 intensity, n = 408 texture) were extracted for all individual lesions and at patient level, where all lesions were aggregated into one VOI. 18 features quantifying dissemination were extracted at patient level. Several lesion selection approaches were tested (largest or hottest lesion, patient level [all with/without dissemination], maximum or median of all lesions) and compared to the predictive value of our previously published model. Several data reduction methods were applied (principal component analysis, recursive feature elimination (RFE), factor analysis, and univariate selection). The predictive value of all models was tested using a fivefold cross-validation approach with 50 repeats with and without oversampling, yielding the mean cross-validated AUC (CV-AUC). Additionally, the relative importance of individual radiomics features was determined.ResultsModels with conventional PET and dissemination features showed the highest predictive value (CV-AUC: 0.72–0.75). Dissemination features had the highest relative importance in these models. No lesion selection approach showed significantly higher predictive value compared to our previous model. Oversampling combined with RFE resulted in highest CV-AUCs.ConclusionRegardless of the applied lesion selection or feature selection approach and feature reduction methods, patient level conventional PET features and dissemination features have the highest predictive value.Trial registration number and date: EudraCT: 2006–005174-42, 01–08-2008.

A novel random subspace method considering complementarity between unsupervised and supervised deep representation features for soft sensors

Unsupervised and supervised deep learning extract effective and abstract features from different perspectives, which have been successfully applied in soft sensors. However, few studies have fused them and explored the complementary effect between the two kinds of features, which limits the utilization of comprehensive prediction information. To address the problem, a novel random subspace method with stacked auto-encoder (SAE) and bidirectional long short-term memory (Bi-LSTM), named RS-SBL, is proposed for soft sensors. Firstly, unsupervised and supervised deep representation features are extracted by SAE and Bi-LSTM, respectively. Secondly, to leverage the complementarity of the fusion features, an improved random subspcae (RS) method with a structure sparsity learning model is designed to discriminate the relative importance of different features and generate ensemble prediction results. Finally, the experiments on two real-world industrial nonlinear processes demonstrate that the proposed RS-SBL with the feature fusion strategy improves the prediction performance, and outperforms the other comparison soft sensor models.