Learning Approach Research Articles

Estimating the volume of penumbra in rodents using DTI and stack-based ensemble machine learning framework

BackgroundThis study investigates the potential of diffusion tensor imaging (DTI) in identifying penumbral volume (PV) compared to the standard gadolinium-required perfusion–diffusion mismatch (PDM), utilizing a stack-based ensemble machine learning (ML) approach with enhanced explainability.MethodsSixteen male rats were subjected to middle cerebral artery occlusion. The penumbra was identified using PDM at 30 and 90 min after occlusion. We used 11 DTI-derived metrics and 14 distance-based features to train five voxel-wise ML models. The model predictions were integrated using stack-based ensemble techniques. ML-estimated and PDM-defined PVs were compared to evaluate model performance through volume similarity assessment, the Pearson correlation analysis, and Bland–Altman analysis. Feature importance was determined for explainability.ResultsIn the test rats, the ML-estimated median PV was 106.4 mL (interquartile range 44.6–157.3 mL), whereas the PDM-defined median PV was 102.0 mL (52.1–144.9 mL). These PVs had a volume similarity of 0.88 (0.79–0.96), a Pearson correlation coefficient of 0.93 (p < 0.001), and a Bland–Altman bias of 2.5 mL (2.4% of the mean PDM-defined PV), with 95% limits of agreement ranging from -44.9 to 49.9 mL. Among the features used for PV prediction, the mean diffusivity was the most important feature.ConclusionsOur study confirmed that PV can be estimated using DTI metrics with a stack-based ensemble ML approach, yielding results comparable to the volume defined by the standard PDM. The model explainability enhanced its clinical relevance. Human studies are warranted to validate our findings.Relevance statementThe proposed DTI-based ML model can estimate PV without the need for contrast agent administration, offering a valuable option for patients with kidney dysfunction. It also can serve as an alternative if perfusion map interpretation fails in the clinical setting.Key points• Penumbral volume can be estimated by DTI combined with stack-based ensemble ML.• Mean diffusivity was the most important feature used for predicting penumbral volume.• The proposed approach can be beneficial for patients with kidney dysfunction.Graphical

A Multipolicy Deep Reinforcement Learning Approach for Multiobjective Joint Routing and Scheduling in Deterministic Networks

A Multipolicy Deep Reinforcement Learning Approach for Multiobjective Joint Routing and Scheduling in Deterministic Networks

Pure tone audiogram classification using deep learning techniques.

Pure tone audiometry has played a critical role in audiology as the initial diagnostic tool, offering vital insights for subsequent analyses. This study aims to develop a robust deep learning framework capable of accurately classifying audiograms across various commonly encountered tasks. This single-centre retrospective study was conducted in accordance with the STROBE guidelines. A total of 12 518 audiograms were collected from 6259 patients aged between 4 and 96 years, who underwent pure tone audiometry testing between February 2018 and April 2022 at Tongji Hospital, Tongji Medical College, Wuhan, China. Three experienced audiologists independently annotated the audiograms, labelling the hearing loss in degrees, types and configurations of each audiogram. A deep learning framework was developed and utilised to classify audiograms across three tasks: determining the degrees of hearing loss, identifying the types of hearing loss, and categorising the configurations of audiograms. The classification performance was evaluated using four commonly used metrics: accuracy, precision, recall and F1-score. The deep learning method consistently outperformed alternative methods, including K-Nearest Neighbors, ExtraTrees, Random Forest, XGBoost, LightGBM, CatBoost and FastAI Net, across all three tasks. It achieved the highest accuracy rates, ranging from 96.75% to 99.85%. Precision values fell within the range of 88.93% to 98.41%, while recall values spanned from 89.25% to 98.38%. The F1-score also exhibited strong performance, ranging from 88.99% to 98.39%. This study demonstrated that a deep learning approach could accurately classify audiograms into their respective categories and could contribute to assisting doctors, particularly those lacking audiology expertise or experience, in better interpreting pure tone audiograms, enhancing diagnostic accuracy in primary care settings, and reducing the misdiagnosis rate of hearing conditions. In scenarios involving large-scale audiological data, the automated classification system could be used as a research tool to efficiently provide a comprehensive overview and statistical analysis. In the era of mobile audiometry, our deep learning framework can also help patients quickly and reliably understand their self-tested audiograms, potentially encouraging timely consultations with audiologists for further evaluation and intervention.

Teachers’ Views on the Integration of Information and Communication Technology in the Teaching of Slovene (First Language) in Primary School

&lt;p style="text-align:justify"&gt;Information and Communication Technology (ICT) brings modern learning approaches, forms and methods to education, and with them new opportunities for learning and teaching. The use of ICT in the classroom motivates students, enables them to participate, provides individualisation and differentiation, interactivity, and encourages creativity and critical thinking. Teachers play a very important role in this, as they are the ones who can provide students with a stimulating learning environment, including through the use of ICT. As Slovene (first language) is the subject with the most teaching hours in public primary schools in the Republic of Slovenia, we conducted an empirical study on the views of general and subject teachers of Slovene (N = 190) on the integration of information and communication technology in teaching. Teachers’ opinions were investigated through quantitative research. We found that teachers frequently use ICT in Slovene lessons, namely for repeating and consolidating learning material and for presenting new learning material, and that ICT is used most often in frontal forms of learning. Teachers mostly use LCD projectors, computers, interactive whiteboards, e-workbooks, audio recordings and videos. We also found that teachers are moderately satisfied with the available ICT equipment and e-materials, and that they feel moderately competent when it comes to using ICT in Slovene lessons. Teachers also used ICT during the COVID-19 epidemic. We also found that there are no statistically significant differences in the use of ICT between teachers teaching Slovene according to their teaching level, age and years of service.&lt;/p&gt;

Enhancing equity in access to automated insulin delivery systems in an ethnically and socioeconomically diverse group of children with type 1 diabetes

IntroductionManufacturer-supported didactic teaching programmes offer effective automated insulin delivery (AID) systems onboarding in children and young people (CYP) with type 1 diabetes (T1D). However, this approach has limited flexibility to...

Assessing and predicting green gentrification susceptibility using an integrated machine learning approach

ABSTRACT Greenery initiatives, such as green infrastructures (GIs), create sustainable and climate-resilient environments. However, they can also have unintended consequences, such as displacement and gentrification in low-income areas. This paper proposes an integrated machine learning (ML) approach that combines both unsupervised and supervised ML algorithms. First, 35 indicators that contribute to green gentrification were identified and categorised into 7 categories: social, economic, demographic, housing, household, amenities, and GIs. Second, data was collected for all census tracts in New York City. Third, the green gentrification susceptibility was modelled into 6 levels using k-means clustering analysis, which is an unsupervised ML model. Fourth, the Technique for Order of Preference by Similarity to the Ideal Solution (TOPSIS) was used to map the census tracts to their green gentrification susceptibility level. Finally, different supervised ML algorithms were trained and tested to predict the green gentrification susceptibility. The results showed that the artificial neural network (ANN) model is the most accurate in classifying and predicting the green gentrification susceptibility with an overall accuracy of 96%. Moreover, the outcomes showed that the Normal Difference Vegetation Index (NDVI), the proximity to GIs, the GIs frequency, and the total area of GIs were identified as the most important indicators to predict green gentrification susceptibility. Ultimately, the proposed approach allows practitioners and researchers to perform micro-level (i.e. on the census-tracts level) predictions and inferences about green gentrification susceptibility. This allows more focused and targeted mitigation actions to be designed and implemented in the most affected communities, thus promoting environmental justice.

MAC: a meta-learning approach for feature learning and recombination

MAC: a meta-learning approach for feature learning and recombination

Multi-class boosting for the analysis of multiple incomplete views on microbiome data

BackgroundMicrobiome dysbiosis has recently been associated with different diseases and disorders. In this context, machine learning (ML) approaches can be useful either to identify new patterns or learn predictive models. However, data to be fed to ML methods can be subject to different sampling, sequencing and preprocessing techniques. Each different choice in the pipeline can lead to a different view (i.e., feature set) of the same individuals, that classical (single-view) ML approaches may fail to simultaneously consider. Moreover, some views may be incomplete, i.e., some individuals may be missing in some views, possibly due to the absence of some measurements or to the fact that some features are not available/applicable for all the individuals. Multi-view learning methods can represent a possible solution to consider multiple feature sets for the same individuals, but most existing multi-view learning methods are limited to binary classification tasks or cannot work with incomplete views.ResultsWe propose irBoost.SH, an extension of the multi-view boosting algorithm rBoost.SH, based on multi-armed bandits. irBoost.SH solves multi-class classification tasks and can analyze incomplete views. At each iteration, it identifies one winning view using adversarial multi-armed bandits and uses its predictions to update a shared instance weight distribution in a learning process based on boosting. In our experiments, performed on 5 multi-view microbiome datasets, the model learned by irBoost.SH always outperforms the best model learned from a single view, its closest competitor rBoost.SH, and the model learned by a multi-view approach based on feature concatenation, reaching an improvement of 11.8% of the F1-score in the prediction of the Autism Spectrum disorder and of 114% in the prediction of the Colorectal Cancer disease.ConclusionsThe proposed method irBoost.SH exhibited outstanding performances in our experiments, also compared to competitor approaches. The obtained results confirm that irBoost.SH can fruitfully be adopted for the analysis of microbiome data, due to its capability to simultaneously exploit multiple feature sets obtained through different sequencing and preprocessing pipelines.

Winter Conference 2023 Editorial.

Multimorbidity, the existence of two or more concurrent chronic conditions in a single individual, represents a major global health challenge. The Nutrition Society's 2023 Winter Conference at the Royal Society, London focused on the topic of 'Diet and lifestyle strategies for prevention and management of multimorbidity', with symposia designed to explore pathways for prevention of multimorbidity across the lifecourse, the role of ageing, the gut-brain-heart connection and lifestyle strategies for prevention and management of multimorbidity. It also considered machine learning and precision nutrition approaches for addressing research challenges in multimorbidity. The opening plenary lecture discussed advancing diet and lifestyle research to address the increasing burden and complexity of multimorbidity. The two-day programme concluded with a plenary which addressed the key dietary risk factors and policies in multimorbidity prevention.

The Development of an Opioid Misuse Training Program for Physical Therapists: A Learning Community Approach.

Musculoskeletal pain and opioid misuse frequently co-occur. The US Preventive Services Task Force calls for all health care providers to be trained to screen for misuse and/or opioid use disorder. The purpose of this study was to develop and implement an opioid misuse training program that could be used by physical therapists. Thirteen practicing physical therapists were invited to participate in a curriculum development project. Using the Curriculum Framework, a collaborative learning approach was used to develop an opioid misuse training program and training manual for physical therapists. Four training sessions were provided virtually every 2 weeks. Topics included an introduction to the opioid crisis, screening, assessing, and communicating with patients and with the health care team about opioid misuse. Each didactic session was followed by a participant feedback session where participants provided recommendations on improving the training content and their impressions on the barriers and facilitators to incorporating the training into practice. A companion training manual was created and sent to participants for comment. Participants were asked over email to describe whether and how they incorporated training materials into clinical practice during the training curriculum. All participants attended sessions 1-3. Twelve (92.3%) attended the fourth session. Based on the participants' feedback, training sessions were edited, and a companion training manual was finalized and distributed to each participant. After the fourth session, 9 participants (69.2%) reported using what they learned in the training. During the participant feedback sessions, participants regarded the training as important. Some participants expressed barriers to discussing opioids with patients and concerns about whether the training was within physical therapists' scope of practice. An iteratively developed training program for physical therapists to address opioid misuse was acceptable, feasible, and provided immediate practice change by most participants.

Teaching of Physics in Engineering from Problem-Based and Project-Based Learning Approaches

Teaching of Physics in Engineering from Problem-Based and Project-Based Learning Approaches

Assessing the quality of chest compressions with a DIY low-cost manikin (LoCoMan) versus a standard manikin: a quasi-experimental study in primary education.

Extending the access to cardiopulmonary resuscitation (CPR) training to a wider public is an important step in increasing survivability of out-of-hospital cardiac arrest. However, often price and maintenance of CPR manikins are barriers that prevent training at schools. This study aims to evaluate the learning of hands-only (HO) CPR by practicing with a low-cost manikin (LoCoMan) with visual qualitative feedback and to compare the results with the skills acquired by practice on a conventional manikin. A quasi-experimental study with 193 schoolchildren (10 to 12years old) who were allocated to two groups: the LoCoMan group was taught via an integrative approach (science combined with physical education (PE)) and practiced on a handmade manikin, and a control group practiced in a traditional setting with a commercial manikin (Resusci Junior, Laerdal, Norway). All participants practiced for 1 hands-on skill session before performing a post-test on an instrumented CPR manikin. The outcomes including HO-CPR performance variables were compared between groups. The LoCoMan and control groups both achieved acceptable percentage of HO-CPR quality (57% and 71%, p = 0.004). Among 6th-graders, there were no significant differences in HO-CPR quality between LoCoMan 68% and control 71%, p = 0.66. The control group achieved better chest compression depth while the LoCoMan group showed more compressions with adequate chest recoil. Conclusion: Schoolchildren are able to build and use a low-cost manikin with visual feedback. The integrative learning approach used in this study may be a feasible alternative methodology for training and learning HO-CPR in schools when commercial manikins are not available. What is Known: • Access to CPR training should be universal and independent of age, location, financial means, or access to qualified instructors. • Scientific societies promote the implementation of CPR in schools, so that teachers and schoolchildren can play a multiplier role in their environment, but the gap in CPR learning is related to cultural, economic factors or access to resources and materials. What is New: • LoCoMan may be a useful device for teaching and learning CPR in schoolchildren from the age of 10 and upwards. • LOCOMAN shows that it is feasible and possible to build a low-cost manikin (about €5 in the European Region) and to integrate it into an integrative educational project, and outlines how this could be done. this approach can be an incentive for teachers to attempt teaching CPR, but also for education outside the formal environment.

Machine Learning-Based Integrated Multiomics Characterization of Colorectal Cancer Reveals Distinctive Metabolic Signatures.

The metabolic signature identification of colorectal cancer is critical for its early diagnosis and therapeutic approaches that will significantly block cancer progression and improve patient survival. Here, we combined an untargeted metabolic analysis strategy based on internal extractive electrospray ionization mass spectrometry and the machine learning approach to analyze metabolites in 173 pairs of cancer samples and matched normal tissue samples to build robust metabolic signature models for diagnostic purposes. Screening and independent validation of metabolic signatures from colorectal cancers via machine learning methods (Logistic Regression_L1 for feature selection and eXtreme Gradient Boosting for classification) was performed to generate a panel of seven signatures with good diagnostic performance (the accuracy of 87.74%, sensitivity of 85.82%, and specificity of 89.66%). Moreover, seven signatures were evaluated according to their ability to distinguish between cancer and normal tissues, with the metabolic molecule PC (30:0) showing good diagnostic performance. In addition, genes associated with PC (30:0) were identified by multiomics analysis (combining metabolic data with transcriptomic data analysis) and our results showed that PC (30:0) could promote the proliferation of colorectal cancer cell SW480, revealing the correlation between genetic changes and metabolic dysregulation in cancer. Overall, our results reveal potential determinants affecting metabolite dysregulation, paving the way for a mechanistic understanding of altered tissue metabolites in colorectal cancer and design interventions for manipulating the levels of circulating metabolites.

Deep Learning for Predicting Progression of Patellofemoral Osteoarthritis Based on Lateral Knee Radiographs, Demographic Data, and Symptomatic Assessments.

In this study, we propose a novel framework that utilizes deep learning and attention mechanisms to predict the radiographic progression of patellofemoral osteoarthritis (PFOA) over a period of 7 years. This study included subjects (1,832 subjects, 3,276 knees) from the baseline of the Multicenter Osteoarthritis Study (MOST). Patellofemoral joint regions of interest were identified using an automated landmark detection tool (BoneFinder) on lateral knee X-rays. An end-to-end deep learning method was developed for predicting PFOA progression based on imaging data in a five-fold cross-validation setting. To evaluate the performance of the models, a set of baselines based on known risk factors were developed and analyzed using gradient boosting machine (GBM). Risk factors included age, sex, body mass index, and Western Ontario and McMaster Universities Arthritis Index score, and the radiographic osteoarthritis stage of the tibiofemoral joint (Kellgren and Lawrence [KL] score). Finally, to increase predictive power, we trained an ensemble model using both imaging and clinical data. Among the individual models, the performance of our deep convolutional neural network attention model achieved the best performance with an area under the receiver operating characteristic curve (AUC) of 0.856 and average precision (AP) of 0.431, slightly outperforming the deep learning approach without attention (AUC = 0.832, AP = 0.4) and the best performing reference GBM model (AUC = 0.767, AP = 0.334). The inclusion of imaging data and clinical variables in an ensemble model allowed statistically more powerful prediction of PFOA progression (AUC = 0.865, AP = 0.447), although the clinical significance of this minor performance gain remains unknown. The spatial attention module improved the predictive performance of the backbone model, and the visual interpretation of attention maps focused on the joint space and the regions where osteophytes typically occur. This study demonstrated the potential of machine learning models to predict the progression of PFOA using imaging and clinical variables. These models could be used to identify patients who are at high risk of progression and prioritize them for new treatments. However, even though the accuracy of the models were excellent in this study using the MOST dataset, they should be still validated using external patient cohorts in the future.

Recognition of local fiber orientation state in prepreg platelet molded composites via deep learning

This paper presents a novel deep learning approach for reconstruction of local through-the-thickness fiber orientation distribution (FOD) in discontinuous long-fiber, prepreg-platelet molded composites (PPMC). The thermal-residual strains on composite surfaces are used as inputs to train a fully convolutional neural network, named U-Net, to predict spatially varying, local FOD. High fidelity synthetic data was generated via computational simulation of PPMC with stochastic material orientation state and was used for training the deep learning model. The proposed U-Net model allowed for rapid recognition of PPMC morphology by solving the inverse structural mechanics problem of determining the average fiber orientation through the composite thickness based on the provided surface strain measurements. Upon training and validation, the U-Net deep learning model was deployed to rapidly predict complex distributions of the local through-the-thickness FOD in PPMC.

AI-Assisted Detection of Interproximal, Occlusal, and Secondary Caries on Bite-Wing Radiographs: A Single-Shot Deep Learning Approach.

Tooth decay is a common oral disease worldwide, but errors in diagnosis can often be made in dental clinics, which can lead to a delay in treatment. This study aims to use artificial intelligence (AI) for the automated detection and localization of secondary, occlusal, and interproximal (D1, D2, D3) caries types on bite-wing radiographs. The eight hundredand sixty bite-wing radiographs were collected from the School of Dentistry database. Pre-processing and data augmentation operations were performed. Interproximal (D1, D2, D3), secondary, and occlusal caries on bite-wing radiographs were annotated by two oral radiologists. The data were split into 80% for training, 10% for validation, and 10% for testing. The AI-based training process was conducted using the YOLOv8 algorithm. A clinical decision support system interface was designed using the Python PyQT5 library, allowing for the use of dental caries detection without the need for complex programming procedures. In the test images, the average precision, average sensitivity, and average F1 score values for secondary, occlusal, and interproximal caries were obtained as 0.977, 0.932, and 0.954, respectively. The AI-based dental caries detection system yielded highly successful results in the test, receiving full approval from dentists for clinical use. YOLOv8 has the potential to increase sensitivity and reliability while reducing the burden on dentists and can prevent diagnostic errors in dental clinics.

ST-Phys: Unsupervised Spatio-Temporal Contrastive Remote Physiological Measurement.

Remote photoplethysmography (rPPG) is a non-contact method that employs facial videos for measuring physiological parameters. Existing rPPG methods have achieved remarkable performance. However, the success mainly profits from supervised learning over massive labeled data. On the other hand, existing unsupervised rPPG methods fail to fully utilize spatio-temporal features and encounter challenges in low-light or noise environments. To address these problems, we propose an unsupervised contrast learning approach, ST-Phys. We incorporate a low-light enhancement module, a temporal dilated module, and a spatial enhanced module to better deal with long-term dependencies under the random low-light conditions. In addition, we design a circular margin loss, wherein rPPG signals originating from identical videos are attracted, while those from distinct videos are repelled. Our method is assessed on six openly accessible datasets, including RGB and NIR videos. Extensive experiments reveal the superior performance of our proposed ST-Phys over state-of-the-art unsupervised rPPG methods. Moreover, it offers advantages in parameter reduction and noise robustness.

NeuralCMS: A deep learning approach to study Jupiter's interior

NASA's Juno mission provided exquisite measurements of Jupiter's gravity field that together with the Galileo entry probe atmospheric measurements constrains the interior structure of the giant planet. Inferring its interior structure range remains a challenging inverse problem requiring a computationally intensive search of combinations of various planetary properties, such as the cloud-level temperature, composition, and core features, requiring the computation of sim \(10^9\) interior models. We propose an efficient deep neural network (DNN) model to generate high-precision wide-ranged interior models based on the very accurate but computationally demanding concentric MacLaurin spheroid (CMS) method. We trained a sharing-based DNN with a large set of CMS results for a four-layer interior model of Jupiter, including a dilute core, to accurately predict the gravity moments and mass, given a combination of interior features. We evaluated the performance of the trained DNN (NeuralCMS) to inspect its predictive limitations. NeuralCMS shows very good performance in predicting the gravity moments, with errors comparable with the uncertainty due to differential rotation, and a very accurate mass prediction. This allowed us to perform a broad parameter space search by computing only sim \(10^4\) actual CMS interior models, resulting in a large sample of plausible interior structures, and reducing the computation time by a factor of \(10^5\). Moreover, we used a DNN explainability algorithm to analyze the impact of the parameters setting the interior model on the predicted observables, providing information on their nonlinear relation.

A machine learning approach to evaluate the impact of virtual balance/cognitive training on fall risk in older women

A machine learning approach to evaluate the impact of virtual balance/cognitive training on fall risk in older women

DLEmotion: Deep learning-based emotion classification using visual attention location information

The correct recognition of each person's emotion can be an important step in knowing mental states and reactions. Because of the complexities of human personality, this is a difficult task and providing a method that can classify emotions with high accuracy can be useful. The purpose of doing this work is to provide a suitable method for recognizing emotions in human subjects. In other words, in this research the aim is to analyze eye tracking data using deep learning approaches and for this purpose, deep neural networks are used to classify fixations in different emotional states. The eye data of 10 women and 7 men who watched 79 emotional images are first collected. In this way, for each fixation point obtained from the subjects, a patch of 80 x 80 pixels was selected from the images and we used these patches of the obtained image as the input to the deep neural networks. The images shown to the subjects have three basic dimensions of emotion, i.e. Valence, Arousal and Dominance. Data classification was done by transfer learning and using Xception, InceptionV3 and DenseNet201 deep neural networks whose weights were trained on ImageNet datasets. We used these three neural networks in our work because they have relatively equal number of parameters and will help us in better analysis of the obtained results. In general, the results show that the points obtained from subjects' fixations location on images with three basic dimensions of emotion can be used to classify emotions. We will also see that changing the size of the patches obtained from the subject's fixation will not have much effect on the test results, which shows that deep neural networks are a powerful and accurate tool for processing people's eye fixation points.