Machine Learning applied to student attentiveness detection: Using emotional and non-emotional measures.

Electronic learning (e-learning) is considered the new norm of learning. One of the significant drawbacks of e-learning in comparison to the traditional classroom is that teachers cannot monitor the students' attentiveness. Previous literature used physical facial features or emotional states in detecting attentiveness. Other studies proposed combining physical and emotional facial features; however, a mixed model that only used a webcam was not tested. The study objective is to develop a machine learning (ML) model that automatically estimates students' attentiveness during e-learning classes using only a webcam. The model would help in evaluating teaching methods for e-learning. This study collected videos from seven students. The webcam of personal computers is used to obtain a video, from which we build a feature set that characterizes a student's physical and emotional state based on their face. This characterization includes eye aspect ratio (EAR), Yawn aspect ratio (YAR), head pose, and emotional states. A total of eleven variables are used in the training and validation of the model. ML algorithms are used to estimate individual students' attention levels. The ML models tested are decision trees, random forests, support vector machines (SVM), and extreme gradient boosting (XGBoost). Human observers' estimation of attention level is used as a reference. Our best attention classifier is the XGBoost, which achieved an average accuracy of 80.52%, with an AUROC OVR of 92.12%. The results indicate that a combination of emotional and non-emotional measures can generate a classifier with an accuracy comparable to other attentiveness studies. The study would also help assess the e-learning lectures through students' attentiveness. Hence will assist in developing the e-learning lectures by generating an attentiveness report for the tested lecture.

Machine learning prediction of mortality in Acute Myocardial Infarction.

BACKGROUND: Acute Myocardial Infarction (AMI) is the leading cause of death in Portugal and globally. The present investigation created a model based on machine learning for predictive analysis of mortality in patients with AMI upon admission, using different variables to analyse their impact on predictive models. METHODS: Three experiments were built for mortality in AMI in a Portuguese hospital between 2013 and 2015 using various machine learning techniques. The three experiments differed in the number and type of variables used. We used a discharged patients' episodes database, including administrative data, laboratory data, and cardiac and physiologic test results, whose primary diagnosis was AMI. RESULTS: Results show that for Experiment 1, Stochastic Gradient Descent was more suitable than the other classification models, with a classification accuracy of 80%, a recall of 77%, and a discriminatory capacity with an AUC of 79%. Adding new variables to the models increased AUC in Experiment 2 to 81% for the Support Vector Machine method. In Experiment 3, we obtained an AUC, in Stochastic Gradient Descent, of 88% and a recall of 80%. These results were obtained when applying feature selection and the SMOTE technique to overcome imbalanced data. CONCLUSIONS: Our results show that the introduction of new variables, namely laboratory data, impacts the performance of the methods, reinforcing the premise that no single approach is adapted to all situations regarding AMI mortality prediction. Instead, they must be selected, considering the context and the information available. Integrating Artificial Intelligence (AI) and machine learning with clinical decision-making can transform care, making clinical practice more efficient, faster, personalised, and effective. AI emerges as an alternative to traditional models since it has the potential to explore large amounts of information automatically and systematically.

Big data and machine learning to tackle diabetes management.

BACKGROUND: Type 2 Diabetes (T2D) diagnosis is based solely on glycaemia, even though it is an endpoint of numerous dysmetabolic pathways. Type 2 Diabetes complexity is challenging in a real-world scenario; thus, dissecting T2D heterogeneity is a priority. Cluster analysis, which identifies natural clusters within multidimensional data based on similarity measures, poses a promising tool to unravel Diabetes complexity. METHODS: In this review, we scrutinize and integrate the results obtained in most of the works up to date on cluster analysis and T2D. RESULTS: To correctly stratify subjects and to differentiate and individualize a preventive or therapeutic approach to Diabetes management, cluster analysis should be informed with more parameters than the traditional ones, such as etiological factors, pathophysiological mechanisms, other dysmetabolic co-morbidities, and biochemical factors, that is the millieu. Ultimately, the above-mentioned factors may impact on Diabetes and its complications. Lastly, we propose another theoretical model, which we named the Integrative Model. We differentiate three types of components: etiological factors, mechanisms and millieu. Each component encompasses several factors to be projected in separate 2D planes allowing an holistic interpretation of the individual pathology. CONCLUSION: Fully profiling the individuals, considering genomic and environmental factors, and exposure time, will allow the drive to precision medicine and prevention of complications.

A stacking-based artificial intelligence framework for an effective detection and localization of colon polyps.

Polyp detection through colonoscopy is a widely used method to prevent colorectal cancer. The automation of this process aided by artificial intelligence allows faster and improved detection of polyps that can be missed during a standard colonoscopy. In this work, we propose to implement various object detection algorithms for polyp detection. To improve the mean average precision (mAP) of the detection, we combine the baseline models through a stacking approach. The experiments demonstrate the potential of this new methodology, which can reduce the workload for oncologists and increase the precision of the localization of polyps. Our proposal achieves a mAP of 0.86, translated into an improvement of 34.9% compared to the best baseline model and 28.8% with respect to the weighted boxes fusion ensemble technique.

Object detection for automatic cancer cell counting in zebrafish xenografts.

Cell counting is a frequent task in medical research studies. However, it is often performed manually; thus, it is time-consuming and prone to human error. Even so, cell counting automation can be challenging to achieve, especially when dealing with crowded scenes and overlapping cells, assuming different shapes and sizes. In this paper, we introduce a deep learning-based cell detection and quantification methodology to automate the cell counting process in the zebrafish xenograft cancer model, an innovative technique for studying tumor biology and for personalizing medicine. First, we implemented a fine-tuned architecture based on the Faster R-CNN using the Inception ResNet V2 feature extractor. Second, we performed several adjustments to optimize the process, paying attention to constraints such as the presence of overlapped cells, the high number of objects to detect, the heterogeneity of the cells' size and shape, and the small size of the data set. This method resulted in a median error of approximately 1% of the total number of cell units. These results demonstrate the potential of our novel approach for quantifying cells in poorly labeled images. Compared to traditional Faster R-CNN, our method improved the average precision from 71% to 85% on the studied data set.

Metabolic Footprint, towards Understanding Type 2 Diabetes beyond Glycemia.

Type 2 diabetes (T2D) heterogeneity is a major determinant of complications risk and treatment response. Using cluster analysis, we aimed to stratify glycemia within metabolic multidimensionality and extract pathophysiological insights out of metabolic profiling. We performed a cluster analysis to stratify 974 subjects (PREVADIAB2 cohort) with normoglycemia, prediabetes, or non-treated diabetes. The algorithm was informed by age, anthropometry, and metabolic milieu (glucose, insulin, C-peptide, and free fatty acid (FFA) levels during the oral glucose tolerance test OGTT). For cluster profiling, we additionally used indexes of metabolism mechanisms (e.g., tissue-specific insulin resistance, insulin clearance, and insulin secretion), non-alcoholic fatty liver disease (NAFLD), and glomerular filtration rate (GFR). We found prominent heterogeneity within two optimal clusters, mainly representing normometabolism (Cluster-I) or insulin resistance and NAFLD (Cluster-II), at higher granularity. This was illustrated by sub-clusters showing similar NAFLD prevalence but differentiated by glycemia, FFA, and GFR (Cluster-II). Sub-clusters with similar glycemia and FFA showed dissimilar insulin clearance and secretion (Cluster-I). This work reveals that T2D heterogeneity can be captured by a thorough metabolic milieu and mechanisms profiling-metabolic footprint. It is expected that deeper phenotyping and increased pathophysiology knowledge will allow to identify subject's multidimensional profile, predict their progression, and treat them towards precision medicine.

Finding manoeuvre motifs in vehicle telematics.

Driving behaviour has a great impact on road safety. A popular way of analysing driving behaviour is to move the focus to the manoeuvres as they give useful information about the driver who is performing them. In this paper, we investigate a new way of identifying manoeuvres from vehicle telematics data, through motif detection in time-series. We implement a modified version of the Extended Motif Discovery (EMD) algorithm, a classical variable-length motif detection algorithm for time-series and we applied it to the UAH-DriveSet, a publicly available naturalistic driving dataset. After a systematic exploration of the extracted motifs, we were able to conclude that the EMD algorithm was not only capable of extracting simple manoeuvres such as accelerations, brakes and curves, but also more complex manoeuvres, such as lane changes and overtaking manoeuvres, which validates motif discovery as a worthwhile line for future research.

Virtual genetic diagnosis for familial hypercholesterolemia powered by machine learning.

AIMS: Familial hypercholesterolemia (FH) is the most common genetic disorder of lipid metabolism. The gold standard for FH diagnosis is genetic testing, available, however, only in selected university hospitals. Clinical scores - for example, the Dutch Lipid Score - are often employed as alternative, more accessible, albeit less accurate FH diagnostic tools. The aim of this study is to obtain a more reliable approach to FH diagnosis by a "virtual" genetic test using machine-learning approaches. METHODS AND RESULTS: We used three machine-learning algorithms (a classification tree (CT), a gradient boosting machine (GBM), a neural network (NN)) to predict the presence of FH-causative genetic mutations in two independent FH cohorts: the FH Gothenburg cohort (split into training data (N = 174) and internal test (N = 74)) and the FH-CEGP Milan cohort (external test, N = 364). By evaluating their area under the receiver operating characteristic (AUROC) curves, we found that the three machine-learning algorithms performed better (AUROC 0.79 (CT), 0.83 (GBM), and 0.83 (NN) on the Gothenburg cohort, and 0.70 (CT), 0.78 (GBM), and 0.76 (NN) on the Milan cohort) than the clinical Dutch Lipid Score (AUROC 0.68 and 0.64 on the Gothenburg and Milan cohorts, respectively) in predicting carriers of FH-causative mutations. CONCLUSION: In the diagnosis of FH-causative genetic mutations, all three machine-learning approaches we have tested outperform the Dutch Lipid Score, which is the clinical standard. We expect these machine-learning algorithms to provide the tools to implement a virtual genetic test of FH. These tools might prove particularly important for lipid clinics without access to genetic testing.

Clustering Clinical Data in R.

We are currently witnessing a paradigm shift from evidence-based medicine to precision medicine, which has been made possible by the enormous development of technology. The advances in data mining algorithms will allow us to integrate trans-omics with clinical data, contributing to our understanding of pathological mechanisms and massively impacting on the clinical sciences. Cluster analysis is one of the main data mining techniques and allows for the exploration of data patterns that the human mind cannot capture.This chapter focuses on the cluster analysis of clinical data, using the statistical software, R. We outline the cluster analysis process, underlining some clinical data characteristics. Starting with the data preprocessing step, we then discuss the advantages and disadvantages of the most commonly used clustering algorithms and point to examples of their applications in clinical work. Finally, we briefly discuss how to perform validation of clusters. Throughout the chapter we highlight R packages suitable for each computational step of cluster analysis.

Correction: Modelling innovation performance of European regions using multi-output neural networks.

[This corrects the article DOI: 10.1371/journal.pone.0185755.].

Modelling innovation performance of European regions using multi-output neural networks.

Regional innovation performance is an important indicator for decision-making regarding the implementation of policies intended to support innovation. However, patterns in regional innovation structures are becoming increasingly diverse, complex and nonlinear. To address these issues, this study aims to develop a model based on a multi-output neural network. Both intra- and inter-regional determinants of innovation performance are empirically investigated using data from the 4th and 5th Community Innovation Surveys of NUTS 2 (Nomenclature of Territorial Units for Statistics) regions. The results suggest that specific innovation strategies must be developed based on the current state of input attributes in the region. Thus, it is possible to develop appropriate strategies and targeted interventions to improve regional innovation performance. We demonstrate that support of entrepreneurship is an effective instrument of innovation policy. We also provide empirical support that both business and government R&D activity have a sigmoidal effect, implying that the most effective R&D support should be directed to regions with below-average and average R&D activity. We further show that the multi-output neural network outperforms traditional statistical and machine learning regression models. In general, therefore, it seems that the proposed model can effectively reflect both the multiple-output nature of innovation performance and the interdependency of the output attributes.