Tenfold Cross-validation Procedure Research Articles

Detection and classification of adult epilepsy using hybrid deep learning approach

The electroencephalogram (EEG) has emerged over the past few decades as one of the key tools used by clinicians to detect seizures and other neurological abnormalities of the human brain. The proper diagnosis of epilepsy is crucial due to its distinctive nature and the subsequent negative effects of epileptic seizures on patients. The classification of minimally pre-processed, raw multichannel EEG signal recordings is the foundation of this article’s unique method for identifying seizures in pre-adult patients. The new method makes use of the automatic feature learning capabilities of a three-dimensional deep convolution auto-encoder (3D-DCAE) associated with a neural network-based classifier to build an integrated framework that endures training in a supervised manner to attain the highest level of classification precision among brain state signals, both ictal and interictal. A pair of models were created and evaluated for testing and assessing our method, utilizing three distinct EEG data section lengths, and a tenfold cross-validation procedure. Based on five evaluation criteria, the labelled hybrid convolutional auto-encoder (LHCAE) model, which utilizes a classifier based on bidirectional long short-term memory (Bi-LSTM) and an EEG segment length of 4 s, had the best efficiency. This proposed model has 99.08 ± 0.54% accuracy, 99.21 ± 0.50% sensitivity, 99.11 ± 0.57% specificity, 99.09 ± 0.55% precision, and an F1-score of 99.16 ± 0.58%, according to the publicly available Children’s Hospital Boston (CHB) dataset. Based on the obtained outcomes, the proposed seizure classification model outperforms the other state-of-the-art method’s performance in the same dataset.

Toward precision medicine using a “digital twin” approach: modeling the onset of disease-specific brain atrophy in individuals with multiple sclerosis

Digital Twin (DT) is a novel concept that may bring a paradigm shift for precision medicine. In this study we demonstrate a DT application for estimating the age of onset of disease-specific brain atrophy in individuals with multiple sclerosis (MS) using brain MRI. We first augmented longitudinal data from a well-fitted spline model derived from a large cross-sectional normal aging data. Then we compared different mixed spline models through both simulated and real-life data and identified the mixed spline model with the best fit. Using the appropriate covariate structure selected from 52 different candidate structures, we augmented the thalamic atrophy trajectory over the lifespan for each individual MS patient and a corresponding hypothetical twin with normal aging. Theoretically, the age at which the brain atrophy trajectory of an MS patient deviates from the trajectory of their hypothetical healthy twin can be considered as the onset of progressive brain tissue loss. With a tenfold cross validation procedure through 1000 bootstrapping samples, we found the onset age of progressive brain tissue loss was, on average, 5–6 years prior to clinical symptom onset. Our novel approach also discovered two clear patterns of patient clusters: earlier onset versus simultaneous onset of brain atrophy.

3DCNN predicting brain age using diffusion tensor imaging.

Neuroimaging-based brain age prediction using deep learning is gaining popularity. However, few studies have attempted to leverage diffusion tensor imaging (DTI) to predict brain age. In this study, we proposed a 3D convolutional neural network model (3DCNN) and trained it on fractional anisotropy (FA) data from six publicly available datasets (n = 2406, age = 17-60) to estimate brain age. Implementing a two-loop nested cross-validation scheme with a tenfold cross-validation procedure, we achieved a robust prediction performance of a mean absolute error (MAE) of 2.785 and a correlation coefficient of 0.932. We also employed Grad-Cam++ to visualize the salient features of the proposed model. We identified a few highly salient fiber tracts, including the genu of corpus callosum and the left cerebellar peduncle, among others that play a pivotal role in our model. In sum, our model reliably predicted brain age and provided novel insight into age-related changes in brains' axonal structure.

Writing Quality Predictive Modeling: Integrating Register-Related Factors

The primary purpose of this study is to investigate the degree to which register knowledge, register-specific motivation, and diverse linguistic features are predictive of human judgment of writing quality in three registers—narrative, informative, and opinion. The secondary purpose is to compare the evaluation metrics of register-partitioned automated writing evaluation models in three conditions: (1) register-related factors alone, (2) linguistic features alone, and (3) the combination of these two. A total of 1006 essays ( n = 327, 342, and 337 for informative, narrative, and opinion, respectively) written by 92 fourth- and fifth-graders were examined. A series of hierarchical linear regression analyses controlling for the effects of demographics were conducted to select the most useful features to capture text quality, scored by humans, in the three registers. These features were in turn entered into automated writing evaluation predictive models with tuning of the parameters in a tenfold cross-validation procedure. The average validity coefficients (i.e., quadratic-weighed kappa, Pearson correlation r, standardized mean score difference, score deviation analysis) were computed. The results demonstrate that (1) diverse feature sets are utilized to predict quality in the three registers, and (2) the combination of register-related factors and linguistic features increases the accuracy and validity of all human and automated scoring models, especially for the registers of informative and opinion writing. The findings from this study suggest that students’ register knowledge and register-specific motivation add additional predictive information when evaluating writing quality across registers beyond that afforded by linguistic features of the paper itself, whether using human scoring or automated evaluation. These findings have practical implications for educational practitioners and scholars in that they can help strengthen consideration of register-specific writing skills and cognitive and motivational forces that are essential components of effective writing instruction and assessment.

Preoperative CT features to predict risk stratification of non-muscle invasive bladder cancer.

To investigate whether preoperative CT features can be used to predict risk stratification of non-muscle invasive bladder cancer (NMIBC). The 168 patients with pathologically confirmed NMIBC who underwent preoperative CT urography were retrospectively analyzed and were divided into training (n = 117) and testing (n = 51) sets. According to the European Association of Urology Guidelines, patients were classified into low-risk (n = 50), medium-risk (n = 23), and high-risk (n = 95) groups. A random over-sample was performed to handle the offset caused by the unbalanced groups. We measured some CT features that may help stratify which for modeling were determined using an F-test-based feature selection with a tenfold cross-validation procedure, and the Gaussian Naive Bayes model was trained on the entire training set. In the testing set, the performance of the model was evaluated. The selected CT features were the maximum and the minimum diameter of the largest tumor, whether the largest tumor is located at the trigone, and tumor number. In the testing set, the model reached a macro- and micro- AUC of 0.783 and 0.745 with an accuracy of 0.529. As for the one-vs-rest problem, the model was most effective in identifying low-risk individuals, with an AUC, accuracy, sensitivity, and specificity of 0.870, 0.647, 1.000, and 0.438, respectively; the medium-risk group reached 0.814, 0.882, 0.250, and 0.936, respectively; the identification of the high-risk group was harder, going 0.665, 0.529, 0.250, and 0.870, respectively. It is feasible to predict the risk stratification of NMIBC using preoperative CT features.

Four-Class Classification of Neuropsychiatric Disorders by Use of Functional Near-Infrared Spectroscopy Derived Biomarkers

Diagnosis of most neuropsychiatric disorders relies on subjective measures, which makes the reliability of final clinical decisions questionable. The aim of this study was to propose a machine learning-based classification approach for objective diagnosis of three disorders of neuropsychiatric or neurological origin with functional near-infrared spectroscopy (fNIRS) derived biomarkers. Thirteen healthy adolescents and sixty-seven patients who were clinically diagnosed with migraine, obsessive compulsive disorder, or schizophrenia performed a Stroop task, while prefrontal cortex hemodynamics were monitored with fNIRS. Hemodynamic and cognitive features were extracted for training three supervised learning algorithms (naïve bayes (NB), linear discriminant analysis (LDA), and support vector machines (SVM)). The performance of each algorithm in correctly predicting the class of each participant across the four classes was tested with ten runs of a ten-fold cross-validation procedure. All algorithms achieved four-class classification performances with accuracies above 81% and specificities above 94%. SVM had the highest performance in terms of accuracy (85.1 ± 1.77%), sensitivity (84 ± 1.7%), specificity (95 ± 0.5%), precision (86 ± 1.6%), and F1-score (85 ± 1.7%). fNIRS-derived features have no subjective report bias when used for automated classification purposes. The presented methodology might have significant potential for assisting in the objective diagnosis of neuropsychiatric disorders associated with frontal lobe dysfunction.

Machine learning predicts clinically significant health related quality of life improvement after sensorimotor rehabilitation interventions in chronic stroke

Health related quality of life (HRQOL) reflects individuals perceived of wellness in health domains and is often deteriorated after stroke. Precise prediction of HRQOL changes after rehabilitation interventions is critical for optimizing stroke rehabilitation efficiency and efficacy. Machine learning (ML) has become a promising outcome prediction approach because of its high accuracy and easiness to use. Incorporating ML models into rehabilitation practice may facilitate efficient and accurate clinical decision making. Therefore, this study aimed to determine if ML algorithms could accurately predict clinically significant HRQOL improvements after stroke sensorimotor rehabilitation interventions and identify important predictors. Five ML algorithms including the random forest (RF), k-nearest neighbors (KNN), artificial neural network, support vector machine and logistic regression were used. Datasets from 132 people with chronic stroke were included. The Stroke Impact Scale was used for assessing multi-dimensional and global self-perceived HRQOL. Potential predictors included personal characteristics and baseline cognitive/motor/sensory/functional/HRQOL attributes. Data were divided into training and test sets. Tenfold cross-validation procedure with the training data set was used for developing models. The test set was used for determining model performance. Results revealed that RF was effective at predicting multidimensional HRQOL (accuracy: 85%; area under the receiver operating characteristic curve, AUC-ROC: 0.86) and global perceived recovery (accuracy: 80%; AUC-ROC: 0.75), and KNN was effective at predicting global perceived recovery (accuracy: 82.5%; AUC-ROC: 0.76). Age/gender, baseline HRQOL, wrist/hand muscle function, arm movement efficiency and sensory function were identified as crucial predictors. Our study indicated that RF and KNN outperformed the other three models on predicting HRQOL recovery after sensorimotor rehabilitation in stroke patients and could be considered for future clinical application.

Revision and Extension of a Generally Applicable Group-Additivity Method for the Calculation of the Standard Heat of Combustion and Formation of Organic Molecules.

The calculation of the heats of combustion ΔH°c and formation ΔH°f of organic molecules at standard conditions is presented using a commonly applicable computer algorithm based on the group-additivity method. This work is a continuation and extension of an earlier publication. The method rests on the complete breakdown of the molecules into their constituting atoms, these being further characterized by their immediate neighbor atoms. The group contributions are calculated by means of a fast Gauss–Seidel fitting calculus using the experimental data of 5030 molecules from literature. The applicability of this method has been tested by a subsequent ten-fold cross-validation procedure, which confirmed the extraordinary accuracy of the prediction of ΔH°c with a correlation coefficient R2 and a cross-validated correlation coefficient Q2 of 1, a standard deviation σ of 18.12 kJ/mol, a cross-validated standard deviation S of 19.16 kJ/mol, and a mean absolute deviation of 0.4%. The heat of formation ΔH°f has been calculated from ΔH°c using the standard enthalpies of combustion for the elements, yielding a correlation coefficient R2 for ΔH°f of 0.9979 and a corresponding standard deviation σ of 18.14 kJ/mol.

Calculation of the Vapour Pressure of Organic Molecules by Means of a Group-Additivity Method and Their Resultant Gibbs Free Energy and Entropy of Vaporization at 298.15 K.

The calculation of the vapour pressure of organic molecules at 298.15 K is presented using a commonly applicable computer algorithm based on the group-additivity method. The basic principle of this method rests on the complete breakdown of the molecules into their constituting atoms, further characterized by their immediate neighbour atoms. The group contributions are calculated by means of a fast Gauss–Seidel fitting algorithm using the experimental data of 2036 molecules from literature. A ten-fold cross-validation procedure has been carried out to test the applicability of this method, which confirmed excellent quality for the prediction of the vapour pressure, expressed in log(pa), with a cross-validated correlation coefficient Q2 of 0.9938 and a standard deviation σ of 0.26. Based on these data, the molecules’ standard Gibbs free energy ΔG°vap has been calculated. Furthermore, using their enthalpies of vaporization, predicted by an analogous group-additivity approach published earlier, the standard entropy of vaporization ΔS°vap has been determined and compared with experimental data of 1129 molecules, exhibiting excellent conformance with a correlation coefficient R2 of 0.9598, a standard error σ of 8.14 J/mol/K and a medium absolute deviation of 4.68%.

Foliar fungal disease classification in banana plants using elliptical local binary pattern on multiresolution dual tree complex wavelet transform domain

The fungal diseases in banana cause major yield losses for millions of farmers around the globe. Early detection of these diseases helps the farmers to devise successful management strategies. The characteristic leaf blade discoloration pattern at the earlier stages of infection could be used to understand the onset of each disease. This paper demonstrates a methodology for classification of three important foliar diseases in banana, using local texture features. The disease affected regions are identified using image enhancement and color segmentation. Segmented images are converted to transform domain using three image transforms (DWT, DTCWT and Ranklet transform). Feature vector is extracted from transform domain images using LBP and its variants (ELBP, MeanELBP and MedianELBP). These texture based features are applied to five popular image classifiers and comparative performance analysis is done using ten-fold cross validation procedure. Experimental results showed best classification performance for ELBP features extracted from DTCWT domain (accuracy 95.4%, precision 93.2%, sensitivity 93.0%, Fscore 93.0% and specificity 96.4%). Compared with traditional methods of feature extraction, this novel method of fusing DTCWT with ELBP features has attained high degree of accuracy in precisely detecting and classifying fungal diseases in banana at an early stage.

Is a COVID-19 Second Wave Possible in Emilia-Romagna (Italy)? Forecasting a Future Outbreak with Particulate Pollution and Machine Learning

The Nobel laureate Niels Bohr once said that: “Predictions are very difficult, especially if they are about the future”. Nonetheless, models that can forecast future COVID-19 outbreaks are receiving special attention by policymakers and health authorities, with the aim of putting in place control measures before the infections begin to increase. Nonetheless, two main problems emerge. First, there is no a general agreement on which kind of data should be registered for judging on the resurgence of the virus (e.g., infections, deaths, percentage of hospitalizations, reports from clinicians, signals from social media). Not only this, but all these data also suffer from common defects, linked to their reporting delays and to the uncertainties in the collection process. Second, the complex nature of COVID-19 outbreaks makes it difficult to understand if traditional epidemiological models, such as susceptible, infectious, or recovered (SIR), are more effective for a timely prediction of an outbreak than alternative computational models. Well aware of the complexity of this forecasting problem, we propose here an innovative metric for predicting COVID-19 diffusion based on the hypothesis that a relation exists between the spread of the virus and the presence in the air of particulate pollutants, such as PM2.5, PM10, and NO2. Drawing on the recent assumption of 239 experts who claimed that this virus can be airborne, and further considering that particulate matter may favor this airborne route, we developed a machine learning (ML) model that has been instructed with: (i) all the COVID-19 infections that occurred in the Italian region of Emilia-Romagna, one of the most polluted areas in Europe, in the period of February–July 2020, (ii) the daily values of all the particulates taken in the same period and in the same region, and finally (iii) the chronology according to which restrictions were imposed by the Italian Government to human activities. Our ML model was then subjected to a classic ten-fold cross-validation procedure that returned a promising 90% accuracy value. Finally, the model was used to predict a possible resurgence of the virus in all the nine provinces of Emilia-Romagna, in the period of September–December 2020. To make those predictions, input to our ML model were the daily measurements of the aforementioned pollutants registered in the periods of September–December 2017/2018/2019, along with the hypothesis that the mild containment measures taken in Italy in the so-called Phase 3 are obeyed. At the time we write this article, we cannot have a confirmation of the precision of our predictions. Nevertheless, we are projecting a scenario based on an original hypothesis that makes our COVID-19 prediction model unique in the world. Its accuracy will be soon judged by history—and this, too, is science at the service of society.

Electrodiagnostic accuracy in polyneuropathies: supervised learning algorithms as a tool for practitioners.

The interpretation of electrophysiological findings may lead to misdiagnosis in polyneuropathies. We investigated the electrodiagnostic accuracy of three supervised learning algorithms (SLAs): shrinkage discriminant analysis, multinomial logistic regression, and support vector machine (SVM), and three expert and three trainee neurophysiologists. We enrolled 434 subjects with the following diagnoses: chronic inflammatory demyelinating polyneuropathy (99), Charcot-Marie-Tooth disease type 1A (124), hereditary neuropathy with liability to pressure palsy (46), diabetic polyneuropathy (67), and controls (98). In each diagnostic class, 90% of subjects were used as training set for SLAs to establish the best performing SLA by tenfold cross validation procedure and 10% of subjects were employed as test set. Performance indicators were accuracy, precision, sensitivity, and specificity. SVM showed the highest overall diagnostic accuracy both in training and test sets (90.5 and 93.2%) and ranked first in a multidimensional comparison analysis. Overall accuracy of neurophysiologists ranged from 54.5 to 81.8%. This proof of principle study shows that SVM provides a high electrodiagnostic accuracy in polyneuropathies. We suggest that the use of SLAs in electrodiagnosis should be exploited to possibly provide a diagnostic support system especially helpful for the less experienced practitioners.

A convolutional neural network to detect scoliosis treatment in radiographs.

The aim of this work is to propose a classification algorithm to automatically detect treatment for scoliosis (brace, implant or no treatment) in postero-anterior radiographs. Such automatic labelling of radiographs could represent a step towards global automatic radiological analysis. Seven hundred and ninety-six frontal radiographies of adolescents were collected (84 patients wearing a brace, 325 with a spinal implant and 387 reference images with no treatment). The dataset was augmented to a total of 2096 images. A classification model was built, composed by a forward convolutional neural network (CNN) followed by a discriminant analysis; the output was a probability for a given image to contain a brace, a spinal implant or none. The model was validated with a stratified tenfold cross-validation procedure. Performance was estimated by calculating the average accuracy. 98.3% of the radiographs were correctly classified as either reference, brace or implant, excluding 2.0% unclassified images.99.7% of brace radiographs were correctly detected, while most of the errors occurred in the reference group (i.e. 2.1% of reference images were wrongly classified). The proposed classification model, the originality of which is the coupling of a CNN with discriminant analysis, can be used to automatically label radiographs for the presence of scoliosis treatment. This information is usually missing from DICOM metadata, so such method could facilitate the use of large databases. Furthermore, the same model architecture could potentially be applied for other radiograph classifications, such as sex and presence of scoliotic deformity.

Prediction of compressive strength and portland cement composition using cross-validation and feature ranking techniques

Prediction of compressive strength and concrete composition using machine learning models is an essential feature in civil engineering applications. In the present paper, a methodology is proposed for the prediction of compressive strength of concrete and portland cement composition using three experimental data sets. Ten-fold cross-validation procedure is applied to four machine learning models Isotonic regression, Artificial neural network, Support vector machine, and Random forest. Further comparison was made with feature ranking and without feature ranking to reduce the computational time and to achieve better prediction accuracy. The accuracy of machine learning models are analyzed with four parameters, Correlation coefficient, Kendall’s tau, Mean absolute error, and Root mean square error. It is observed that better prediction capability is achieved with the ten-fold cross-validation, since it gives a statistically unbiased result. Results obtain reveals a high correlation and less error between the experimental and predicted values for all the three experimental datasets consider.

Estimating minute ventilation and air pollution inhaled dose using heart rate, breath frequency, age, sex and forced vital capacity: A pooled-data analysis.

Air pollution inhaled dose is the product of pollutant concentration and minute ventilation (). Previous studies have parameterized the relationship between and variables such as heart rate (HR) and have observed substantial inter-subject variability. In this paper, we evaluate a method to estimate with easy-to-measure variables in an analysis of pooled-data from eight independent studies. We compiled a large diverse data set that is balanced with respect to age, sex and fitness level. We used linear mixed models to estimate with HR, breath frequency (fB), age, sex, height, and forced vital capacity (FVC) as predictors. FVC was estimated using the Global Lung Function Initiative method. We log-transformed the dependent and independent variables to produce a model in the form of a power function and assessed model performance using a ten-fold cross-validation procedure. The best performing model using HR as the only field-measured parameter was = e-9.59HR2.39age0.274sex-0.204FVC0.520 with HR in beats per minute, age in years, sex is 1 for males and 2 for females, FVC in liters, and a median(IQR) cross-validated percent error of 0.664(45.4)%. The best performing model overall was = e-8.57HR1.72fB0.611age0.298sex-0.206FVC0.614, where fB is breaths per minute, and a median(IQR) percent error of 1.20(37.9)%. The performance of these models is substantially better than any previously-published model when evaluated using this large pooled-data set. We did not observe an independent effect of height on , nor an effect of race, though this may have been due to insufficient numbers of non-white participants. We did observe an effect of FVC such that these models over- or under-predict in persons whose measured FVC was substantially lower or higher than estimated FVC, respectively. Although additional measurements are necessary to confirm this finding regarding FVC, we recommend using measured FVC when possible.

Automated detection of microaneurysms using Stockwell transform and statistical features

Microaneurysms (MAs) are the earliest pre-eminent indicators of diabetic retinopathy (DR) and are hard to distinguish for ophthalmologists on standard fundus images. This study proposes a method based on discrete orthonormal Stockwell transform and statistical features for discriminating between normal and diseased retinal images. Feature extraction by the two different approaches is consolidated and a total of 24 features are extracted for classifier models. Training and testing of the proposed method have been accomplished using 1140 retinal colour photographs. A comparative study using eight best-known classifiers is showcased for detection of MAs and the performance of the classifiers is evaluated using retinal images by performing ten-fold cross-validation procedure. Simulation results demonstrate the efficiency and adequacy of the proposed method which mainly characterise the textural features. The proposed method is compared with existing algorithms and the results show that the algorithm detects DR with high veracity. With the high accuracy and positive prediction, the proposed system assures promising results in early diagnosis of DR.

Mapping EORTC-QLQ-C30 and QLQ-CR29 onto EQ-5D-5L in Colorectal Cancer Patients.

Patient-level utility data are needed for cost-utility analysis; in oncology, however, the data are commonly gathered using disease-specific questionnaires that are often not appropriate. Present study aimed to derive an algorithm which can map the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30 (EORTC QLQ-C30) scales and the Colorectal Cancer-Specific Quality Of Life Questionnaire (QLQ-CR29) scales onto the EuroQoL 5-Dimension 5-Level (EQ-5D-5L) values in patients with colorectal cancer (CRC). Using the Ordinary Least Square (OLS) model, a cross-sectional dataset of 252 patients with CRC were gathered from three academic centers of cancer treatment in Tehran in 2017. The predicted R2 (Pred R2) and adjusted R2 (Adj R2) are used to evaluate model goodness of fit. Additionally, mean absolute error (MAE), root mean square error (RMSE), Spearman's correlation coefficients (ρ), and intraclass correlation (ICC) are applied to assess predictive ability of models. The tenfold cross-validation procedure was applied for validation models. According to the results of our study, the model C4 from EORTC QLQ-C30 was the best predictive model (Pred R2 = 66.57%, Adj R2 = 67.67%, RMSE = 0.10173, MAE = 0.07840). Also, the model R4 from QLQ-CR29 performed the best for EQ-5D-5L (Adj R2 = 48.42%, Pred R2 = 45.54%, MAE = 0.10051, RMSE = 0.12997). The mapping algorithm successfully mapped the EORTC QLQ-C30 and QLQ-CR29 scales onto the EQ-5D-5L values; therefore, it enables policymakers to convert cancer-specific questionnaires scores to the preference-based scores.

Predicting life expectancy with a long short-term memory recurrent neural network using electronic medical records

BackgroundLife expectancy is one of the most important factors in end-of-life decision making. Good prognostication for example helps to determine the course of treatment and helps to anticipate the procurement of health care services and facilities, or more broadly: facilitates Advance Care Planning. Advance Care Planning improves the quality of the final phase of life by stimulating doctors to explore the preferences for end-of-life care with their patients, and people close to the patients. Physicians, however, tend to overestimate life expectancy, and miss the window of opportunity to initiate Advance Care Planning. This research tests the potential of using machine learning and natural language processing techniques for predicting life expectancy from electronic medical records.MethodsWe approached the task of predicting life expectancy as a supervised machine learning task. We trained and tested a long short-term memory recurrent neural network on the medical records of deceased patients. We developed the model with a ten-fold cross-validation procedure, and evaluated its performance on a held-out set of test data. We compared the performance of a model which does not use text features (baseline model) to the performance of a model which uses features extracted from the free texts of the medical records (keyword model), and to doctors’ performance on a similar task as described in scientific literature.ResultsBoth doctors and the baseline model were correct in 20% of the cases, taking a margin of 33% around the actual life expectancy as the target. The keyword model, in comparison, attained an accuracy of 29% with its prognoses. While doctors overestimated life expectancy in 63% of the incorrect prognoses, which harms anticipation to appropriate end-of-life care, the keyword model overestimated life expectancy in only 31% of the incorrect prognoses.ConclusionsPrognostication of life expectancy is difficult for humans. Our research shows that machine learning and natural language processing techniques offer a feasible and promising approach to predicting life expectancy. The research has potential for real-life applications, such as supporting timely recognition of the right moment to start Advance Care Planning.

Deep Learning for an Effective Nonorthogonal Multiple Access Scheme

Nonorthogonal multiple access (NOMA) has been considered as an essential multiple access technique for enhancing system capacity and spectral efficiency in future communication scenarios. However, the existing NOMA systems have a fundamental limit: high computational complexity and a sharply changing wireless channel make exploiting the characteristics of the channel and deriving the ideal allocation methods very difficult tasks. To break this fundamental limit, in this paper, we propose a novel and effective deep learning (DL)-aided NOMA system, in which several NOMA users with random deployment are served by one base station. Since DL is advantageous in that it allows training the input signals and detecting sharply changing channel conditions, we exploit it to address wireless NOMA channels in an end-to-end manner. Specifically, it is employed in the proposed NOMA system to learn a completely unknown channel environment. A long short-term memory (LSTM) network based on DL is incorporated into a typical NOMA system, enabling the proposed scheme to detect the channel characteristics automatically. In the proposed strategy, the LSTM is first trained by simulated data under different channel conditions via offline learning, and then the corresponding output data can be obtained based on the current input data used during the online learning process. In general, we build, train and test the proposed cooperative framework to realize automatic encoding, decoding and channel detection in an additive white Gaussian noise channel. Furthermore, we regard one conventional user activity and data detection scheme as an unknown nonlinear mapping operation and use LSTM to approximate it to evaluate the data detection capacity of DL based on NOMA. Simulation results demonstrate that the proposed scheme is robust and efficient compared with conventional approaches. In addition, the accuracy of the LSTM-aided NOMA scheme is studied by introducing the well-known tenfold cross-validation procedure.

Quantifying Cognitive Workload in Simulated Flight Using Passive, Dry EEG Measurements

A reliable method for quantifying cognitive workload in pilots could find uses in flight training and scheduling, cockpit design, and improving flight safety. Many proposed methods for monitoring cognitive workload in this population rely on measuring physiological responses to externally delivered probe stimuli and/or use traditional gel-based electroencephalography (EEG) sensors. Here we develop passive, probe-independent algorithms for classifying three levels of flight task complexity based on 4-channel, gel-free EEG during simulated flight. Using a library of 168 input features drawn from different data science application domains, we evaluated 13 different classifiers, using a nested tenfold cross-validation procedure to estimate generalization performance. The best subsets of features yielded a median classification accuracy of 90.17% across subjects, with perfect accuracy in one subject and greater than 75% in 16 of 21 subjects. Though EEG line length and linear discriminant analysis were generally among the most effective features and classifiers, respectively, we find that to maximize prediction accuracy, feature set-classifier combinations should be individualized. No single channel proved more valuable than another in predicting flight task complexity, but combining EEG features across channels maintained or improved performance in 81% of subjects.