Search

IntroductionFew artificial intelligence models exist to predict severe forms of COVID-19. Most rely on post-infection laboratory data, hindering early treatment for high-risk individuals.MethodsThis study developed a machine learning model to predict inherent risk of severe symptoms after contracting SARS-CoV-2. Using a Decision Tree trained on 153 Alpha variant patients, demographic, clinical and immunogenetic markers were considered. Model performance was assessed on Alpha and Delta variant datasets. Key risk factors included age, gender, absence of KIR2DS2 gene (alone or with HLA-C C1 group alleles), presence of 14-bp polymorphism in HLA-G gene, presence of KIR2DS5 gene, and presence of KIR telomeric region A/A.ResultsThe model achieved 83.01% accuracy for Alpha variant and 78.57% for Delta variant, with True Positive Rates of 80.82 and 77.78%, and True Negative Rates of 85.00% and 79.17%, respectively. The model showed high sensitivity in identifying individuals at risk.DiscussionThe present study demonstrates the potential of AI algorithms, combined with demographic, epidemiologic, and immunogenetic data, in identifying individuals at high risk of severe COVID-19 and facilitating early treatment. Further studies are required for routine clinical integration.

IntroductionMultiple sclerosis (MS) is a persistent neurological condition impacting the central nervous system (CNS). The precise cause of multiple sclerosis is still uncertain; however, it is thought to arise from a blend of genetic and environmental factors. MS diagnosis includes assessing medical history, conducting neurological exams, performing magnetic resonance imaging (MRI) scans, and analyzing cerebrospinal fluid. While there is currently no cure for MS, numerous treatments exist to address symptoms, decelerate disease progression, and enhance the quality of life for individuals with MS.MethodsThis paper introduces a novel machine learning (ML) algorithm utilizing decision trees to address a key objective: creating a predictive tool for assessing the likelihood of MS development. It achieves this by combining prevalent demographic risk factors, specifically gender, with crucial immunogenetic risk markers, such as the alleles responsible for human leukocyte antigen (HLA) class I molecules and the killer immunoglobulin-like receptors (KIR) genes responsible for natural killer lymphocyte receptors.ResultsThe study included 619 healthy controls and 299 patients affected by MS, all of whom originated from Sardinia. The gender feature has been disregarded due to its substantial bias in influencing the classification outcomes. By solely considering immunogenetic risk markers, the algorithm demonstrates an ability to accurately identify 73.24% of MS patients and 66.07% of individuals without the disease.DiscussionGiven its notable performance, this system has the potential to support clinicians in monitoring the relatives of MS patients and identifying individuals who are at an increased risk of developing the disease.

This paper introduces a disruption predictor constructed through a fully unsupervised two-dimensional mapping of the high-dimensional JET operational space. The primary strength of this disruption predictor lies in its inherent self-organization capability. Diverging from both supervised disruption predictors and earlier approaches suggested by the same authors, which were based on unsupervised models such as Self-Organizing or Generative Topographic Maps, this predictor eliminates the need for labeling data of disruption terminated pulses during training. In prior methods, labels were indeed required post-mapping to inform the model about the presence or absence of disruption precursors at each time instant during the disrupted discharges. In contrast, our approach in this study involves no labeling of data from disruption-terminated experiments. The Self-Organizing Map, operating without any a priori information, adeptly identifies the regions characterizing the pre-disruptive phase. Moreover, SOM discovers non-trivial relationships and captures the complicated interplay of device diagnostics on the internal plasma states from the experimental data. The provided model is highly interpretable; it allows the visualization of high-dimensional data and facilitates easy interrogation of the model to understand the reasons behind its correlations. Hence, utilizing SOMs across various devices can prove invaluable in extracting rules and identifying common patterns, thereby facilitating extrapolation to ITER of the knowledge acquired from existing tokamaks.

Reliable disruption prediction (DP) and disruption mitigation systems are considered unavoidable during international thermonuclear experimental reactor (ITER) operations and in the view of the next fusion reactors such as the DEMOnstration Power Plant (DEMO) and China Fusion Engineering Test Reactor (CFETR). In the last two decades, a great number of DP systems have been developed using data-driven methods. The performance of the DP models has been improved over the years both for a more appropriate choice of diagnostics and input features and for the availability of increasingly powerful data-driven modelling techniques. However, a direct comparison among the proposals has not yet been conducted. Such a comparison is mandatory, at least for the same device, to learn lessons from all these efforts and finally choose the best set of diagnostic signals and the best modelling approach. A first effort towards this goal is made in this paper, where different DP models will be compared using the same performance indices and the same device. In particular, the performance of a conventional Multilayer Perceptron Neural Network (MLP-NN) model is compared with those of two more sophisticated models, based on Generative Topographic Mapping (GTM) and Convolutional Neural Networks (CNN), on the same real time diagnostic signals from several experiments at the JET tokamak. The most common performance indices have been used to compare the different DP models and the results are deeply discussed. The comparison confirms the soundness of all the investigated machine learning approaches and the chosen diagnostics, enables us to highlight the pros and cons of each model, and helps to consciously choose the approach that best matches with the plasma protection needs.

Nowadays, the demand for communication multi-carriers’ channels, where the sub-channels are made mutually independent by using orthogonal frequency division multiplexing (OFDM), is widespread both for wireless and wired communication systems. Even if OFDM is a spectrally efficient modulation scheme, due to the allowed number of subcarriers, high data rate, and good coverage, the transmitted signal can present high peak values in the time domain, due to inverse fast Fourier transform operations. This gives rise to high peak-to-average power ratio (PAPR) with respect to single carrier systems. These peaks can saturate the transmitting amplifiers, modifying the shape of the OFDM symbol and affecting its information content, and they give rise to electromagnetic compatibility issues for the surrounding electric devices. In this paper, a closed form PAPR reduction algorithm is proposed, which belongs to selected mapping (SLM) methods. These methods consist in shifting the phases of the components to minimize the amplitude of the peaks. The determination of the optimal set of phase shifts is a very complex problem; therefore, the SLM approaches proposed in literature all resort to iterative algorithms. Moreover, as this calculation must be performed online, both the computational cost and the effect on the bit rate (BR) cannot be established a priori. The proposed analytic algorithm finds the optimal phase shifts of an approximated formulation of the PAPR. Simulation results outperform unprocessed conventional OFDM transmission by several dBs. Moreover, the complementary cumulative distribution function (CCDF) shows that, in most of the packets, the proposed algorithm reduces the PAPR if compared with randomly selected phase shifts. For example, with a number of shifted phases U=8, the CCFD curves corresponding to the analytical and random methods intersect at a probability value equal to 10−2, which means that in 99% of cases the former method reduces the PAPR more than the latter one. This is also confirmed by the value of the gain, which, at the same number of shifted phases and at the probability value equal to 10−1, changes from 2.09 dB for the analytical to 1.68 dB for the random SLM.

The Nocturnal Frontal Lobe Epilepsy (NFLE) is a form of epilepsy in which seizures occur predominantly during sleep. In other forms of epilepsy, the commonly used clinical approach mainly involves manual inspection of encephalography (EEG) signals, a laborious and time-consuming process which often requires the contribution of more than one experienced neurologist. In the last decades, numerous approaches to automate this detection have been proposed and, more recently, machine learning has shown very promising performance. In this paper, an original Convolutional Neural Network (CNN) architecture is proposed to develop patient-specific seizure detection models for three patients affected by NFLE. The performances, in terms of accuracy, sensitivity, and specificity, exceed by several percentage points those in the most recent literature. The capability of the patient-specific models has been also tested to compare the obtained seizure onset times with those provided by the neurologists, with encouraging results. Moreover, the same CNN architecture has been used to develop a cross-patient seizure detection system, resorting to the transfer-learning paradigm. Starting from a patient-specific model, few data from a new patient are enough to customize his model. This contribution aims to alleviate the task of neurologists, who may have a robust indication to corroborate their clinical conclusions.

In tokamak nuclear fusion reactors, one of the main issues is to know the total emission of radiation, which is mandatory to understand the plasma physics and is very useful to monitor and control the plasma evolution. This radiation can be measured by means of a bolometer system that consists in a certain number of elements sensitive to the integral of the radiation along straight lines crossing the plasma. By placing the sensors in such a way to have families of crossing lines, sophisticated tomographic inversion algorithms allow to reconstruct the radiation tomography in the 2D poloidal cross-section of the plasma. In tokamaks, the number of projection cameras is often quite limited resulting in an inversion mathematic problem very ill conditioned so that, usually, it is solved by means of a grid-based, iterative constrained optimization procedure, whose convergence time is not suitable for the real time requirements. In this paper, to illustrate the method, an assumption not valid in general is made on the correlation among the grid elements, based on the statistical distribution of the radiation emissivity over a set of tomographic reconstructions, performed off-line. Then, a regularization procedure is carried out, which merge highly correlated grid elements providing a squared coefficients matrix with an enough low condition number. This matrix, which is inverted offline once for all, can be multiplied by the actual bolometer measures returning the tomographic reconstruction, with calculations suitable for real time application. The proposed algorithm is applied, in this paper, to a synthetic case study.

Non-intrusive load monitoring (NILM) is a process of determining the operating states and the energy consumption of single electric devices using a single energy meter providing aggregate load measurements. Due to the large spread of power electronic-based and nonlinear devices connected to the network, the time signals of both voltage and current are typically non-sinusoidal. The effectiveness of a NILM algorithm strongly depends on determining a set of discriminative features. In this paper, voltage and current signals were combined to define, according to the definitions provided in Standard IEEE 1459, different power quantities, that can be used to distinguish different types of appliance. Multi-layer perceptron (MLP) classifiers were trained to solve the appliance detection problem as a multi-class event classification problem, varying the electric features in input. This allowed to select an optimal set of features guarantying good classification performance in identifying typical electric loads.

In this paper, a procedure is proposed to design a power line communication (PLC) system to perform the digital transmission in a distributed energy storage system consisting of fleets of electric cars. PLC uses existing power cables or wires as data communication multicarrier channels. For each vehicle, the information to be transmitted can be, for example: the models of the batteries, the level of the charge state, and the schedule of charging/discharging. Orthogonal frequency division multiplexing modulation (OFDM) is used for the bit loading, whose parameters are optimized to find the best compromise between the communication conflicting objectives of minimizing the signal power, maximizing the bit rate, and minimizing the bit error rate. The off-line design is modeled as a multi-objective optimization problem, whose solution supplies a set of Pareto optimal solutions. At the same time, as many charging stations share part of the transmission line, the optimization problem includes also the assignment of the sub-carriers to the single charging stations. Each connection between the control node and a charging station has its own frequency response and is affected by a noise spectrum. In this paper, a procedure is presented, called Chimera, which allows one to solve the multi-objective optimization problem with respect to a unique frequency response, representing the whole set of lines connecting each charging station with the central node. Among the provided Pareto solutions, the designer will make the final decision based on the control system requirements and/or the hardware constraints.