Statistical Machine Research Articles

Women and ethnoracial minorities with poor cardiovascular health measures associated with a higher risk of developing mood disorder

BackgroundMood disorders (MDS) are a type of mental health illness that effects millions of people in the United States. Early prediction of MDS can give providers greater opportunity to treat these disorders. We hypothesized that longitudinal cardiovascular health (CVH) measurements would be informative for MDS prediction.MethodsTo test this hypothesis, the American Heart Association’s Guideline Advantage (TGA) dataset was used, which contained longitudinal EHR from 70 outpatient clinics. The statistical analysis and machine learning models were employed to identify the associations of the MDS and the longitudinal CVH metrics and other confounding factors.ResultsPatients diagnosed with MDS consistently had a higher proportion of poor CVH compared to patients without MDS, with the largest difference between groups for Body mass index (BMI) and Smoking. Race and gender were associated with status of CVH metrics. Approximate 46% female patients with MDS had a poor hemoglobin A1C compared to 44% of those without MDS; 62% of those with MDS had poor BMI compared to 47% of those without MDS; 59% of those with MDS had poor blood pressure (BP) compared to 43% of those without MDS; and 43% of those with MDS were current smokers compared to 17% of those without MDS.ConclusionsWomen and ethnoracial minorities with poor cardiovascular health measures were associated with a higher risk of development of MDS, which indicated the high utility for using routine medical records data collected in care to improve detection and treatment for MDS among patients with poor CVH.

Scaling up a Hybrid MT System: From low to full resources

This article describes a hybrid approach to machine translation (MT) that is inspired by the rule-based, statistical, example-based, and other hybrid machine translation approaches currently used or described in academic literature. It describes how the approach was implemented for language pairs using only limited monolingual resources and hardly any parallel resources (the METIS-II system), and how it is currently implemented with rich resources on both the source and target side as well as rich parallel data (the PaCo-MT system). We aim to illustrate that a similar paradigm can be used, irrespectively of the resources available, but of course with an impact on translation quality.

Online Nonlinear Error Compensation Circuit Based on Neural Networks

Nonlinear errors of sensor output signals are common in the field of inertial measurement and can be compensated with statistical models or machine learning models. Machine learning solutions with large computational complexity are generally offline or implemented on additional hardware platforms, which are difficult to meet the high integration requirements of microelectromechanical system inertial sensors. This paper explored the feasibility of an online compensation scheme based on neural networks. In the designed solution, a simplified small-scale network is used for modeling, and the peak-to-peak value and standard deviation of the error after compensation are reduced to 17.00% and 16.95%, respectively. Additionally, a compensation circuit is designed based on the simplified modeling scheme. The results show that the circuit compensation effect is consistent with the results of the algorithm experiment. Under SMIC 180 nm complementary metal-oxide semiconductor (CMOS) technology, the circuit has a maximum operating frequency of 96 MHz and an area of 0.19 mm2. When the sampling signal frequency is 800 kHz, the power consumption is only 1.12 mW. This circuit can be used as a component of the measurement and control system on chip (SoC), which meets real-time application scenarios with low power consumption requirements.

Information criteria and cross validation for Bayesian inference in regular and singular cases

In data science, an unknown information source is estimated by a predictive distribution defined from a statistical model and a prior. In an older Bayesian framework, it was explained that the Bayesian predictive distribution should be the best on the assumption that a statistical model is convinced to be correct and a prior is given by a subjective belief in a small world. However, such a restricted treatment of Bayesian inference cannot be applied to highly complicated statistical models and learning machines in a large world. In 1980, a new scientific paradigm of Bayesian inference was proposed by Akaike, in which both a model and a prior are candidate systems and they had better be designed by mathematical procedures so that the predictive distribution is the better approximation of unknown information source. Nowadays, Akaike’s proposal is widely accepted in statistics, data science, and machine learning. In this paper, in order to establish a mathematical foundation for developing a measure of a statistical model and a prior, we show the relation among the generalization loss, the information criteria, and the cross-validation loss, then compare them from three different points of view. First, their performances are compared in singular problems where the posterior distribution is far from any normal distribution. Second, they are studied in the case when a leverage sample point is contained in data. And last, their stochastic properties are clarified when they are used for the prior optimization problem. The mathematical and experimental comparison shows the equivalence and the difference among them, which we expect useful in practical applications.

P143 The effect of JAK inhibition on the metabolome of neutrophils from patients with rheumatoid arthritis

Abstract Background/Aims Janus kinase inhibitors (JAKi) including baricitinib and tofacitinib target the JAK/STAT pathway and are clinically effective in treating rheumatoid arthritis (RA). Neutrophils play a key role in the pathophysiology of RA and their function is directly affected by these drugs as shown by clinical trials of JAKi which report increased infection rates and transient neutropenia during therapy. The aim of this work is to determine difference in the metabolome of neutrophils from healthy controls and patients with RA, and the effect of JAKi on the metabolome of healthy and RA neutrophils. Methods Neutrophils were isolated from healthy controls (HC) (n = 10) and patients with RA (n = 10 DMARD-naïve, n = 10 Biologics-naive) and pre-incubated with baricitinib, tofacitinib or a pan-JAK inhibitor (all 200ng/mL) for 2h before snap-freezing. Metabolites were extracted by 50:50 v/v AcN:H2O and analysed by 1H NMR spectroscopy using a 700 MHz Avance IIIHD Bruker NMR spectrometer equipped with a TCI cryoprobe. Chenomx, Bruker TopSpin and tameNMR software were used to identify metabolites and process spectra. Statistical analysis and machine learning models were implemented using R and the Mixomics package. Pathway analysis was carried out using Ingenuity (IPA). ROS production in response to 15 min fMLP stimulation was measured using DHR-123 and flow cytometry. Results At 0h, changes in levels of NADP+, NADPH and hypoxanthine in RA neutrophils indicated increased production of ROS via NADPH oxidase and xanthine oxidase compared to HC. Lower levels of glucose and increased levels of lactic acid also suggested increased glycolysis in RA neutrophils compared to HC. However after 2h incubation with JAKi, a difference in the response to these drugs was observed. 80 metabolites were higher in RA neutrophils compared to HC following 2h JAKi treatment. IPA analysis revealed the metabolites higher in RA neutrophils were involved in NAD biosynthesis, NAD phosphorylation/dephosphorylation, amino acid synthesis and transport, and carbohydrate metabolism. Changes in the levels of NADP+, NADPH and hypoxanthine in RA neutrophils indicate that JAKi inhibit ROS production. We therefore confirmed experimentally that JAKi inhibit ROS production in response to fMLP (p < 0.05). IPA also predicted altered neutrophil cell functions including cell death and cell-to-cell signalling. We previously showed that JAKi did not affect the constitutive rate of neutrophil apoptosis, however JAKi significantly inhibit the delay in apoptosis induced by GM-CSF (p < 0.05). No significant difference in the metabolome was observed when comparing tofacinib-treatment and baricitinib-treatment in either the HC or RA patient group. Conclusion The metabolome of RA and HC neutrophils is different at baseline. JAKi alter the levels of intracellular metabolites involved in neutrophil activation and ROS production in RA. This may explain, in part, the efficacy of JAKi in decreasing disease activity and joint damage in RA. Disclosure M. Fresneda Alarcon: None. R. Grosman: None. S. Chokesuwattanaskul: None. M. Phelan: None. H. Wright: None.

Interpretable machine learning approach for predicting COVID-19 risk status of an individual

This study aimed to ascertained using Statistical feature selection methods and interpretable Machine learning models, the best features that predict risk status (“Low”, “Medium”, “High”) to COVID-19 infection. This study utilizes a publicly available dataset obtained via; online web-based risk assessment calculator to ascertain the risk status of COVID-19 infection. 57 out of 59 features were first filtered for multicollinearity using the Pearson correlation coefficient and further shrunk to 55 features by the LASSO GLM approach. SMOTE resampling technique was used to incur the problem of imbalanced class distribution. The interpretable ML algorithms were implored during the classification phase. The best classifier predictions were saved as a new instance and perturbed using a single Decision tree classifier. To further build trust and explainability of the best model, the XGBoost classifier was used as a global surrogate model to train predictions of the best model. The XGBoost individual’s explanation was done using the SHAP explainable AI-framework. Random Forest classifier with a validation accuracy score of 96.35 % from 55 features reduced by feature selection emerged as the best classifier model. The decision tree classifier approximated the best classifier correctly with a prediction accuracy score of 92.23 % and Matthew’s correlation coefficient of 0.8960. The XGBoost classifier approximated the best classifier model with a prediction score of 99.7 %. This study identified COVID-19 positive, COVID-19 contacts, COVID-19 symptoms, Health workers, and Public transport count as the five most consistent features that predict an individual’s risk exposure to COVID-19.

Data analytics using statistical methods and machine learning: a case study of power transfer units

Sensors can produce large amounts of data related to products, design, and materials; however, it is important to use the right data for the right purposes. Therefore, detailed analysis of data accumulated from different sensors in production and assembly manufacturing lines is necessary to minimize faulty products and understand the production process. Additionally, when selecting analytical methods, manufacturing companies must select the most suitable techniques. This paper presents a data analytics approach to extract useful information, such as important measurements for the dimensions of a shim, a small part for aligning shafts, from the manufacturing data of a power transfer unit (PTU). This paper also identifies the best techniques and analytical approaches within the following six individual areas: (1) identifying measurements associated with faults; (2) identifying measurements associated with shim dimensions; (3) identifying associations between station codes; (4) predicting shim dimensions; (5) identifying duplicate samples in faulty data; and (6) identifying error distributions associated with measurement. These areas are analysed in accordance with two analytical approaches: (a) statistical analysis and (b) machine learning (ML)-based analysis. The results show (a) the relative importance of measurements with regard to the faulty unit and shim dimensions, (b) the error distribution of measurements, and (c) the reproduction rate of faulty units. Additionally, both statistical analysis and ML-based analysis have shown that the measurement ‘PTU housing measurement’ is the most important measurement among available shim dimensions. Additionally, certain faulty stations correlated with one another. ML is shown to be the most suitable technique in three areas (e.g. identifying measurements associated with faults), while statistical analysis is sufficient for the other three areas (e.g. identifying measurements associated with shim dimensions) because they do not require a complex analytical model. This study provides a clearer understanding of assembly line production and identifies highly correlated and significant measurements of a faulty unit.

Predicting Energy Demand in Semi-Remote Arctic Locations

Forecasting energy demand within a distribution network is essential for developing strategies to manage and optimize available energy resources and the associated infrastructure. In this study, we consider remote communities in the Arctic located at the end of the radial distribution network without alternative energy supply. Therefore, it is crucial to develop an accurate forecasting model to manage and optimize the limited energy resources available. We first compare the accuracy of several models that perform short-and medium-term load forecasts in rural areas, where a single industrial customer dominates the electricity consumption. We consider both statistical methods and machine learning models to predict energy demand. Then, we evaluate the transferability of each method to a geographical rural area different from the one considered for training. Our results indicate that statistical models achieve higher accuracy on longer forecast horizons relative to neural networks, while the machine-learning approaches perform better in predicting load at shorter time intervals. The machine learning models also exhibit good transferability, as they manage to predict well the load at new locations that were not accounted for during training. Our work will serve as a guide for selecting the appropriate prediction model and apply it to perform energy load forecasting in rural areas and in locations where historical consumption data may be limited or even not available.

No-Reference Stereoscopic Image Quality Assessment Based on Binocular Statistical Features and Machine Learning

Learning a deep structure representation for complex information networks is a vital research area, and assessing the quality of stereoscopic images or videos is challenging due to complex 3D quality factors. In this paper, we explore how to extract effective features to enhance the prediction accuracy of perceptual quality assessment. Inspired by the structure representation of the human visual system and the machine learning technique, we propose a no-reference quality assessment scheme for stereoscopic images. More specifically, the statistical features of the gradient magnitude and Laplacian of Gaussian responses are extracted to form binocular quality-predictive features. After feature extraction, these features of distorted stereoscopic image and its human perceptual score are used to construct a statistical regression model with the machine learning technique. Experimental results on the benchmark databases show that the proposed model generates image quality prediction well correlated with the human visual perception and delivers highly competitive performance with the typical and representative methods. The proposed scheme can be further applied to the real-world applications on video broadcasting and 3D multimedia industry.

Exploring musical style in the anonymous and doubtfully attributed mass movements of the Coimbra manuscripts: a statistical and machine learning approach

This paper studies the music from the sixteenth century Coimbra manuscripts using both traditional musicological approaches and techniques based on statistical analysis and machine learning. A particular focus is placed on gaining insights into the origins of the anonymous and doubtfully attributed mass movements, looked at through the lens of potential stylistic differences between the Iberian and Franco-Flemish musical traditions of the time. Another goal is to explore the origins of the Coimbra works in the historical, political and social context of the reception of foreign repertoires.

An Evaluation of Some Machine Learning Algorithms as Tools for Predicting Soil Characteristics Based on Their Spectral Response in the Vis‑NIR Range

Using the Land Use and Coverage Frame Survey (LUCAS) database of European soil surface layer properties, statistical and machine learning predictive models for several key soil characteristics (clay content, pH in CaCl2, concentration of organic carbon, calcium carbonates and nitrogen and exchange cations capacity) were compared on the basis of processing their spectral responses in the visible (Vis) and near‑infrared (NIR) parts. Standard methods of relationship modeling were used: stepwise regression, partial least squares regression and linear regression with input data obtained from principal components analysis. Using the inputs extracted by statistical algorithms various machine learning algorithms were used in the modeling. The usefulness of the models was analyzed by comparison with the values of the determination coefficients, the root mean square error and the distribution of residual values. The mean square error of estimation in the cross‑validation procedure for the stack model using the multilayer perceptron and the distributed random forest were as follows: for clay content – ca. 4.5%; for pH – ca. 0.35; for SOC – ca. 7.5 g/kg (0.75% by weight); for CaCO3 content – ca. 19 g/kg; for N content – ca. 0.50 g/kg; and for CEC – ca. 3.5 cmol(+)/kg.

A Rough Set Based Approach to Compute The Impact of Non-Academic Parameters on Academic Performance

Background: Our academic performance can differ due various to non academic factors namely, self motivation and mind stability where peace of mind plays a pivotal role; self confidence after any kind of set back, environment and financial pressure, family support, and capability of an individual to make a combat. These nonacademic factors may lead to stress affecting our academic performance. Objective: Inthispaper,weaimtoassess the academicfactorsaffect our academic performance. Methods: The paper is modeled on survey performed on college students and depending upon their responses related to these non-academicfactors. We have carried out the work through statistical analysis and machine learning algorithms. Density bases Clustering algorithm is being used to cluster similar kind of performance and analyze how the factors differ from each other. Moreover we also presented a comparative study between academic parameters such as attendance, interest in subject and travelling time in comparison to such non academic performances. A rough set is also being created comprising the parameterswhichgivesthebestresultamongthestudents. Results: Proposed algorithm based on RST outplayed the accuracy of current models which is existing, there is a sharp drop in RMSE value for the parameters chosen in reduct dataset rather than whole dataset. Conclusion: Our proposed model’s accuracy overshadows the current analyzing model.

Multimodal affect analysis of psychodynamic play therapy

Objective: We explore state of the art machine learning based tools for automatic facial and linguistic affect analysis to allow easier, faster, and more precise quantification and annotation of children’s verbal and non-verbal affective expressions in psychodynamic child psychotherapy. Method: The sample included 53 Turkish children: 41 with internalizing, externalizing and comorbid problems; 12 in the non-clinical range. We collected audio and video recordings of 148 sessions, which were manually transcribed. Independent raters coded children’s expressions of pleasure, anger, sadness and anxiety using the Children’s Play Therapy Instrument (CPTI). Automatic facial and linguistic affect analysis modalities were adapted, developed, and combined in a system that predicts affect. Statistical regression methods (linear and polynomial regression) and machine learning techniques (deep learning, support vector regression and extreme learning machine) were used for predicting CPTI affect dimensions. Results: Experimental results show significant associations between automated affect predictions and CPTI affect dimensions with small to medium effect sizes. Fusion of facial and linguistic features work best for pleasure predictions; however, for other affect predictions linguistic analyses outperform facial analyses. External validity analyses partially support anger and pleasure predictions. Discussion: The system enables retrieving affective expressions of children, but needs improvement for precision.

Fast secure matrix multiplications over ring-based homomorphic encryption

ABSTRACT As widespread development of biometrics, concerns about security and privacy are rapidly increasing. Secure matrix computation is one of the most fundamental and useful operations for statistical analysis and machine learning with protecting the confidentiality of input data. Secure computation can be achieved by homomorphic encryption, supporting meaningful operations over encrypted data. HElib is a software library that implements the Brakerski-Gentry-Vaikuntanathan (BGV) homomorphic scheme, in which secure matrix-vector multiplication is proposed for operating matrices. Recently, Duong et al. (Tatra Mt. Publ) proposed a new method for secure single matrix multiplication over a ring-LWE-based scheme. In this paper, we generalize Duong et al.’s method for secure multiple matrix multiplications over the BGV scheme. We also implement our method using HElib, and show that our method is much faster than the matrix-vector multiplication in HElib for secure matrix multiplications.

Intelligent socio-economic status prediction system using machine learning models on Rajahmundry A.P., SES dataset

Background: Developing economic and social systems and assuring the efficiency of economic and social processes is the major task for the government of any country. Predictable machine learning (ML) models are used for analyzing data sets that allow more efficient enterprise management. Now a day, the research on Socio-Economic Status (SES) and Machine Learning (ML) is very crucial to find socio-economic inequalities, and take further actions that are preventions, protections, and suppressions. Objectives: The mainobjective of this research is to understand the Socio Economic System issues and predicting SES levels on particular area like Rajahmundry, AP, India using statistical analysis and machine learning methodologies. Methods: In this, we analyze the data that is collected from Rajahmundry (Rajamahandravaram),Andhra Pradesh, India with 48 feature attributes (dimensions), and one target four class attribute (poor, rich, middle, upper-middle ). The SES levels like poor, rich, middle, and upper-middle classes are predicted by 5 ML algorithms. Findings: In this paper, we conduct the statistical analysis of each attribute, and analyze and compare the performance accuracies using confusion matrix, performance parameter (classification accuracy, Precision,Recall, and F1) values and receive operating characteristic (ROC) under AUC values of five efficient ML algorithms like Naïve Bayes, Decision Trees (DTs), k-NN, SVM (kernel RBF) and Random Forest (RF). We observed that the RF algorithm showed better results when compared with other algorithms for the Rajahmundry AP SES dataset. The RF algorithm performs 97.82% of classification accuracy (CA) and time is taken for model construction 0.41 seconds. The next superior performed ML model is DTs with 96.67% of CA and 0.16 seconds for model construction. Novelty: Comprehensive analysis indicates that the novel AP SES Dataset with empirical statistical analysis gives the good results and predicts the SES levels with RF model is very effective. Keywords: Machine Learning; socio-economic status; Rajahmundry;household; poverty

Efficiency Enhancement for Underwater Adaptive Modulation and Coding Systems: Via Sparse Principal Component Analysis

In this letter, to explore key channel state information (CSI) as a more efficient switching metric in the task of underwater adaptive modulation and coding (AMC), a sparse principal component analysis (SPCA) based approach is proposed from the perspective of statistical analysis plus machine learning (ML). This data-driven sparse learning method can offer significant system efficiency enhancement in the procedures of both channel estimation and communication scheme switching. By leveraging a dataset that contains real-world channel measurements collected from three field experiments, simulations demonstrate the effectiveness of the proposed scheme.

Learning from Data to Optimize Control in Precision Farming

Precision farming is one way of many to meet a 55 percent increase in global demand for agricultural products on current agricultural land by 2050 at reduced need of fertilizers and efficient use of water resources. The catalyst for the emergence of precision farming has been satellite positioning and navigation followed by Internet-of-Things, generating vast information that can be used to optimize farming processes in real-time. Statistical tools from data mining, predictive modeling, and machine learning analyze patterns in historical data, to make predictions about future events as well as intelligent actions. This special issue presents the latest development in statistical inference, machine learning, and optimum control for precision farming.

MLsurvival: an automated machine learning tool for survival analysis of cancer patients.

e13555 Background: Survival analysis is used to establish a connection between covariates and the time of event with censored data. Compared with traditional statistical methods, machine learning approaches based on sophisticated and effective computational algorithms are more capable for handling complex multi-dimensional medical data. Methods: We developed an automated machine learning tool MLsurvival to analyze survival data of cancer patients, algorithms of which include the statistical cox regression and machine learning based on linear model (elastic net), ensemble model (gradient boosting with least squares or regression trees and random forest) and support vector kernel (linear and non-linear). The workflow of MLsurvival is comprised with four modules: preprocessing (missing data remove or imputation and feature standardization), feature selection (unsupervised multi-statistics and supervised machine recursive feature elimination with cross-validation), modeling (hyperparameter and performance evaluation) and prediction. To evaluate the performance of this tool, we analyzed medical data for 222 hepatocellular carcinoma (HCC) patients at stage II-III who underwent surgical resection and developed five machine learning approach based estimation models for overall survival (OS). Models were trained on 155 patients with 300 features, including clinical information, somatic mutation and copy number variation, and independently validated on the rest 67 patients. Results: The ensemble model of gradient boosting fitted by MLsurvival using 48 selected features for the data of 155 HCC patients possessed the highest mean AUC and C-Index value. For 67 patients in validation set, this model predicted half year mortality of patients with an AUC of 0.9 (95% CI, 0.771-1.029) and one year mortality with an AUC of 0.897 (95% CI, 0.816-0.978). In addition to that, this model was also predictive for the time of recurrence (pvalue < 0.0001). Furthermore, we also utilized this tool in survival analysis for extensive real data from patients with breast, lung, and esophagus cancers, while most of results showed superior accuracy and stable performance. Conclusions: MLsurvial is an automate tool for survival analysis of cancer patients with well performance. The risk scoring system implemented in this tool offers a novel strategy for incorporating multi-dimensional risk factors to predict clinical outcome, contributes to the better understanding of disease background and helps to optimize the clinical follow-up and therapeutic treatment for cancer patients.

Designing bioinspired brick-and-mortar composites using machine learning and statistical learning

The brick-and-mortar structure inspired by nature, such as in nacre, is considered one of the most optimal designs for structural composites. Given the large number of design possibilities, extensive computational work is required to guide their manufacturing. Here, we propose a computational framework that combines statistical analysis and machine learning with finite element analysis to establish structure–property design strategies for brick-and-mortar composites. Approximately 20,000 models with different geometrical designs were categorized into good and bad based on their failure modes, with statistical analysis of the results used to find the importance of each feature. Aspect ratio of the bricks and horizontal mortar thickness were identified as the main influencing features. A decision tree machine learning model was then established to draw the boundaries of good design space. This approach might be used for the design of brick-and-mortar composites with improved mechanical properties.

Pitfalls in Article Submissions for Publication.

The aim of every academician and clinical dermatologist is to publish their research in reputed biomedical journals. But from conceptualization to completion, myriad shortcomings creep into the article and by the time it is ready for publication, by default and certainly not by design, the article discourse gets flawed, sometime fatally so. The endeavor of this article is to discuss these pitfalls from conceptualization, statistical machinations, authorial misconcepts, article structuring, and final journal selection. The article can function as a prophylactic checklist, albeit not comprehensive, by any prospective author and is an appreciation of the most oft repeated fallacies usually detected in publication submissions.