Traditional Statistics Research Articles

Harmonization for Parkinson’s Disease Multi-Dataset T1 MRI Morphometry Classification

Classification of disease and healthy volunteer cohorts provides a useful clinical alternative to traditional group statistics due to individualized, personalized predictions. Classifiers for neurodegenerative disease can be trained on structural MRI morphometry, but require large multi-scanner datasets, introducing confounding batch effects. We test ComBat, a common harmonization model, in an example application to classify subjects with Parkinson’s disease from healthy volunteers and identify common pitfalls, including data leakage. We used a multi-dataset cohort of 372 subjects (216 with Parkinson’s disease, 156 healthy volunteers) from 11 identified scanners. We extracted both FreeSurfer and the determinant of Jacobian morphometry to compare single-scanner and multi-scanner classification pipelines. We confirm the presence of batch effects by running single scanner classifiers which could achieve wildly divergent AUCs on scanner-specific datasets (mean:0.651 ± 0.144). Multi-scanner classifiers that considered neurobiological batch effects between sites could easily achieve a test AUC of 0.902, though pipelines that prevented data leakage could only achieve a test AUC of 0.550. We conclude that batch effects remain a major issue for classification problems, such that even impressive single-scanner classifiers are unlikely to generalize to multiple scanners, and that solving for batch effects in a classifier problem must avoid circularity and reporting overly optimistic results.

Sexually dimorphic differences in angiogenesis markers are associated with brain aging trajectories in humans.

Aberrant angiogenesis could contribute to the development of cognitive impairment and represent a therapeutic target for preventing dementia. However, most studies addressing angiogenesis and cognitive impairment focus on model organisms. To test the relevance of angiogenesis to human cognitive aging, we evaluated associations of circulating blood markers of angiogenesis with brain aging trajectories in a pooled two-center sample from deeply phenotyped longitudinal human cohorts (n = 435; female = 207, age = 74 ± 9) using cognitive assessments, biospecimens, structural brain imaging, and clinical data. Blood markers included ligands involved in angiogenesis and vascular function such as basic fibroblast growth factor (bFGF), members of the vascular endothelial growth factor family (VEGFA, VEGFB, and VEGFC), and placental growth factor (PlGF), in addition to their receptors VEGF receptor 1 (VEGFR1) and tyrosine kinase with immunoglobulin and EGF homology domain 2 (Tie2). Machine learning and traditional statistics revealed sexually dimorphic associations of plasma angiogenic growth factors with brain aging outcomes, including executive function and gray matter atrophy. Specifically, markers of angiogenesis were associated with higher executive function and less brain atrophy in younger women (not men), a directionality of association that reversed around age 75. Higher concentrations of bFGF, known for pleiotropic effects on multiple cell types, predicted favorable cognitive trajectories in both women and men. An independent sample from a multicenter dataset (MarkVCID; n = 80; female = 30, age = 73 ± 9) was used to externally validate these findings. In conclusion, this analysis demonstrates the association of angiogenesis to human brain aging, with potential therapeutic implications for vascular cognitive impairment and dementia.

Abstract 4113047: Discovering novel biomarkers and predicting cardiovascular disease using AI/ML techniques for precision medicine

Personalized interventions are deemed vital given the intricate characteristics, advancement, inherent genetic composition, and diversity of cardiovascular diseases (CVDs). The appropriate utilization of artificial intelligence (AI) and machine learning (ML) methodologies can yield novel understandings of CVDs, enabling improved personalized treatments through predictive analysis and deep phenotyping. In this study, we proposed and employed a novel approach combining traditional statistics and a nexus of cutting-edge AI/ML techniques to identify significant biomarkers for our predictive engine by analyzing the complete transcriptome of CVD patients (Figure 1). After robust gene expression data pre-processing, we utilized three statistical tests to assess the differences in transcriptomic expression and clinical characteristics between healthy individuals and CVD patients. Next, a classifier assigned rankings to transcriptomic features based on their relation to the case-control variable. The top ten percent of commonly observed significant biomarkers were evaluated using four unique ML classifiers. After optimizing hyperparameters, the ensembled models, which were implemented using a soft voting classifier, accurately differentiated between patients and healthy individuals. We have uncovered 18 transcriptomic biomarkers that are highly significant in the CVD population that were used to predict disease with up to 96% accuracy (Figure 2). Additionally, we cross-validated our results with clinical records collected from patients in our cohort (Figure 3). The identified biomarkers served as potential indicators for early detection of CVDs. With its successful implementation, our newly developed predictive engine provides a valuable framework for identifying patients with CVDs based on their biomarker profiles.

A Course-Based Undergraduate Research Experience (CURE) in Statistics: Enhancing Educational Outcomes and Promoting Access to Data-Driven Careers

Course-based undergraduate research experiences (CUREs) have been shown to help students gain research skills. Few studies, however, have evaluated the impact of CUREs on long-term educational and employment outcomes. This article examines postgraduation outcomes for students enrolling in introductory statistics delivered as a CURE compared with those taking a traditional mathematics-based introductory statistics course. Participating in the CURE was found to be associated with a higher likelihood of holding a job in which a primary responsibility was working with data, greater confidence in working with data, and a higher likelihood of earning more than $100K annually. This study suggests that CUREs may not only impart technical skills but also boost students’ self-efficacy in applying statistical methods and analyzing data.

Artificial intelligence in predicting postoperative heterotopic ossification following anterior cervical disc replacement.

This study aimed to develop and validate a machine learning (ML) model to predict high-grade heterotopic ossification (HO) following Anterior cervical disc replacement (ACDR). Retrospective review of prospectively collected data of patients undergoing ACDR or hybrid surgery (HS) at a quaternary referral medical center was performed. Patients diagnosed as C3-7 single- or multi-level cervical disc degeneration disease with > 2 years of follow-up and complete pre- and postoperative radiological imaging were included. An ML-based algorithm was developed to predict high grade HO based on perioperative demographic, clinical, and radiographic parameters. Furthermore, model performance was evaluated according to discrimination and overall performance. In total, 339 ACDR segments were included (61.65% female, mean age 45.65 ± 8.03 years). Over 45.65 ± 8.03 months of follow-up, 48 (14.16%) segments developed high grade HO. The model demonstrated good discrimination and overall performance according to precision (High grade HO: 0.71 ± 0.01, none-low grade HO: 0.85 ± 0.02), recall (High grade HO: 0.68 ± 0.03, none-low grade HO: 0.87 ± 0.01), F1-score (High grade HO: 0.69 ± 0.02, none-low grade HO: 0.86 ± 0.01), and AUC (0.78 ± 0.08), with lower prosthesis‑endplate depth ratio, higher height change, male, and lower postoperative-shell ROM identified as the most important predictive features. Through an ML approach, the model identified risk factors and predicted development of high grade HO following ACDR with good discrimination and overall performance. By addressing the shortcomings of traditional statistics and adopting a new logical approach, ML techniques can support discovery, clinical decision-making, and intraoperative techniques better.

Validation of the Internet entrepreneurial self-efficacy scale among Romanian technical students.

This study's aim is to evaluate the Romanian version of the Internet Entrepreneurial Self-Efficacy Scale (IESES), a new assessment scale for online entrepreneurial self-efficacy, and estimate its psychometric properties in a sample of engineering students (N = 644; 317 females). The scale was translated and adapted into Romanian through a forward-backward method. The factorial structure was investigated using Confirmatory Factor Analysis (CFA) and multigroup-CFA for gender invariance. The scale was associated with instruments that measure entrepreneurial intention both traditionally and in the online environment (Individual Entrepreneurial Internet Scale and Entrepreneurial Intention Questionnaire) and components from the Theory of Planned Behavior (attitudes, subjective norms and perceived behavioral control). McDonald's omega and Cronbach's alpha coefficients were used to assess reliability. Traditional statistics are complemented by network analysis. Results show that adequate matching items for the 3-factor model and complete gender invariance is maintained. The three factors are associated with the internet and traditional entrepreneurial intention, and the components from the theory of planned behavior (attitudes, subjective norms, and perceived behavioral control). Internal consistency is excellent because α and ω values range between 0.834 and 0.919. In addition, discriminant validity was demonstrated. The network analysis suggests the relevance of technology utilization in the self-efficacy of online businesses in the case of engineering students. Overall, findings enable us to conclude that the Romanian version of the IESES is a valid, accurate instrument that can be implemented to evaluate self-efficacy related to the successful execution of an online business.

Theoretical wavelet ℓ1-norm from one-point probability density function prediction

Context. Weak gravitational lensing, which results from the bending of light by matter along the line of sight, is a potent tool for exploring large-scale structures, particularly in quantifying non-Gaussianities. It is a pivotal objective for upcoming surveys. In the realm of current and forthcoming full-sky weak-lensing surveys, convergence maps, which represent a line-of-sight integration of the matter density field up to the source redshift, facilitate field-level inference. This provides an advantageous avenue for cosmological exploration. Traditional two-point statistics fall short of capturing non-Gaussianities, necessitating the use of higher-order statistics to extract this crucial information. Among the various available higher-order statistics, the wavelet ℓ1 -norm has proven its efficiency in inferring cosmology. However, the lack of a robust theoretical framework mandates reliance on simulations, which demand substantial resources and time. Aims. Our novel approach introduces a theoretical prediction of the wavelet ℓ1-norm for weak-lensing convergence maps that is grounded in the principles of large-deviation theory. This method builds upon recent work and offers a theoretical prescription for an aperture mass one-point probability density function. Methods. We present for the first time a theoretical prediction of the wavelet ℓ1-norm for convergence maps that is derived from the theoretical prediction of their one-point probability distribution. Additionally, we explored the cosmological dependence of this prediction and validated the results on simulations. Results. A comparison of our predicted wavelet ℓ1 -norm with simulations demonstrates a high level of accuracy in the weakly nonlinear regime. Moreover, we show its ability to capture cosmological dependence. This paves the way for a more robust and efficient parameter-inference process.

Evaluating Higher Education Performance via Machine Learning During Disruptive Times: A Case of Applied Education in Türkiye

ABSTRACTIn response to the COVID‐19 pandemic, an abrupt wave of digitisation and online migration swept the higher education institutions around the globe. In the aftermath of this digital transformation which endures as the legacy of the pandemic, what lacks in knowledge is how effective the anti‐COVID measures were in maintaining quality education. Using machine learning to analyse student grades as a proxy for educational standards, this study investigates and demonstrates the evaluative potential of machine learning (vs. traditional statistics) with respect to not only crisis responses in education but also applied studies such as Information Systems and Tourism. Main implication of this study is the analytical utility of machine learning even when educational data are irregular and small. However, incorporating accurate and meaningful data points into the existing online educational systems is crucial to leverage this utility of machine learning.

Evaluation of particle tracking codes for dispersing particles in porous media

Particle tracking (PT) is a popular technique in microscopy, microfluidics and colloidal transport studies, where image analysis is used to reconstruct trajectories from bright spots in a video. The performance of many PT algorithms has been rigorously tested for directed and Brownian motion in open media. However, PT is frequently used to track particles in porous media where complex geometries and viscous flows generate particles with high velocity variability over time. Here, we present an evaluation of four PT algorithms for a simulated dispersion of particles in porous media across a range of particle speeds and densities. Of special note, we introduce a new velocity-based PT linking algorithm (V-TrackMat) that achieves high accuracy relative to the other PT algorithms. Our findings underscore that traditional statistics, which revolve around detection and linking proficiency, fall short in providing a holistic comparison of PT codes because they tend to underpenalize aggressive linking techniques. We further elucidate that all codes analyzed show a decrease in performance due to high speeds, particle densities, and trajectory noise. However, linking algorithms designed to harness velocity data show superior performance, especially in the case of high-speed advective motion. Lastly, we emphasize how PT error can influence transport analysis.

Stress Drives Soccer Athletes'Wellness and Movement: Using Convergent Cross-Mapping to Identify Causal Relationships in a Dynamic Environment.

Prediction of athlete wellness is difficult-or, many sports-medicine practitioners and scientists would argue, impossible. Instead, one settles for correlational relationships of variables gathered at fixed moments in time. The issue may be an inherent mismatch between usual methods of data collection and analysis and the complex nature of the variables governing athlete wellness. Variables such as external load, stress, muscle soreness, and sleep quality may affect each other and wellness in a dynamic, nonlinear, way over time. In such an environment, traditional data-collection methods and statistics will fail to capture causal effects. If we are to move this area of sport science forward, a different approach is required. We analyzed data from 2 different soccer teams that showed no significance between player load and wellness or among individual measures of wellness. Our analysis used methods of attractor reconstruction to examine possible causal relationships between GPS/accelerometer-measured external training load and wellness variables. Our analysis showed that player self-rated stress, a component of wellness, seems a fundamental driving variable. The influence of stress is so great that stress can predict other components of athlete wellness, and, in turn, self-rated stress can be predicted by observing a player's load data. We demonstrate the ability of nonlinear methods to identify interactions between and among variables to predict future athlete stress. These relationships are indicative of the causal relationships playing out in athlete wellness over the course of a soccer season.

A Comparative Analysis of Deep Learning and Traditional Statistics for Stock Price and Return Forecasting

A Comparative Analysis of Deep Learning and Traditional Statistics for Stock Price and Return Forecasting

Fundamentals of Nonparametric Statistical Tests for Dental Clinical Research.

This article provides the foundation for employing nonparametric testing in dental clinical research. To make wise judgments in their research, investigators should learn more about the main nonparametric tests and their particular uses. Biostatistical analysis is essential in dental research; dental research frequently deviates from the assumptions that underpin traditional parametric statistics. Nonparametric statistics are useful for studies with small sample sizes, nominal- or ordinal-level data, and non-normally distributed variables. These statistical tests make no assumptions about the sampled population. Nonparametric tests are statistical methods based on signs and ranks. For dental research to be conducted effectively and accurately, statistical approaches must be applied correctly. Therefore, dental researchers must understand the many statistical methods at their disposal and know when to use them.

Understanding the Relationship Between Comorbidities, Medication Nonadherence, Activities of Daily Living, and Heart Condition Status Among Older Adults in the United States: A Regression Analysis and Machine Learning Approach.

Nonadherence to medication among patients with heart disease poses serious risks, including worsened heart failure and increased mortality rates. This study aims to explore the complex interplay between comorbidities, medication nonadherence, activities of daily living, and heart condition status in older American adults, using both traditional statistical methods and machine learning. Data from 326 older adults with heart conditions, drawn from the Health and Retirement Study, were analyzed. Descriptive statistics characterized demographic profiles and comorbidities, whereas logistic regression, multiple regression analyses, and decision tree models were used to address our research inquiries. In addition, a machine learning approach, specifically decision tree models, was integrated to enhance predictive accuracy. Our analysis showed that factors like age, gender, hypertension, and stroke history were significantly linked to worsening heart conditions. Notably, depression emerged as a robust predictor of medication nonadherence. Further adjusted analyses underscored significant correlations between stroke and challenges in basic activities such as dressing, bathing, and eating. Depression correlated significantly with difficulties in dressing, bed mobility, and toileting, whereas lung disease was associated with bathing hindrances. Intriguingly, our decision tree model revealed that patients experiencing dressing challenges, but not toileting difficulties, were more prone to report no improvement in heart condition status over the preceding 2 years. Blending traditional statistics with machine learning in this study reveals significant implications for crafting personalized interventions to improve patients' depression, leading to increased activities of daily living, medication adherence, reduced severity of comorbidities, and ultimately better management of heart conditions.

Enhancing the Introductory Statistics Course: Comparing Student Learning and Performance in Traditional and Web-Enhanced Traditional Courses

A redesign of a traditional introductory statistics course was performed with the intent of increasing student success in the course without the extreme burden of implementing a flipped course. A study was then conducted to compare the new format (web-enhanced traditional) to the traditional format taught by the same professor with the same assignments, quizzes, and exams during the same semester. Students selected their own sections of the course to take but were not aware of the different class formats when they enrolled. Students in both course formats were similar demographically, with similar math ACT scores. It was found that students who participated in the web-enhanced traditional course were more successful in completing the course and performed better on course assessments than those students who participated in the traditional introductory statistics course. Supplementary materials for this article are available online.

Europe has the fastest Ironman race courses and the fastest Ironman age group triathletes

The majority of participants in Ironman triathlon races are age group athletes. We have extensive knowledge about recreational athletes’ training and competition participation. Nonetheless, Ironman age group triathletes must achieve fast race times to qualify for the Ironman World Championship in Hawaii. They can, therefore, benefit from knowing where the fastest Ironman racecourses in the world are. The aim of the present study was to investigate where the fastest Ironman racecourses for age group triathletes are located in the world. Data from 677,702 Ironman age group finishers’ records (544,963 from men and 132,739 from women) originating from 228 countries and participating in 444 events across 66 different Ironman race locations between 2002 and 2022 were analyzed. Data was analyzed through traditional descriptive statistics and with machine learning regression models. Four algorithms were tested (Random Forest Regressor, XG Boost Regressor, Cat Boot Regressor, and Decision Tree Regressor). The models used gender, age group, country of origin, environmental factors (average air and water temperatures), and the event location as independent variables to predict the final overall race time. Despite the majority of successful Ironman age group triathletes originating from the USA (274,553), followed by athletes from the United Kingdom (55,410) and Canada (38,264), these countries exhibited average overall race times that were significantly slower compared to the fastest countries. Most of the triathletes competed in Ironman Wisconsin (38,545), followed by Ironman Florida (38,157) and Ironman Lake Placid (34,341). The fastest overall race times were achieved in Ironman Copenhagen (11.68 ± 1.38 h), followed by Ironman Hawaii (11.72 ± 1.86 h), Ironman Barcelona (11.78 ± 1.43 h), Ironman Florianópolis (11.80 ± 1.52 h), Ironman Frankfurt (12.03 ± 1.38 h) and Ironman Kalmar (12.08 ± 1.47 h). The fastest athletes originated from Belgium (11.48 ± 1.47 h), followed by athletes from Denmark (11.59 ± 1.40 h), Switzerland (11.62 ± 1.49 h), Austria (11.68 ± 1.50), Finland (11.68 ± 1.40 h) and Germany (11.74 ± 15.1 h). Flat running and cycling courses were associated with faster overall race times. Three of the predictive models identified the ‘country’ and ‘age group’ variables as the most important predictors. Environmental characteristics showed the lowest influence regarding the other variables. The origin of the athlete was the most predictive variable whereas environmental characteristics showed the lowest influence. Flat cycling and flat running courses were associated with faster overall race times. The fastest overall race times were achieved mainly in European races such as Ironman Copenhagen, Ironman Hawaii, Ironman Barcelona, Ironman Florianópolis, Ironman Frankfurt and Ironman Kalmar. The fastest triathletes originated from European countries such as Belgium, Denmark, Switzerland, Austria, Finland, and Germany.

A nowcasting model of industrial production using alternative data and machine learning approaches

Recent years have seen a growing trend to utilize "alternative data" in addition to traditional statistical data in order to understand and assess economic conditions in real time. In this paper, we construct a nowcasting model for the Indices of Industrial Production (IIP), which measure production activity in the manufacturing sector in Japan. The model has the following characteristics: First, it uses alternative data (mobility data and electricity demand data) that is available in real-time and can nowcast the IIP one to two months before their official release. Second, the model employs machine learning techniques to improve the nowcasting accuracy by endogenously changing the mixing ratio of nowcast values based on traditional economic statistics (the Indices of Industrial Production Forecast) and nowcast values based on alternative data, depending on the economic situation. The estimation results show that by applying machine learning techniques to alternative data, production activity can be nowcasted with high accuracy, including when it went through large fluctuations during the spread of the COVID-19 pandemic.

Ultrasonic signal classification for composite materials via deep convolutional neural networks

ABSTRACT Ultrasonic testing (UT) is the most commonly used non-destructive testing method in the aerospace field. However, for the detection of delamination defect in multi-layer composite adhesive materials with metal and non-metal bonding, traditional energy statistics methods are limited and cannot achieve high-precision automatic signal classification. This article proposes a signal classification model based on hybrid neural networks. By studying the classification accuracy of convolutional neural networks (CNN) and long short-term memory networks (LSTM) for different types of signals, a relatively balanced classification is achieved, which effectively improves the classification accuracy. By integrating attention mechanisms, the ability of the detection model to identify key features is further enhanced. Experimental results demonstrate that the proposed model achieves an accuracy of 98.57%. Bayesian Optimisation (BO) can effectively and automatically select the optimal hyperparameters, and achieve global optimisation. In the experiment, the accuracy increased by 0.73% compared to the benchmark value. Comparative experiments show that the signal classification model established in this paper has good performance.

Formation Mechanism and Source Apportionment of Hydrochemical Components in Groundwater in the Yinchuan Plain

Groundwater is one of the major water sources for production, living, and agricultural irrigation in the Yinchuan Plain. Owing to the influence of the regional environmental background and long-term effects of human activities, groundwater quality is generally inferior. To deeply analyze the formation mechanism and source of hydrochemical components in groundwater in the Yinchuan Plain, the traditional hydrochemical graphic method and mathematical statistics and principal component analysis-multivariate linear statistical model were used. Based on inorganic component contents of 100 phreatic water samples and 46 confined groundwater samples, the hydrochemical characteristics and quality status, spatial distribution of over-limit toxicological components, and contribution rate of hydrochemical components were analyzed. The results showed that the chemical components of groundwater were controlled by rock weathering and evaporation concentration. Dissolution-enrichment （F1）, original geological environment （F2）, and human activities（F3） were the principal factors that influenced groundwater hydrochemistry with the contribution rates of 73.67%, 14.45%, and 11.88%, respectively. The major over-limit toxicity indices in groundwater were NO3--N and F-. High NO3--N phreatic water was mainly influenced by agriculture activities, followed by the discharge of domestic sewage. Enrichment of groundwater F- was mainly caused by leaching of F-bearing minerals and cation exchange adsorption.

A comparative study of cosmological constraints from weak lensing using Convolutional Neural Networks

Weak Lensing (WL) surveys are reaching unprecedented depths, enabling the investigation of very small angular scales. At these scales, nonlinear gravitational effects lead to higher-order correlations making the matter distribution highly non-Gaussian. Extracting this information using traditional statistics has proven difficult, and Machine Learning based summary statistics have emerged as a powerful alternative. We explore the capabilities of a discriminative, Convolutional Neural Networks (CNN) based approach, focusing on parameter constraints in the (Ω m , σ8) cosmological parameter space. Leveraging novel training loss functions and network representations on WL mock datasets without baryons, we show that our models achieve ~ 5 times higher figure of merit in the σ8-Ω m plane than the power spectrum, ~ 3 times higher than peak counts, and ~ 2 times higher than previous CNN-learned summary statistics and scattering transforms, for noise levels relevant to Rubin or Euclid. For WL convergence maps with baryonic physics, our models achieve ~ 2.3 times stronger constraining power than the power spectrum at these noise levels, also outperforming previous summary statistics. To further explore the possibilities of CNNs for this task, we also discuss transfer learning where we adapt pre-trained models, trained on different tasks or datasets, for cosmological inference, finding that these do not improve the performance.

Neoliberal transformation through circulation of economic elites in times of critical junctures: the case of 1999–2001 economic crises in Turkey

ABSTRACT While it is well-established that exogenous shocks can trigger policy changes, political elites have often been depicted as the initiators of this process. This paper adopts an elite sociology perspective, extending beyond the exclusive focus on political elites, to uncover the pathway through which critical junctures foster policy changes. It employs a case study of the economic crises in Turkey spanning from 1999 to 2001 as an illustrative example. During this period, a significant shift occurred, moving from traditional statist economic policies to widespread neoliberal reforms. Our analysis reveals that this policy shift was not solely influenced by external factors, but facilitated by an intricate process of elite circulation within the economic bureaucracy. This process, which began in the 1980s and gained momentum in the 2000s, coincided with an influx of bureaucratic elites aligned with neoliberal principles and played a crucial role in shaping the trajectory of policy changes.