Bayesian information criterion approximations to Bayes factors for univariate and multivariate logistic regression models

Introduction to Bayesian Analyses for Clinical Research

Bayesian analyses are becoming more popular as a means of analyzing data, yet the Bayesian approach is novel to many members of the broad clinical audience. While Bayesian analyses are foundational to anesthesia pharmacokinetic/pharmacodynamic modeling, they also can be used for analyzing data from clinical trials or observational studies. The traditional null hypothesis significance testing (frequentist) approach uses only the data collected from the current study to make inferences. On the other hand, the Bayesian approach quantifies the external information or expert knowledge and combines the external information with the study data, then makes inference from this combined information. We introduce to the clinical and translational science researcher what it means to do Bayesian statistics, why a researcher would choose to perform their analyses using the Bayesian approach, when it would be advantageous to use a Bayesian instead of a frequentist approach, and how Bayesian analyses and interpretations differ from the more traditional frequentist methods. Throughout this paper, we use various pain- and anesthesia-related examples to highlight the ideas and statistical concepts that should be relatable to other areas of research as well.

A longitudinal investigation of non-suicidal self-injury persistence patterns, risk factors, and clinical outcomes during the college period

BACKGROUND

Although non-suicidal self-injury (NSSI) is known typically to begin in adolescence, longitudinal information is lacking about patterns, predictors, and clinical outcomes of NSSI persistence among emerging adults. The present study was designed to (1) estimate NSSI persistence during the college period, (2) identify risk factors and high-risk students for NSSI persistence patterns, and (3) evaluate the association with future mental disorders and suicidal thoughts and behaviors (STB).

METHODS

Using prospective cohorts from the Leuven College Surveys (n = 5915), part of the World Mental Health International College Student Initiative, web-based surveys assessed mental health and psychosocial problems at college entrance and three annual follow-up assessments.

RESULTS

Approximately one in five (20.4%) students reported lifetime NSSI at college entrance. NSSI persistence was estimated at 56.4%, with 15.6% reporting a high-frequency repetitive pattern (≥five times yearly). Many hypothesized risk factors were associated with repetitive NSSI persistence, with the most potent effects observed for pre-college NSSI characteristics. Multivariate models suggest that an intervention focusing on the 10-20% at the highest predicted risk could effectively reach 34.9-56.7% of students with high-frequency repetitive NSSI persistence (PPV = 81.8-93.4, AUC = 0.88-0.91). Repetitive NSSI persistence during the first two college years predicted 12-month mental disorders, role impairment, and STB during the third college year, including suicide attempts.

CONCLUSIONS

Most emerging adults with a history of NSSI report persistent self-injury during their college years. Web-based screening may be a promising approach for detecting students at risk for a highly persistent NSSI pattern characterized by subsequent adverse outcomes.

Prehospital stratification and prioritisation of non-ST-segment elevation acute coronary syndrome patients (NSTEACS): the MARIACHI scale

OBJECTIVE

The objective of this study was to develop and validate a risk scale (MARIACHI) for patients classified as non-ST-segment elevation acute coronary syndrome (NSTEACS) in a prehospital setting with the ability to identify patients at an increased risk of mortality at an early stage.

METHODS

A retrospective observational study conducted in Catalonia over two periods: 2015-2017 (development and internal validation cohort) and Aug 2018-Jan 2019 (external validation cohort). We included patients classified as prehospital NSTEACS, assisted by an advanced life support unit and requiring hospital admission. The primary outcome was in-hospital mortality. Cohorts were compared using logistic regression and a predictive model was created using bootstrapping techniques.

RESULTS

The development and internal validation cohort included 519 patients. The model is composed of five variables associated with hospital mortality: age, systolic blood pressure, heart rate > 95 bpm, Killip-Kimball III-IV and ST depression ≥ 0.5 mm. The model showed good overall performance (Brier = 0.043) and consistency in discrimination (AUC 0.88, 95% CI 0.83-0.92) and calibration (slope = 0.91; 95% CI 0.89-0.93). We included 1316 patients for the external validation sample. There was no difference in discrimination (AUC 0.83, 95% CI 0.78-0.87; DeLong Test p = 0.071), but there was in calibration (p < 0.001), so it was recalibrated. The finally model obtained was stratified and scored into three groups according to the predicted risk of patient in-hospital mortality: low risk: < 1% (-8 to 0 points), moderate risk: 1-5% (+ 1 to + 5 points) and high risk: > 5% (6-12 points).

CONCLUSION

The MARIACHI scale showed correct discrimination and calibration to predict high-risk NSTEACS. Identification of high-risk patients may help with treatment and low referral decisions at the prehospital level.

Airway Phenotypes and Nocturnal Wearing of Dentures in Elders with Sleep Apnea

The objective of this study was to examine to what extent the anatomic characteristics of the upper airway can influence the effect of nocturnal wearing of dentures on the sleep of edentulous elders with untreated sleep apnea. This study used the data from a randomized crossover clinical trial and an exploratory approach to address its objectives. Cone beam computed tomography scans of 65 edentulous individuals (female, n = 37; male, n = 28; mean ± SD age, 74.54 ± 6.42 y) with untreated obstructive sleep apnea (OSA) were used to identify anatomic variables. Polysomnography data were collected by means of one portable overnight recording. The respiratory variable values, including apnea-hypopnea index (AHI), with and without denture worn during sleep were used to calculate the change. Statistical analyses included multiple linear regressions, cluster analysis, and binary logistic regressions. A receiver operator characteristic curve was used to illustrate the accuracy of the statistical model. The regression model explained 15.8% (R²) of AHI change. An increase in the lateral dimension of the minimum cross-sectional area was associated with a decrease in AHI, oxygen desaturation index, and respiratory arousal index changes (P ≤ 0.041). Furthermore, an increase in the length of the hypopharynx was associated with an increase in AHI and oxygen desaturation index changes (P ≤ 0.027). An increase in the lateral dimension of the minimum cross-sectional area of the upper airway was associated with a decreased likelihood of being in the group having a worsened AHI (odds ratio = 0.85; 95% CI, 0.76 to 0.95; P = 0.006). An increase in the length of the oropharynx was associated with an increased likelihood of having increased AHI (odds ratio = 1.10; 95% CI, 1.01 to 1.20; P = 0.026). The nocturnal aggravation of respiratory variables in edentulous individuals with OSA who wear dentures at night can be linked to certain anatomic characteristics of the upper airway. Replication of these findings may open novel avenues for personalized advice regarding nocturnal wearing of dentures in edentulous individuals with OSA (ClinicalTrials.gov: NCT01868295).

Risk and Protective Profile of Men Who Have Sex With Men Using Mobile Voluntary HIV Counseling and Testing: Latent Class Analysis

BACKGROUND

Mobile voluntary counseling and testing (VCT) for HIV has been carried out to improve the targeting of at-risk populations and HIV case detection for men who have sex with men (MSM). However, the HIV-positive detection rate using this screening strategy has declined in recent years. This may imply unknown changes in risk-taking and protective features jointly influencing the testing results. These changing patterns in this key population remain unexplored.

OBJECTIVE

The aim of this study was to identify the nuanced group classification of MSM who underwent mobile VCT using latent class analysis (LCA), and to compare the difference in characteristics and testing results between subgroups.

METHODS

A cross-sectional research design and purposive sampling were applied between May 21, 2019, and December 31, 2019. Participants were recruited by a well-trained research assistant through social networking platforms, including the most popular instant messenger app Line, geosocial network apps dedicated to MSM, and online communities. Mobile VCT was provided to participants at an assigned time and place. Demographic characteristics and risk-taking and protective features of the MSM were collected via online questionnaires. LCA was used to identify discrete subgroups based on four risk-taking indicators-multiple sexual partners (MSP), unprotected anal intercourse (UAI), recreational drug use within the past 3 months, and history of sexually transmitted diseases-and three protective indicators-experience of postexposure prophylaxis, preexposure prophylaxis use, and regular HIV testing.

RESULTS

Overall, 1018 participants (mean age 30.17, SD 7.29 years) were included. A three-class model provided the best fit. Classes 1, 2, and 3 corresponded to the highest risk (n=175, 17.19%), highest protection (n=121, 11.89%), and low risk and low protection (n=722, 70.92%), respectively. Compared to those of class 3, class 1 participants were more likely to have MSP and UAI within the past 3 months, to be ≥40 years of age (odds ratio [OR] 2.197, 95% CI 1.357-3.558; P=.001), to have HIV-positive results (OR 6.47, 95% CI 2.272-18.482; P<.001), and a CD4 count ≤349/μL (OR 17.50, 95% CI 1.223-250.357; P=.04). Class 2 participants were more likely to adopt biomedical preventions and have marital experience (OR 2.55, 95% CI 1.033-6.277; P=.04).

CONCLUSIONS

LCA helped derive a classification of risk-taking and protection subgroups among MSM who underwent mobile VCT. These results may inform policies for simplifying the prescreening assessment and more precisely recognizing those who have higher probabilities of risk-taking features but remain undiagnosed targets, including MSM engaging in MSP and UAI within the past 3 months and those ≥40 years old. These results could be applied to tailor HIV prevention and testing programs.

Construction of Mathematics Teaching Environment Based on Big Data on the Wisdom Cloud Platform of Higher Vocational Education

As a public basic course in the system of higher vocational colleges, mathematics has received more and more attention from education administrators. However, there are still problems in higher education that mathematics teaching resources are scatter and data are difficult to integrate. Big data (DT) and smart clouds are the burgeon of intelligent technology, which has provided support for teaching informatization and teaching data integration in colleges and universities. This article aims to solve the problems of mathematics teaching in higher vocational education, such as more content and less class hours, difficult, higher vocational education due to the expansion of enrollment, and the decline in mathematics teaching quality, and build a smart cloud platform based on big data mathematics instructing environment system and mathematics wisdom instructing. The reconstruction of the environment is the deep integration of information technology and mathematics instructing, which can promote the deep integration of information technology with teaching, management, and environment. The experiments in this article show that the application of the smart cloud platform to mathematics teaching can greatly increase students' enthusiasm for learning and improve the school's mathematics teaching level, and the improvement rate of students' performance can reach 99.89%. The research in this article is of great significance to the adjustment of mathematics teaching environment in higher professional education.

Effective Connectivity for Decoding Electroencephalographic Motor Imagery Using a Probabilistic Neural Network

Motor imagery (MI)-based brain-computer interfaces have gained much attention in the last few years. They provide the ability to control external devices, such as prosthetic arms and wheelchairs, by using brain activities. Several researchers have reported the inter-communication of multiple brain regions during motor tasks, thus making it difficult to isolate one or two brain regions in which motor activities take place. Therefore, a deeper understanding of the brain's neural patterns is important for BCI in order to provide more useful and insightful features. Thus, brain connectivity provides a promising approach to solving the stated shortcomings by considering inter-channel/region relationships during motor imagination. This study used effective connectivity in the brain in terms of the partial directed coherence (PDC) and directed transfer function (DTF) as intensively unconventional feature sets for motor imagery (MI) classification. MANOVA-based analysis was performed to identify statistically significant connectivity pairs. Furthermore, the study sought to predict MI patterns by using four classification algorithms-an SVM, KNN, decision tree, and probabilistic neural network. The study provides a comparative analysis of all of the classification methods using two-class MI data extracted from the PhysioNet EEG database. The proposed techniques based on a probabilistic neural network (PNN) as a classifier and PDC as a feature set outperformed the other classification and feature extraction techniques with a superior classification accuracy and a lower error rate. The research findings indicate that when the PDC was used as a feature set, the PNN attained the greatest overall average accuracy of 98.65%, whereas the same classifier was used to attain the greatest accuracy of 82.81% with the DTF. This study validates the activation of multiple brain regions during a motor task by achieving better classification outcomes through brain connectivity as compared to conventional features. Since the PDC outperformed the DTF as a feature set with its superior classification accuracy and low error rate, it has great potential for application in MI-based brain-computer interfaces.

Bayesian information criterion approximations to Bayes factors for univariate and multivariate logistic regression models

Schwarz's criterion, also known as the Bayesian Information Criterion or BIC, is commonly used for model selection in logistic regression due to its simple intuitive formula. For tests of nested hypotheses in independent and identically distributed data as well as in Normal linear regression, previous results have motivated use of Schwarz's criterion by its consistent approximation to the Bayes factor (BF), defined as the ratio of posterior to prior model odds. Furthermore, under construction of an intuitive unit-information prior for the parameters of interest to test for inclusion in the nested models, previous results have shown that Schwarz's criterion approximates the BF to higher order in the neighborhood of the simpler nested model. This paper extends these results to univariate and multivariate logistic regression, providing approximations to the BF for arbitrary prior distributions and definitions of the unit-information prior corresponding to Schwarz's approximation. Simulations show accuracies of the approximations for small samples sizes as well as comparisons to conclusions from frequentist testing. We present an application in prostate cancer, the motivating setting for our work, which illustrates the approximation for large data sets in a practical example.