clusterBMA: Bayesian model averaging for clustering.

Various methods have been developed to combine inference across multiple sets of results for unsupervised clustering, within the ensemble clustering literature. The approach of reporting results from one 'best' model out of several candidate clustering models generally ignores the uncertainty that arises from model selection, and results in inferences that are sensitive to the particular model and parameters chosen. Bayesian model averaging (BMA) is a popular approach for combining results across multiple models that offers some attractive benefits in this setting, including probabilistic interpretation of the combined cluster structure and quantification of model-based uncertainty. In this work we introduce clusterBMA, a method that enables weighted model averaging across results from multiple unsupervised clustering algorithms. We use clustering internal validation criteria to develop an approximation of the posterior model probability, used for weighting the results from each model. From a combined posterior similarity matrix representing a weighted average of the clustering solutions across models, we apply symmetric simplex matrix factorisation to calculate final probabilistic cluster allocations. In addition to outperforming other ensemble clustering methods on simulated data, clusterBMA offers unique features including probabilistic allocation to averaged clusters, combining allocation probabilities from 'hard' and 'soft' clustering algorithms, and measuring model-based uncertainty in averaged cluster allocation. This method is implemented in an accompanying R package of the same name. We use simulated datasets to explore the ability of the proposed technique to identify robust integrated clusters with varying levels of separation between subgroups, and with varying numbers of clusters between models. Benchmarking accuracy against four other ensemble methods previously demonstrated to be highly effective in the literature, clusterBMA matches or exceeds the performance of competing approaches under various conditions of dimensionality and cluster separation. clusterBMA substantially outperformed other ensemble methods for high dimensional simulated data with low cluster separation, with 1.16 to 7.12 times better performance as measured by the Adjusted Rand Index. We also explore the performance of this approach through a case study that aims to identify probabilistic clusters of individuals based on electroencephalography (EEG) data. In applied settings for clustering individuals based on health data, the features of probabilistic allocation and measurement of model-based uncertainty in averaged clusters are useful for clinical relevance and statistical communication.

A global perspective on the intrinsic dimensionality of COVID-19 data.

We develop a novel global perspective of the complexity of the relationships between three COVID-19 datasets, the standardised per-capita growth rate of COVID-19 cases and deaths, and the Oxford Coronavirus Government Response Tracker COVID-19 Stringency Index (CSI) which is a measure describing a country's stringency of lockdown policies. We use a state-of-the-art heterogeneous intrinsic dimension estimator implemented as a Bayesian mixture model, called Hidalgo. Our findings suggest that these highly popular COVID-19 statistics may project onto two low-dimensional manifolds without significant information loss, suggesting that COVID-19 data dynamics are generated from a latent mechanism characterised by a few important variables. The low dimensionality imply a strong dependency among the standardised growth rates of cases and deaths per capita and the CSI for countries over 2020-2021. Importantly, we identify spatial autocorrelation in the intrinsic dimension distribution worldwide. The results show how high-income countries are more prone to lie on low-dimensional manifolds, likely arising from aging populations, comorbidities, and increased per capita mortality burden from COVID-19. Finally, the temporal stratification of the dataset allows the examination of the intrinsic dimension at a more granular level throughout the pandemic.

Being Bayesian in the 2020s: opportunities and challenges in the practice of modern applied Bayesian statistics.

Building on a strong foundation of philosophy, theory, methods and computation over the past three decades, Bayesian approaches are now an integral part of the toolkit for most statisticians and data scientists. Whether they are dedicated Bayesians or opportunistic users, applied professionals can now reap many of the benefits afforded by the Bayesian paradigm. In this paper, we touch on six modern opportunities and challenges in applied Bayesian statistics: intelligent data collection, new data sources, federated analysis, inference for implicit models, model transfer and purposeful software products. This article is part of the theme issue 'Bayesian inference: challenges, perspectives, and prospects'.

Graph neural network-based anomaly detection for river network systems.

Background: Water is the lifeblood of river networks, and its quality plays a crucial role in sustaining both aquatic ecosystems and human societies. Real-time monitoring of water quality is increasingly reliant on in-situ sensor technology.Anomaly detection is crucial for identifying erroneous patterns in sensor data, but can be a challenging task due to the complexity and variability of the data, even under typical conditions. This paper presents a solution to the challenging task of anomaly detection for river network sensor data, which is essential for accurate and continuous monitoring. Methods: We use a graph neural network model, the recently proposed Graph Deviation Network (GDN), which employs graph attention-based forecasting to capture the complex spatio-temporal relationships between sensors. We propose an alternate anomaly threshold criteria for the model, GDN+, based on the learned graph. To evaluate the model's efficacy, we introduce new benchmarking simulation experiments with highly-sophisticated dependency structures and subsequence anomalies of various types. We also introduce software called gnnad. Results: We further examine the strengths and weaknesses of this baseline approach, GDN, in comparison to other benchmarking methods on complex real-world river network data. Conclusions: Findings suggest that GDN+ outperforms the baseline approach in high-dimensional data, while also providing improved interpretability.

EEG-based clusters differentiate psychological distress, sleep quality and cognitive function in adolescents.

INTRODUCTION: To better understand the relationships between neurophysiology, cognitive function and psychopathology risk in adolescence there is value in identifying data-driven subgroups based on measurements of brain activity and function, and then comparing cognition and mental health between such subgroups. METHODS: We developed a flexible and scaleable multi-stage analysis pipeline to identify data-driven clusters of 12-year-olds (M = 12.64, SD = 0.32) based on frequency characteristics calculated from resting state, eyes-closed electroencephalography (EEG) recordings. For this preliminary cross-sectional study, EEG data was collected from 59 individuals in the Longitudinal Adolescent Brain Study (LABS) being undertaken in Queensland, Australia. Applying multiple unsupervised clustering algorithms to these EEG features, we identified well-separated subgroups of individuals. To study patterns of difference in cognitive function and mental health symptoms between clusters, we applied Bayesian regression models to probabilistically identify differences in these measures between clusters. RESULTS: We identified 5 core clusters associated with distinct subtypes of resting state EEG frequency content. Bayesian models demonstrated substantial differences in psychological distress, sleep quality and cognitive function between clusters. By examining associations between neurophysiology and health measures across clusters, we have identified preliminary risk and protective profiles linked to EEG characteristics. CONCLUSION: This method provides the potential to identify neurophysiological subgroups of adolescents in the general population based on resting state EEG, and associated patterns of health and cognition that are not observed at the whole group level. This approach offers potential utility in clinical risk prediction for mental and cognitive health outcomes throughout adolescent development.

Connecting virtual reality and ecology: a new tool to run seamless immersive experiments in R.

Virtual reality (VR) technology is an emerging tool that is supporting the connection between conservation research and public engagement with environmental issues. The use of VR in ecology consists of interviewing diverse groups of people while they are immersed within a virtual ecosystem to produce better information than more traditional surveys. However, at present, the relatively high level of expertise in specific programming languages and disjoint pathways required to run VR experiments hinder their wider application in ecology and other sciences. We present R2VR, a package for implementing and performing VR experiments in R with the aim of easing the learning curve for applied scientists including ecologists. The package provides functions for rendering VR scenes on web browsers with A-Frame that can be viewed by multiple users on smartphones, laptops, and VR headsets. It also provides instructions on how to retrieve answers from an online database in R. Three published ecological case studies are used to illustrate the R2VR workflow, and show how to run a VR experiments and collect the resulting datasets. By tapping into the popularity of R among ecologists, the R2VR package creates new opportunities to address the complex challenges associated with conservation, improve scientific knowledge, and promote new ways to share better understanding of environmental issues. The package could also be used in other fields outside of ecology.

Development and evaluation of a matrix for assessing fatigue-related risk, derived from a national survey of nurses' work patterns.

BACKGROUND: Multiple aspects of nurses' rosters interact to affect the quality of patient care they can provide and their own health, safety and wellbeing. OBJECTIVES: (1) Develop and test a matrix incorporating multiple aspects of rosters and recovery sleep that are individually associated with three fatigue-related outcomes - fatigue-related clinical errors, excessive sleepiness and sleepy driving; and (2) evaluate whether the matrix also predicts nurses' ratings of the effects of rosters on aspects of life outside work. DESIGN: Develop and test the matrix using data from a national survey of nurses' fatigue and work patterns in six hospital-based practice areas with high fatigue risk. METHODS: Survey data included demographics, work patterns (previous 14 days), choice about shifts, and the extent to which work patterns cause problems with social life, home life, personal relationships, and other commitments (rated 1 = not at all to 5 = very much). Matrix variables were selected based on univariate associations with the fatigue-related outcomes, limits in the collective employment contract, and previous research. Each variable was categorised as lower (score 0), significant (score 1), or higher risk (score 2). Logistic multiple regression modelling tested the independent predictive power of matrix scores against models including all the (uncategorised) work pattern and recovery sleep variables with significant univariate associations with each outcome variable. Model fit was measured using Akaike and Bayesian Information Criterion statistics. RESULTS: Data were included from 2358 nurses who averaged at least 30 h/week in the previous fortnight in one of the target practice areas. Final matrix variables were: total hours worked; number of shift extensions >30 min, night shifts; breaks < 9 h; breaks ≥ 24 h; nights with sleep 11pm to 7am; days waking fully rested; and roster change. After controlling for gender, ethnicity, years of nursing experience, and the extent of shift choice, the matrix score was a significant independent predictor of each of the three fatigue-related outcomes, and for all four aspects of life outside work. For all outcome variables, the model including the matrix score was a better fit to the data than the equivalent model including all the (uncategorised) work pattern variables. CONCLUSIONS: A matrix that predicts the likelihood of nurses reporting fatigue-related safety outcomes can be used to compare the impact of rosters both at work and outside work. It can be used for roster design and management, and to guide nurses' choices about the shifts they work.

Fatigue and nurses' work patterns: An online questionnaire survey.

BACKGROUND: Fatigue resulting from shift work and extended hours can compromise patient care and the safety and health of nurses, as well as increasing nursing turnover and health care costs. OBJECTIVES: This research aimed to identify aspects of nurses' work patterns associated with increased risk of reporting fatigue-related outcomes. DESIGN: A national survey of work patterns and fatigue-related outcomes in 6 practice areas expected to have high fatigue risk (child health including neonatology, cardiac care/intensive care, emergency and trauma, in-patient mental health, medical, and surgical nursing). METHODS: The 5-page online questionnaire included questions addressing: demographics, usual work patterns, work in the previous two weeks, choice about shifts, and four fatigue-related outcomes - having a sleep problem for at least 6 months, sleepiness (Epworth Sleepiness Scale), recalling a fatigue-related error in clinical practice in the last 6 months, and feeling close to falling asleep at the wheel in the last 12 months. The target population was all registered and enrolled nurses employed to work in public hospitals at least 30 h/week in one of the 6 practice areas. Participation was voluntary and anonymous. RESULTS: Respondents (n = 3133) were 89.8% women and 8% Maori (indigenous New Zealanders), median age 40 years, range 21-71 years (response rate 42.6%). Nurses were more likely than New Zealand adults in general to report chronic sleep problems (37.73% vs 25.09%, p < 0.0001) and excessive sleepiness (33.75% vs 14.9%, p < 0.0001). Fatigue-related error(s) in the last 6 months were recalled by 30.80% and 64.50% reported having felt sleepy at the wheel in the last 12 months. Logistic regression analyses indicated that fatigue-related outcomes were most consistently associated with shift timing and sleep. Risk increased with more night shifts and decreased with more nights with sleep between 11 p.m. and 7 a.m. and on which nurses had enough sleep to feel fully rested. Risk also increased with roster changes and more shift extensions greater than 30 min and decreased with more choice about shifts. Comparisons between intensive care/cardiac care and in-patient mental health nursing highlight that fatigue has different causes and consequences in different practice areas. CONCLUSIONS: Findings confirm the need for a more comprehensive and adaptable approach to managing fatigue. We advocate an approach that integrates safety management and scientific principles with nursing and management expertise. It should be data-driven, risk-focused, adaptable, and resilient in the face of changes in the services required, the resources available, and the overall goals of the healthcare system.

Monitoring the Effectiveness of Fatigue Risk Management: A Survey of Pilots' Concerns.

INTRODUCTION: Airlines are required to monitor the effectiveness of their pilot fatigue risk management. The present survey sought the views of all pilots at Delta Air Lines on fatigue-related issues raised by their colleagues participating in regular airline safety audits. METHODS: All 13,217 pilots from 9 aircraft fleets were invited to participate in an anonymous online survey. Questions related to aspects of scheduling, fatigue mitigations, and fatigue safety culture. RESULTS: There were 1108 pilots who completed the survey (response rate = 8.4%). On 7/9 fleets, most pilots thought 5- to 7-d rotations were too long (exceptions: B747, median = 14 d; A330 median = 8.5 d). In the previous year, on average across all fleets, 60.6% of pilots had worked up to or beyond their personal rotation limit (minimum, B747 = 6.3%; maximum, MD88/90 = 75.9%). Rotations where duty periods start progressively earlier were considered highly fatiguing by 73.8% of pilots, compared to 14.7% for rotations where duty periods started progressively later and 1.6% for rotations with successive duty periods starting at the same time. The median optimum break length between rotations was 3-4 d. On 7/9 fleets, fewer than 20% of pilots tried to build their monthly schedules with back-to-back rotations (exceptions: B747, 43.8%; A330, 34.3%). Awareness of fatigue and perceptions of company fatigue risk management activities varied widely among fleets. DISCUSSION: The findings identify possible improvements in fatigue risk management and highlight that care is needed when extrapolating from one operational context to another. As a safety assurance exercise, we recommend repeating the survey biannually, or sooner if warranted by specific circumstances.Gander P, Mangie J, Phillips A, Santos-Fernandez E, Wu LJ. Monitoring the effectiveness of fatigue risk management: a survey of pilots' concerns. Aerosp Med Hum Perform. 2018; 89(10):889-895.