Bayesian Inference under Small Sample Sizes Using General Noninformative Priors

This paper proposes a Bayesian inference method for problems with small sample sizes. A general type of noninformative prior is proposed to formulate the Bayesian posterior. It is shown that this type of prior can represent a broad range of priors such as classical noninformative priors and asymptotically locally invariant priors and can be derived as the limiting states of normal-inverse-Gamma conjugate priors, allowing for analytical evaluations of Bayesian posteriors and predictors. The performance of different noninformative priors under small sample sizes is compared using the likelihood combining both fitting and prediction performances. Laplace approximation is used to evaluate the likelihood. A realistic fatigue reliability problem was used to illustrate the method. Following that, an actual aeroengine disk lifing application with two test samples is presented, and the results are compared with the existing method.

Quantum sensing for gravitational cartography

The sensing of gravity has emerged as an important tool in geophysics for applications such as engineering and climate research, where it provides the capability to probe otherwise inaccessible features under the surface of the Earth. Examples include the monitoring of temporal variations such as those found in aquifers and geodesy. However, resolving metre scale underground features is rendered impractical by the long measurement times needed for the removal of vibrational noise. Here, we overcome this limitation and open up the field of gravity cartography by realising a practical quantum gravity gradient sensor. Our design suppresses the effects of micro-seismic and laser noise, as well as thermal and magnetic field variations, and instrument tilt. The instrument achieves an uncertainty of 20 E (1 E = 10^-9 s^-2) and is used to perform a 0.5 m spatial resolution survey across a 8.5 m long line, detecting a 2 m tunnel with a signal to noise ratio of 8. The tunnel centre is localised using a Bayesian inference method, determining the centre to within ± 0.19 m in the horizontal direction and finding the centre depth as (1.89 -0.59/+2.3) m. The removal of vibrational noise enables improvements in instrument performance to directly translate into reduced measurement time in mapping. This opens new applications such as mapping the water distribution of aquifers and evaluating impacts on the water table, detecting new features in archaeology, determination of soil properties and water content ,and reducing the risk of unforeseen ground conditions in the construction of critical energy, transport and utilities infrastructure, providing a new window into the underground.

Unraveling the COVID-19 hospitalization dynamics in Spain using publicly available data

SummaryBackgroundOne of the main challenges of the ongoing COVID-19 pandemic is to be able to make sense of available, but often heterogeneous and noisy data, to characterize the evolution of the SARS-CoV-2 infection dynamics, with the additional goal of having better preparedness and planning of healthcare services. This contribution presents a data-driven methodology that allows exploring the hospitalization dynamics of COVID-19, exemplified with a study of 17 autonomous regions in Spain.MethodsWe use data on new daily cases and hospitalizations reported by the Ministry of Health of Spain to implement a Bayesian inference method that allows making short and mid-term predictions of bed occupancy of COVID-19 patients in each of the autonomous regions of the country.FindingsWe show how to use given and generated temporal series for the number of daily admissions and discharges from hospital to reproduce the hospitalization dynamics of COVID-19 patients. For the case-study of the region of Aragon, we estimate that the probability of being admitted to hospital care upon infection is 0·090 [0·086-0·094], (95% C.I.), with the distribution governing hospital admission yielding a median interval of 3·5 days and an IQR of 7 days. Likewise, the distribution on the length of stay produces estimates of 12 days for the median and 10 days for the IQR. A comparison between model parameters for the regions analyzed allows to detect differences and changes in policies of the health authorities.InterpretationThe amount of data that is currently available is limited, and sometimes unreliable, hindering our understanding of many aspects of this pandemic. We have observed important regional differences, signaling that to properly compare very different populations, it is paramount to acknowledge all the diversity in terms of culture, socio-economic status and resource availability. To better understand the impact of this pandemic, much more data, disaggregated and properly annotated, should be made available.

Machine Learning Solutions for Bridge Scour Forecast Based on Monitoring Data

Bridge scour is a challenge throughout the U.S.A. and other countries. Despite the scale of the issue, there is still a substantial lack of robust methods for scour prediction to support reliable, risk-based management and decision making. Throughout the past decade, the use of real-time scour monitoring systems has gained increasing interest among state departments of transportation across the U.S.A. This paper introduces three distinct methodologies for scour prediction using advanced artificial intelligence (AI)/machine learning (ML) techniques based on real-time scour monitoring data. Scour monitoring data included the riverbed and river stage elevation time series at bridge piers gathered from various sources. Deep learning algorithms showed promising in prediction of bed elevation and water level variations as early as a week in advance. Ensemble neural networks proved successful in the predicting the maximum upcoming scour depth, using the observed sensor data at the onset of a scour episode, and based on bridge pier, flow and riverbed characteristics. In addition, two of the common empirical scour models were calibrated based on the observed sensor data using the Bayesian inference method, showing significant improvement in prediction accuracy. Overall, this paper introduces a novel approach for scour risk management by integrating emerging AI/ML algorithms with real-time monitoring systems for early scour forecast.

Bayesian Estimation of Ammunition Demand Based on Multinomial Distribution

In view of the small sample size of combat ammunition trial data and the difficulty of forecasting the demand for combat ammunition, a Bayesian inference method based on multinomial distribution is proposed. Firstly, considering the different damage grades of ammunition hitting targets, the damage results are approximated as multinomial distribution, and a Bayesian inference model of ammunition demand based on multinomial distribution is established, which provides a theoretical basis for forecasting the ammunition demand of multigrade damage under the condition of small samples. Secondly, the conjugate Dirichlet distribution of multinomial distribution is selected as a prior distribution, and Dempster–Shafer evidence theory (D-S theory) is introduced to fuse multisource previous information. Bayesian inference is made through the Markov chain Monte Carlo method based on Gibbs sampling, and ammunition demand at different damage grades is obtained by referring to cumulative damage probability. The study result shows that the Bayesian inference method based on multinomial distribution is highly maneuverable and can be used to predict ammunition demand of different damage grades under the condition of small samples.

COVID-19 Vaccine Prioritisation in Japan and South Korea

Background Due to a limited initial supply of COVID-19 vaccines, the prioritisation of individuals for vaccination is of utmost importance for public health. Here, we provide the optimal allocation strategy for COVID-19 vaccines according to age in Japan and South Korea. Methods Combining national case reports, age-specific contact matrices, and observed periods between each stages of infection (Susceptible-Exposed-Infectious-Quarantined), we constructed a compartmental model. We estimated the age-stratified probability of transmission given contact (q_i) using Bayesian inference method and simulated different vaccination scenarios to reduce either case numbers or death toll. We also performed sensitivity analyses on the proportion of asymptomatic cases and vaccine efficacy. Findings The model inferred age-stratified probability of transmission given contact (q_i) showed similar age-dependent increase in Japan and South Korea. Assuming the reported COVID-19 vaccine efficacy, our results indicate that Japan and South Korea need to prioritise individuals aged 20-35 years and individuals aged over 60 years, respectively, to minimise case numbers. To minimise the death toll, both countries need to prioritise individuals aged over 75 years. These trends were not changed by proportions of asymptomatic cases and varying vaccine efficacy on individuals under 20 years. Interpretation We presented the optimal vaccination strategy for Japan and South Korea. Comparing the results of these countries demonstrates that not only the effective contact rates containing q_i but also the age-demographics of current epidemic in Japan (dominance in 20s) and South Korea (dominant cases over 50s) affect vaccine allocation strategy.

Ancient DNA reveals that few GWAS loci have been strongly selected during recent human history

Genetic data from ancient humans has provided new evidence in the study of loci thought to be under historic selection, and thus is a powerful tool for identifying instances of selection that might be missed by methods that use present-day samples alone. Using a curated set of disease-associated variants from the NHGRI-EBI GWAS Catalog, we provide an analysis to identify disease-associated variants that bear signatures of selection over time. After accounting for the fact that not every ancient individual contributed equally to modern genomes, a Bayesian inference method was used to infer allele frequency trajectories over time and determine which disease-associated loci exhibit signatures of natural selection. Of the 2,709 variants analyzed in this study, 895 show at least a weak signature of selection (|s| > 0.001), including multiple variants that are introgressed from Neanderthals. However, only nine disease-associated variants show a signature of strong selection (|s| > 0.01). Additionally, we find that many risk-associated alleles have increased in frequency during the past 10,000 years. Overall, we find that disease-associated variants from GWAS are governed by nearly neutral evolution. Exceptions to this broad pattern include GWAS loci that protect against asthma and variants in MHC genes. Ancient samples allow us an unprecedented look at how our species has changed over time, and our results represent an important early step in using this new source of data to better understand the evolution of hereditary disease risks.

Drone phenotyping and machine learning enable discovery of loci regulating daily floral opening in lettuce

Flower opening and closure are traits of reproductive importance in all angiosperms because they determine the success of self- and cross-pollination. The temporal nature of this phenotype rendered it a difficult target for genetic studies. Cultivated and wild lettuce, Lactuca spp., have composite inflorescences that open only once. An L. serriola×L. sativa F6 recombinant inbred line (RIL) population differed markedly for daily floral opening time. This population was used to map the genetic determinants of this trait; the floral opening time of 236 RILs was scored using time-course image series obtained by drone-based phenotyping on two occasions. Floral pixels were identified from the images using a support vector machine with an accuracy >99%. A Bayesian inference method was developed to extract the peak floral opening time for individual genotypes from the time-stamped image data. Two independent quantitative trait loci (QTLs; Daily Floral Opening 2.1 and qDFO8.1) explaining >30% of the phenotypic variation in floral opening time were discovered. Candidate genes with non-synonymous polymorphisms in coding sequences were identified within the QTLs. This study demonstrates the power of combining remote sensing, machine learning, Bayesian statistics, and genome-wide marker data for studying the genetics of recalcitrant phenotypes.

Metal Cation-Binding Mechanisms of Q-Proline Peptoid Macrocycles in Solution

<div>The rational design of foldable and functionalizable peptidomimetic scaffolds requires the concerted application of both computational and experimental methods. Recently, a new class of designed peptoid macrocycle incorporating spiroligomer proline mimics (Q-prolines) has been found to pre-organize when bound by monovalent metal cations. To determine the solution-state structure of these cation-bound macrocycles, we employ a Bayesian inference method (BICePs) to reconcile enhanced-sampling molecular simulations with sparse ROESY correlations from experimental NMR studies. The BICePs approach circumvents the need for bespoke force field parameterization, instead relying on experimental restraints to help narrow the possible set of <i>cis</i>/<i>trans</i> amide isomers in solution. Conformations predicted to be most populated in solution were then simulated in the presence of explicit cations to yield trajectories with observed binding events, revealing a highly-preorganized all-<i>trans</i> amide conformation, whose formation is likely limited by the slow rate of <i>cis</i>/<i>trans</i> isomerization. Interestingly, this conformation differs from a racemic crystal structure solved in the absence of cation. Free energies of cation binding computed from distance-dependent potentials of mean force suggest Na⁺ has higher affinity to the macrocycle than K⁺, with both cations binding much more strongly in acetonitrile than water. The simulated affinities are able to correctly rank the extent to which different macrocycle sequences exhibit preorganization in the presence of different metal cations and solvents, suggesting our approach is suitable for solution-state computational design.</div>

Metal Cation-Binding Mechanisms of Q-Proline Peptoid Macrocycles in Solution

<div>The rational design of foldable and functionalizable peptidomimetic scaffolds requires the concerted application of both computational and experimental methods. Recently, a new class of designed peptoid macrocycle incorporating spiroligomer proline mimics (Q-prolines) has been found to pre-organize when bound by monovalent metal cations. To determine the solution-state structure of these cation-bound macrocycles, we employ a Bayesian inference method (BICePs) to reconcile enhanced-sampling molecular simulations with sparse ROESY correlations from experimental NMR studies. The BICePs approach circumvents the need for bespoke force field parameterization, instead relying on experimental restraints to help narrow the possible set of <i>cis</i>/<i>trans</i> amide isomers in solution. Conformations predicted to be most populated in solution were then simulated in the presence of explicit cations to yield trajectories with observed binding events, revealing a highly-preorganized all-<i>trans</i> amide conformation, whose formation is likely limited by the slow rate of <i>cis</i>/<i>trans</i> isomerization. Interestingly, this conformation differs from a racemic crystal structure solved in the absence of cation. Free energies of cation binding computed from distance-dependent potentials of mean force suggest Na⁺ has higher affinity to the macrocycle than K⁺, with both cations binding much more strongly in acetonitrile than water. The simulated affinities are able to correctly rank the extent to which different macrocycle sequences exhibit preorganization in the presence of different metal cations and solvents, suggesting our approach is suitable for solution-state computational design.</div>