Approximate Bayesian computation

Intrinsic dimension as a multi-scale summary statistics in network modeling

Complex networks are powerful mathematical tools for modelling and understanding the behaviour of highly interconnected systems. However, existing methods for analyzing these networks focus on local properties (e.g. degree distribution, clustering coefficient) or global properties (e.g. diameter, modularity) and fail to characterize the network structure across multiple scales. In this paper, we introduce a rigorous method for calculating the intrinsic dimension of unweighted networks. The intrinsic dimension is a feature that describes the network structure at all scales, from local to global. We propose using this measure as a summary statistic within an Approximate Bayesian Computation framework to infer the parameters of flexible and multi-purpose mechanistic models that generate complex networks. Furthermore, we present a new mechanistic model that can reproduce the intrinsic dimension of networks with large diameters, a task that has been challenging for existing models.

Modeling the risk of aquatic species invasion spread through boater movements and river connections

Aquatic invasive species (AIS) are one of the greatest threats to the functioning of aquatic ecosystems worldwide. Once an invasive species has been introduced to a new region, many governments develop management strategies to reduce further spread. Nevertheless, managing AIS in a new region is challenging because of the vast areas that need protection and limited resources. Spatial heterogeneity in invasion risk is driven by environmental suitability and propagule pressure, which can be used to prioritize locations for surveillance and intervention activities. To better understand invasion risk across aquatic landscapes, we developed a simulation model to estimate the likelihood of a waterbody becoming invaded with an AIS. The model included waterbodies connected via a multilayer network that included boater movements and hydrological connections. In a case study of Minnesota, we used zebra mussels (Dreissena polymorpha) and starry stonewort (Nitellopsis obtusa) as model species. We simulated the impacts of management scenarios developed by stakeholders and created a decision-support tool available through an online application provided as part of the AIS Explorer dashboard. Our baseline model revealed that 89% of new zebra mussel invasions and 84% of new starry stonewort invasions occurred through boater movements, establishing it as a primary pathway of spread and offering insights beyond risk estimates generated by traditional environmental suitability models alone. Our results highlight the critical role of interventions applied to boater movements to reduce AIS dispersal.

The virtual multiple sclerosis patient

Multiple sclerosis (MS) diagnosis typically involves assessing clinical symptoms, MRI findings, and ruling out alternative explanations. While myelin damage broadly affects conduction speeds, traditional tests focus on specific white-matter tracts, which may not reflect overall impairment accurately. In this study, we integrate diffusion tensor immaging (DTI) and magnetoencephalography (MEG) data into individualized virtual brain models to estimate conduction velocities for MS patients and controls. Using Bayesian inference, we demonstrated a causal link between empirical spectral changes and inferred slower conduction velocities in patients. Remarkably, these velocities proved superior predictors of clinical disability compared to structural damage. Our findings underscore a nuanced relationship between conduction delays and large-scale brain dynamics, suggesting that individualized velocity alterations at the whole-brain level contribute causatively to clinical outcomes in MS.

Mathematical models of coagulation-are we there yet?

BACKGROUND

Mathematical models of coagulation have been developed to mirror thrombin generation in plasma, with the aim of investigating how variation in coagulation factor levels regulates hemostasis. However, current models vary in the reactions they capture and the reaction rates used, and their validation is restricted by a lack of large coherent datasets, resulting in questioning of their utility.

OBJECTIVES

To address this debate, we systematically assessed current models against a large dataset, using plasma coagulation factor levels from 348 individuals with normal hemostasis to identify the causes of these variations.

METHODS

We compared model predictions with measured thrombin generation, quantifying and comparing the ability of each model to predict thrombin generation, the contributions of the individual reactions, and their dependence on reaction rates.

RESULTS

We found that no current model predicted the hemostatic response across the whole cohort and all produced thrombin generation curves that did not resemble those obtained experimentally. Our analysis has identified the key reactions that lead to differential model predictions, where experimental uncertainty leads to variability in predictions, and we determined reactions that have a high influence on measured thrombin generation, such as the contribution of factor XI.

CONCLUSION

This systematic assessment of models of coagulation, using large dataset inputs, points to ways in which these models can be improved. A model that accurately reflects the effects of the multiple subtle variations in an individual's hemostatic profile could be used for assessing antithrombotics or as a tool for precision medicine.

Integrating water quality data with a Bayesian network model to improve spatial and temporal phosphorus attribution: Application to the Maumee River Basin

Surface water nutrient pollution, the primary cause of eutrophication, remains a major environmental concern in Western Lake Erie despite intergovernmental efforts to regulate nutrient sources. The Maumee River Basin has been the largest nutrient contributor. The two primary nutrient sources are inorganic fertilizer and livestock manure applied to croplands, which are later carried to the streams via runoff and soil erosion. Prior studies of nutrient source attribution have focused on large watersheds or counties at annual time scales. Source attribution at finer spatiotemporal scales, which enables more effective nutrient management, remains a substantial challenge. This study aims to address this challenge by developing a generalizable Bayesian network model for phosphorus source attribution at the subwatershed scale (12-digit Hydrologic Unit Code). Since phosphorus release is uncertain, we combine excess phosphorus derived from manure and fertilizer application and crop uptake data, flow information simulated by the SWAT model, and in-stream water quality measurements using Approximate Bayesian Computation to derive a posterior that attributes phosphorus contributions to subwatersheds. Our results show significant variability in subwatershed-scale phosphorus release that is lost in coarse-scale attribution. Phosphorus contributions attributed to the subwatersheds are on average lower than the excess phosphorus estimated by the nutrient balance approach currently adopted by environmental agencies. Fertilizer contributes more soluble reactive phosphorus than manure, while manure contributes most of the unreactive phosphorus. While developed for the specific context of Maumee River Basin, our lightweight and generalizable model framework could be adapted to other regions and pollutants and could help inform targeted environmental regulation and enforcement.

A multiscale modeling framework for Scenario Modeling: Characterizing the heterogeneity of the COVID-19 epidemic in the US

The Scenario Modeling Hub (SMH) initiative provides projections of potential epidemic scenarios in the United States (US) by using a multi-model approach. Our contribution to the SMH is generated by a multiscale model that combines the global epidemic metapopulation modeling approach (GLEAM) with a local epidemic and mobility model of the US (LEAM-US), first introduced here. The LEAM-US model consists of 3142 subpopulations each representing a single county across the 50 US states and the District of Columbia, enabling us to project state and national trajectories of COVID-19 cases, hospitalizations, and deaths under different epidemic scenarios. The model is age-structured, and multi-strain. It integrates data on vaccine administration, human mobility, and non-pharmaceutical interventions. The model contributed to all 17 rounds of the SMH, and allows for the mechanistic characterization of the spatio-temporal heterogeneities observed during the COVID-19 pandemic. Here we describe the mathematical and computational structure of our model, and present the results concerning the emergence of the SARS-CoV-2 Alpha variant (lineage designation B.1.1.7) as a case study. Our findings show considerable spatial and temporal heterogeneity in the introduction and diffusion of the Alpha variant, both at the level of individual states and combined statistical areas, as it competes against the ancestral lineage. We discuss the key factors driving the time required for the Alpha variant to rise to dominance within a population, and quantify the impact that the emergence of the Alpha variant had on the effective reproduction number at the state level. Overall, we show that our multiscale modeling approach is able to capture the complexity and heterogeneity of the COVID-19 pandemic response in the US.

Real-Time Use of a Dynamic Model To Measure the Impact of Public Health Interventions on Measles Outbreak Size and Duration - Chicago, Illinois, 2024

Measles is a highly infectious, vaccine-preventable disease that can cause severe illness, hospitalization, and death. A measles outbreak associated with a migrant shelter in Chicago occurred during February-April 2024, in which a total of 57 confirmed cases were identified, including 52 among shelter residents, three among staff members, and two among community members with a known link to the shelter. CDC simulated a measles outbreak among shelter residents using a dynamic disease model, updated in real time as additional cases were identified, to produce outbreak forecasts and assess the impact of public health interventions. As of April 8, the model forecasted a median final outbreak size of 58 cases (IQR = 56-60 cases); model fit and prediction range improved as more case data became available. Counterfactual analysis of different intervention scenarios demonstrated the importance of early deployment of public health interventions in Chicago, with a 69% chance of an outbreak of 100 or more cases had there been no mass vaccination or active case-finding compared with only a 1% chance when those interventions were deployed. This analysis highlights the value of using real-time, dynamic models to aid public health response, set expectations about outbreak size and duration, and quantify the impact of interventions. The model shows that prompt mass vaccination and active case-finding likely substantially reduced the chance of a large (100 or more cases) outbreak in Chicago.

Importance of social inequalities to contact patterns, vaccine uptake, and epidemic dynamics

Individuals' socio-demographic and economic characteristics crucially shape the spread of an epidemic by largely determining the exposure level to the virus and the severity of the disease for those who got infected. While the complex interplay between individual characteristics and epidemic dynamics is widely recognised, traditional mathematical models often overlook these factors. In this study, we examine two important aspects of human behaviour relevant to epidemics: contact patterns and vaccination uptake. Using data collected during the COVID-19 pandemic in Hungary, we first identify the dimensions along which individuals exhibit the greatest variation in their contact patterns and vaccination uptake. We find that generally higher socio-economic groups of the population have a higher number of contacts and a higher vaccination uptake with respect to disadvantaged groups. Subsequently, we propose a data-driven epidemiological model that incorporates these behavioural differences. Finally, we apply our model to analyse the fourth wave of COVID-19 in Hungary, providing valuable insights into real-world scenarios. By bridging the gap between individual characteristics and epidemic spread, our research contributes to a more comprehensive understanding of disease dynamics and informs effective public health strategies.

Artificial neural networks for model identification and parameter estimation in computational cognitive models

Computational cognitive models have been used extensively to formalize cognitive processes. Model parameters offer a simple way to quantify individual differences in how humans process information. Similarly, model comparison allows researchers to identify which theories, embedded in different models, provide the best accounts of the data. Cognitive modeling uses statistical tools to quantitatively relate models to data that often rely on computing/estimating the likelihood of the data under the model. However, this likelihood is computationally intractable for a substantial number of models. These relevant models may embody reasonable theories of cognition, but are often under-explored due to the limited range of tools available to relate them to data. We contribute to filling this gap in a simple way using artificial neural networks (ANNs) to map data directly onto model identity and parameters, bypassing the likelihood estimation. We test our instantiation of an ANN as a cognitive model fitting tool on classes of cognitive models with strong inter-trial dependencies (such as reinforcement learning models), which offer unique challenges to most methods. We show that we can adequately perform both parameter estimation and model identification using our ANN approach, including for models that cannot be fit using traditional likelihood-based methods. We further discuss our work in the context of the ongoing research leveraging simulation-based approaches to parameter estimation and model identification, and how these approaches broaden the class of cognitive models researchers can quantitatively investigate.

Artificial neural networks for model identification and parameter estimation in computational cognitive models

Computational cognitive models have been used extensively to formalize cognitive processes. Model parameters offer a simple way to quantify individual differences in how humans process information. Similarly, model comparison allows researchers to identify which theories, embedded in different models, provide the best accounts of the data. Cognitive modeling uses statistical tools to quantitatively relate models to data that often rely on computing/estimating the likelihood of the data under the model. However, this likelihood is computationally intractable for a substantial number of models. These relevant models may embody reasonable theories of cognition, but are often under-explored due to the limited range of tools available to relate them to data. We contribute to filling this gap in a simple way using artificial neural networks (ANNs) to map data directly onto model identity and parameters, bypassing the likelihood estimation. We test our instantiation of an ANN as a cognitive model fitting tool on classes of cognitive models with strong inter-trial dependencies (such as reinforcement learning models), which offer unique challenges to most methods. We show that we can adequately perform both parameter estimation and model identification using our ANN approach, including for models that cannot be fit using traditional likelihood-based methods. We further discuss our work in the context of the ongoing research leveraging simulation-based approaches to parameter estimation and model identification, and how these approaches broaden the class of cognitive models researchers can quantitatively investigate.

Estimating actual SARS-CoV-2 infections from secondary data

Eminent in pandemic management is accurate information on infection dynamics to plan for timely installation of control measures and vaccination campaigns. Despite huge efforts in diagnostic testing of individuals, the underestimation of the actual number of SARS-CoV-2 infections remains significant due to the large number of undocumented cases. In this paper we demonstrate and compare three methods to estimate the dynamics of true infections based on secondary data i.e., (a) test positivity, (b) infection fatality and (c) wastewater monitoring. The concept is tested with Austrian data on a national basis for the period of April 2020 to December 2022. Further, we use the results of prevalence studies from the same period to generate (upper and lower bounds of) credible intervals for true infections for four data points. Model parameters are subsequently estimated by applying Approximate Bayesian Computation-rejection sampling and Genetic Algorithms. The method is then validated for the case study Vienna. We find that all three methods yield fairly similar results for estimating the true number of infections, which supports the idea that all three datasets contain similar baseline information. None of them is considered superior, as their advantages and shortcomings depend on the specific case study at hand.

Selection among site-dependent structurally constrained substitution models of protein evolution by approximate Bayesian computation

MOTIVATION

The selection among substitution models of molecular evolution is fundamental for obtaining accurate phylogenetic inferences. At the protein level, evolutionary analyses are traditionally based on empirical substitution models but these models make unrealistic assumptions and are being surpassed by structurally constrained substitution (SCS) models. The SCS models often consider site-dependent evolution, a process that provides realism but complicates their implementation into likelihood functions that are commonly used for substitution model selection.

RESULTS

We present a method to perform selection among site-dependent SCS models, also among empirical and site-dependent SCS models, based on the approximate Bayesian computation (ABC) approach and its implementation into the computational framework ProteinModelerABC. The framework implements ABC with and without regression adjustments and includes diverse empirical and site-dependent SCS models of protein evolution. Using extensive simulated data, we found that it provides selection among SCS and empirical models with acceptable accuracy. As illustrative examples, we applied the framework to analyze a variety of protein families observing that SCS models fit them better than the corresponding best-fitting empirical substitution models.

AVAILABILITY AND IMPLEMENTATION

ProteinModelerABC is freely available from https://github.com/DavidFerreiro/ProteinModelerABC, can run in parallel and includes a graphical user interface. The framework is distributed with detailed documentation and ready-to-use examples.

Unlocking ensemble ecosystem modelling for large and complex networks

The potential effects of conservation actions on threatened species can be predicted using ensemble ecosystem models by forecasting populations with and without intervention. These model ensembles commonly assume stable coexistence of species in the absence of available data. However, existing ensemble-generation methods become computationally inefficient as the size of the ecosystem network increases, preventing larger networks from being studied. We present a novel sequential Monte Carlo sampling approach for ensemble generation that is orders of magnitude faster than existing approaches. We demonstrate that the methods produce equivalent parameter inferences, model predictions, and tightly constrained parameter combinations using a novel sensitivity analysis method. For one case study, we demonstrate a speed-up from 108 days to 6 hours, while maintaining equivalent ensembles. Additionally, we demonstrate how to identify the parameter combinations that strongly drive feasibility and stability, drawing ecological insight from the ensembles. Now, for the first time, larger and more realistic networks can be practically simulated and analysed.

Parameter identifiability and model selection for partial differential equation models of cell invasion

When employing mechanistic models to study biological phenomena, practical parameter identifiability is important for making accurate predictions across wide ranges of unseen scenarios, as well as for understanding the underlying mechanisms. In this work, we use a profile-likelihood approach to investigate parameter identifiability for four extensions of the Fisher-Kolmogorov-Petrovsky-Piskunov (Fisher-KPP) model, given experimental data from a cell invasion assay. We show that more complicated models tend to be less identifiable, with parameter estimates being more sensitive to subtle differences in experimental procedures, and that they require more data to be practically identifiable. As a result, we suggest that parameter identifiability should be considered alongside goodness-of-fit and model complexity as criteria for model selection.

Calibration of agent based models for monophasic and biphasic tumour growth using approximate Bayesian computation

Agent-based models (ABMs) are readily used to capture the stochasticity in tumour evolution; however, these models are often challenging to validate with experimental measurements due to model complexity. The Voronoi cell-based model (VCBM) is an off-lattice agent-based model that captures individual cell shapes using a Voronoi tessellation and mimics the evolution of cancer cell proliferation and movement. Evidence suggests tumours can exhibit biphasic growth in vivo. To account for this phenomena, we extend the VCBM to capture the existence of two distinct growth phases. Prior work primarily focused on point estimation for the parameters without consideration of estimating uncertainty. In this paper, approximate Bayesian computation is employed to calibrate the model to in vivo measurements of breast, ovarian and pancreatic cancer. Our approach involves estimating the distribution of parameters that govern cancer cell proliferation and recovering outputs that match the experimental data. Our results show that the VCBM, and its biphasic extension, provides insight into tumour growth and quantifies uncertainty in the switching time between the two phases of the biphasic growth model. We find this approach enables precise estimates for the time taken for a daughter cell to become a mature cell. This allows us to propose future refinements to the model to improve accuracy, whilst also making conclusions about the differences in cancer cell characteristics.

Tuning the potency and selectivity of ImmTAC molecules by affinity modulation

T-cell-engaging bispecifics have great clinical potential for the treatment of cancer and infectious diseases. The binding affinity and kinetics of a bispecific molecule for both target and T-cell CD3 have substantial effects on potency and specificity, but the rules governing these relationships are not fully understood. Using immune mobilizing monoclonal TCRs against cancer (ImmTAC) molecules as a model, we explored the impact of altering affinity for target and CD3 on the potency and specificity of the redirected T-cell response. This class of bispecifics binds specific target peptides presented by human leukocyte antigen on the cell surface via an affinity-enhanced T-cell receptor and can redirect T-cell activation with an anti-CD3 effector moiety. The data reveal that combining a strong affinity TCR with an intermediate affinity anti-CD3 results in optimal T-cell activation, while strong affinity of both targeting and effector domains significantly reduces maximum cytokine release. Moreover, by optimizing the affinity of both parts of the molecule, it is possible to improve the selectivity. These results could be effectively modelled based on kinetic proofreading with limited signalling. This model explained the experimental observation that strong binding at both ends of the molecules leads to reduced activity, through very stable target-bispecific-effector complexes leading to CD3 entering a non-signalling dark state. These findings have important implications for the design of anti-CD3-based bispecifics with optimal biophysical parameters for both activity and specificity.

Moana: Alternate surveillance for COVID-19 in a Unique Population (MASC-UP)

Objective

Create a longitudinal, multi-modal and multi-level surveillance cohort that targets early detection of symptomatic and asymptomatic COVID-19 cases among Native Hawaiian and Pacific Islander adults in the Continental US and identify effective modalities for participatory disease surveillance and sustainably integrate them into ongoing COVID-19 and other public health surveillance efforts.

Materials and methods

We recruited cohorts from three sites: Federal Way, WA; Springdale, AR; and remotely. Participants received a survey that included demographic characteristics and questions regarding COVID-19. Participants completed symptom checks via text message every month and recorded their temperature daily using a Kinsa smart thermometer.

Results

Recruitment and data collection is ongoing. Presently, 441 adults have consented to participate. One-third of participants were classified as essential workers during the pandemic.

Discussion

Over the past 18 months, we have improved our strategies to elicit better data from participants and have learned from some of the weaknesses in our initial deployment of this type of surveillance system. Other limitations stem from historic inequities and barriers which limited Native Hawaiian and Pacific Island representation in academic and clinical environments. One manifestation of this was the limited ability to provide study materials and support in multiple languages. We hope that continued partnership with the community will allow further opportunities to help restore trust in academic and medical institutions, thus generating knowledge to advance health equity.

Conclusion

This participatory disease surveillance mechanism complements traditional surveillance systems by engaging underserved communities. We may also gain insights generalizable to other pathogens of concern.

Extrapolating Metal (Cu, Ni, Zn) Toxicity from Individuals to Populations Across Daphnia Species Using Mechanistic Models: The Roles of Uncertainty Propagation and Combined Physiological Modes of Action

Mechanistic effect modeling is a promising tool to improve the ecological realism of environmental risk assessment. An open question for the mechanistic modeling of metal toxicity is whether the same physiological mode of action (PMoA) could be assumed for closely related species. The implications of various modeling choices, such as the use of parameter point estimates and assumption of simplistic toxicodynamic models, are largely unexplored. We conducted life-table experiments with Daphnia longispina, Daphnia magna, and Daphnia pulex exposed to the single metals Cu, Ni, and Zn, and calibrated toxicokinetic-toxicodynamic (TKTD) models based on dynamic energy budget theory. We developed TKTD models with single and combined PMoAs to compare their goodness-of-fit and predicted population-level sensitivity. We identified the PMoA reproduction efficiency as most probable in all species for Ni and Zn, but not for Cu, and found that combined-PMoA models predicted higher population-level sensitivity than single-PMoA models, which was related to the predicted individual-level sensitivity, rather than to mechanistic differences between models. Using point estimates of parameters, instead of sampling from the probability distributions of parameters, could also lead to differences in the predicted population-level sensitivity. According to model predictions, apical chronic endpoints (cumulative reproduction, survival) are conservative for single-metal population effects across metals and species. We conclude that the assumption of an identical PMoA for different species of Daphnia could be justified for Ni and Zn, but not for Cu. Single-PMoA models are more appropriate than combined-PMoA models from a model selection perspective, but propagation of the associated uncertainty should be considered. More accurate predictions of effects at low concentrations may nevertheless motivate the use of combined-PMoA models. Environ Toxicol Chem 2024;43:338-358. © 2023 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.

Model of methane transport in tree stems: Case study of sap flow and radial diffusion

The transport processes of methane (CH4 ) in tree stems remain largely unknown, although they are critical in assessing the whole-forest CH4 dynamics. We used a physically based dynamic model to study the spatial and diurnal dynamics of stem CH4 transport and fluxes. We parameterised the model using data from laboratory experiments with Pinus sylvestris and Betula pendula and compared the model to experimental data from a field study. Stem CH4 flux in laboratory and field conditions were explained by the axial advective CH4 transport from soil with xylem sap flow and the radial CH4 diffusion through the stem conditions. Diffusion resistance caused by the bark permeability did not significantly affect gas transport or stem CH4 flux in the laboratory experiments. The role of axial diffusion of CH4 in trees was unresolved and requires further studies. Due to the transit time of CH4 in the stem, the diurnal dynamics of stem CH4 fluxes can deviate markedly from the diurnal dynamics of sap flow.

Global stability and parameter analysis reinforce therapeutic targets of PD-L1-PD-1 and MDSCs for glioblastoma

Glioblastoma (GBM) is an aggressive primary brain cancer that currently has minimally effective treatments. Like other cancers, immunosuppression by the PD-L1-PD-1 immune checkpoint complex is a prominent axis by which glioma cells evade the immune system. Myeloid-derived suppressor cells (MDSCs), which are recruited to the glioma microenviroment, also contribute to the immunosuppressed GBM microenvironment by suppressing T cell functions. In this paper, we propose a GBM-specific tumor-immune ordinary differential equations model of glioma cells, T cells, and MDSCs to provide theoretical insights into the interactions between these cells. Equilibrium and stability analysis indicates that there are unique tumorous and tumor-free equilibria which are locally stable under certain conditions. Further, the tumor-free equilibrium is globally stable when T cell activation and the tumor kill rate by T cells overcome tumor growth, T cell inhibition by PD-L1-PD-1 and MDSCs, and the T cell death rate. Bifurcation analysis suggests that a treatment plan that includes surgical resection and therapeutics targeting immune suppression caused by the PD-L1-PD1 complex and MDSCs results in the system tending to the tumor-free equilibrium. Using a set of preclinical experimental data, we implement the approximate Bayesian computation (ABC) rejection method to construct probability density distributions that estimate model parameters. These distributions inform an appropriate search curve for global sensitivity analysis using the extended fourier amplitude sensitivity test. Sensitivity results combined with the ABC method suggest that parameter interaction is occurring between the drivers of tumor burden, which are the tumor growth rate and carrying capacity as well as the tumor kill rate by T cells, and the two modeled forms of immunosuppression, PD-L1-PD-1 immune checkpoint and MDSC suppression of T cells. Thus, treatment with an immune checkpoint inhibitor in combination with a therapeutic targeting the inhibitory mechanisms of MDSCs should be explored.

Effective population size of Culex quinquefasciatus under insecticide-based vector management and following Hurricane Harvey in Harris County, Texas

Introduction: Culex quinquefasciatus is a mosquito species of significant public health importance due to its ability to transmit multiple pathogens that can cause mosquito-borne diseases, such as West Nile fever and St. Louis encephalitis. In Harris County, Texas, Cx. quinquefasciatus is a common vector species and is subjected to insecticide-based management by the Harris County Public Health Department. However, insecticide resistance in mosquitoes has increased rapidly worldwide and raises concerns about maintaining the effectiveness of vector control approaches. This concern is highly relevant in Texas, with its humid subtropical climate along the Gulf Coast that provides suitable habitat for Cx. quinquefasciatus and other mosquito species that are known disease vectors. Therefore, there is an urgent and ongoing need to monitor the effectiveness of current vector control programs. Methods: In this study, we evaluated the impact of vector control approaches by estimating the effective population size of Cx. quinquefasciatus in Harris County. We applied Approximate Bayesian Computation to microsatellite data to estimate effective population size. We collected Cx. quinquefasciatus samples from two mosquito control operation areas; 415 and 802, during routine vector monitoring in 2016 and 2017. No county mosquito control operations were applied at area 415 in 2016 and 2017, whereas extensive adulticide spraying operations were in effect at area 802 during the summer of 2016. We collected data for eighteen microsatellite markers for 713 and 723 mosquitoes at eight timepoints from 2016 to 2017 in areas 415 and 802, respectively. We also investigated the impact of Hurricane Harvey's landfall in the Houston area in August of 2017 on Cx. quinquefasciatus population fluctuation. Results: We found that the bottleneck scenario was the most probable historical scenario describing the impact of the winter season at area 415 and area 802, with the highest posterior probability of 0.9167 and 0.4966, respectively. We also detected an expansion event following Hurricane Harvey at area 802, showing a 3.03-fold increase in 2017. Discussion: Although we did not detect significant effects of vector control interventions, we found considerable influences of the winter season and a major hurricane on the effective population size of Cx. quinquefasciatus. The fluctuations in effective population size in both areas showed a significant seasonal pattern. Additionally, the significant population expansion following Hurricane Harvey in 2017 supports the necessity for post-hurricane vector-control interventions.

An integrated model for pre- and post-harvest aflatoxin contamination in maize

Aflatoxin contamination caused by colonization of maize by Aspergillus flavus continues to pose a major human and livestock health hazard in the food chain. Increasing attention has been focused on the development of models to predict risk and to identify effective intervention strategies. Most risk prediction models have focused on elucidating weather and site variables on the pre-harvest dynamics of A. flavus growth and aflatoxin production. However fungal growth and toxin accumulation continue to occur after harvest, especially in countries where storage conditions are limited by logistical and cost constraints. In this paper, building on previous work, we introduce and test an integrated meteorology-driven epidemiological model that covers the entire supply chain from planting to delivery. We parameterise the model using approximate Bayesian computation with monthly time-series data over six years for contamination levels of aflatoxin in daily shipments received from up to three sourcing regions at a high-volume maize processing plant in South Central India. The time series for aflatoxin levels from the parameterised model successfully replicated the overall profile, scale and variance of the historical aflatoxin datasets used for fitting and validation. We use the model to illustrate the dynamics of A. flavus growth and aflatoxin production during the pre- and post-harvest phases in different sourcing regions, in short-term predictions to inform decision making about sourcing supplies and to compare intervention strategies to reduce the risks of aflatoxin contamination.

Continuous time causal structure induction with prevention and generation

Most research into causal learning has focused on atemporal contingency data settings while fewer studies have examined learning and reasoning about systems exhibiting events that unfold in continuous time. Of these, none have yet explored learning about preventative causal influences. How do people use temporal information to infer which components of a causal system are generating or preventing activity of other components? In what ways do generative and preventative causes interact in shaping the behavior of causal mechanisms and their learnability? We explore human causal structure learning within a space of hypotheses that combine generative and preventative causal relationships. Participants observe the behavior of causal devices as they are perturbed by fixed interventions and subject to either regular or irregular spontaneous activations. We find that participants are capable learners in this setting, successfully identifying the large majority of generative, preventative and non-causal relationships but making certain attribution errors. We lay out a computational-level framework for normative inference in this setting and propose a family of more cognitively plausible algorithmic approximations. We find that participants' judgment patterns can be both qualitatively and quantitatively captured by a model that approximates normative inference via a simulation and summary statistics scheme based on structurally local computation using temporally local evidence.

The influence of inter-regional delays in generating large-scale brain networks of phase synchronization

Large-scale networks of phase synchronization are considered to regulate the communication between brain regions fundamental to cognitive function, but the mapping to their structural substrates, i.e., the structure-function relationship, remains poorly understood. Biophysical Network Models (BNMs) have demonstrated the influences of local oscillatory activity and inter-regional anatomical connections in generating alpha-band (8-12 Hz) networks of phase synchronization observed with Electroencephalography (EEG) and Magnetoencephalography (MEG). Yet, the influence of inter-regional conduction delays remains unknown. In this study, we compared a BNM with standard "distance-dependent delays", which assumes constant conduction velocity, to BNMs with delays specified by two alternative methods accounting for spatially varying conduction velocities, "isochronous delays" and "mixed delays". We followed the Approximate Bayesian Computation (ABC) workflow, i) specifying neurophysiologically informed prior distributions of BNM parameters, ii) verifying the suitability of the prior distributions with Prior Predictive Checks, iii) fitting each of the three BNMs to alpha-band MEG resting-state data (N = 75) with Bayesian optimization for Likelihood-Free Inference (BOLFI), and iv) choosing between the fitted BNMs with ABC model comparison on a separate MEG dataset (N = 30). Prior Predictive Checks revealed the range of dynamics generated by each of the BNMs to encompass those seen in the MEG data, suggesting the suitability of the prior distributions. Fitting the models to MEG data yielded reliable posterior distributions of the parameters of each of the BNMs. Finally, model comparison revealed the BNM with "distance-dependent delays", as the most probable to describe the generation of alpha-band networks of phase synchronization seen in MEG. These findings suggest that distance-dependent delays might contribute to the neocortical architecture of human alpha-band networks of phase synchronization. Hence, our study illuminates the role of inter-regional delays in generating the large-scale networks of phase synchronization that might subserve the communication between regions vital to cognition.

PyMC: a modern, and comprehensive probabilistic programming framework in Python

PyMC is a probabilistic programming library for Python that provides tools for constructing and fitting Bayesian models. It offers an intuitive, readable syntax that is close to the natural syntax statisticians use to describe models. PyMC leverages the symbolic computation library PyTensor, allowing it to be compiled into a variety of computational backends, such as C, JAX, and Numba, which in turn offer access to different computational architectures including CPU, GPU, and TPU. Being a general modeling framework, PyMC supports a variety of models including generalized hierarchical linear regression and classification, time series, ordinary differential equations (ODEs), and non-parametric models such as Gaussian processes (GPs). We demonstrate PyMC's versatility and ease of use with examples spanning a range of common statistical models. Additionally, we discuss the positive role of PyMC in the development of the open-source ecosystem for probabilistic programming.

Profile-Wise Analysis: A profile likelihood-based workflow for identifiability analysis, estimation, and prediction with mechanistic mathematical models

Interpreting data using mechanistic mathematical models provides a foundation for discovery and decision-making in all areas of science and engineering. Developing mechanistic insight by combining mathematical models and experimental data is especially critical in mathematical biology as new data and new types of data are collected and reported. Key steps in using mechanistic mathematical models to interpret data include: (i) identifiability analysis; (ii) parameter estimation; and (iii) model prediction. Here we present a systematic, computationally-efficient workflow we call Profile-Wise Analysis (PWA) that addresses all three steps in a unified way. Recently-developed methods for constructing 'profile-wise' prediction intervals enable this workflow and provide the central linkage between different workflow components. These methods propagate profile-likelihood-based confidence sets for model parameters to predictions in a way that isolates how different parameter combinations affect model predictions. We show how to extend these profile-wise prediction intervals to two-dimensional interest parameters. We then demonstrate how to combine profile-wise prediction confidence sets to give an overall prediction confidence set that approximates the full likelihood-based prediction confidence set well. Our three case studies illustrate practical aspects of the workflow, focusing on ordinary differential equation (ODE) mechanistic models with both Gaussian and non-Gaussian noise models. While the case studies focus on ODE-based models, the workflow applies to other classes of mathematical models, including partial differential equations and simulation-based stochastic models. Open-source software on GitHub can be used to replicate the case studies.

An in silico and in vitro human neuronal network model reveals cellular mechanisms beyond NaV1.1 underlying Dravet syndrome

Human induced pluripotent stem cell (hiPSC)-derived neuronal networks on multi-electrode arrays (MEAs) provide a unique phenotyping tool to study neurological disorders. However, it is difficult to infer cellular mechanisms underlying these phenotypes. Computational modeling can utilize the rich dataset generated by MEAs, and advance understanding of disease mechanisms. However, existing models lack biophysical detail, or validation and calibration to relevant experimental data. We developed a biophysical in silico model that accurately simulates healthy neuronal networks on MEAs. To demonstrate the potential of our model, we studied neuronal networks derived from a Dravet syndrome (DS) patient with a missense mutation in SCN1A, encoding sodium channel NaV1.1. Our in silico model revealed that sodium channel dysfunctions were insufficient to replicate the in vitro DS phenotype, and predicted decreased slow afterhyperpolarization and synaptic strengths. We verified these changes in DS patient-derived neurons, demonstrating the utility of our in silico model to predict disease mechanisms.

Statistical methods to identify mechanisms in studies of eco-evolutionary dynamics

While the reciprocal effects of ecological and evolutionary dynamics are increasingly recognized as an important driver for biodiversity, detection of such eco-evolutionary feedbacks, their underlying mechanisms, and their consequences remains challenging. Eco-evolutionary dynamics occur at different spatial and temporal scales and can leave signatures at different levels of organization (e.g., gene, protein, trait, community) that are often difficult to detect. Recent advances in statistical methods combined with alternative hypothesis testing provides a promising approach to identify potential eco-evolutionary drivers for observed data even in non-model systems that are not amenable to experimental manipulation. We discuss recent advances in eco-evolutionary modeling and statistical methods and discuss challenges for fitting mechanistic models to eco-evolutionary data.

Growth exponents reflect evolutionary processes and treatment response in brain metastases

Tumor growth is the result of the interplay of complex biological processes in huge numbers of individual cells living in changing environments. Effective simple mathematical laws have been shown to describe tumor growth in vitro, or simple animal models with bounded-growth dynamics accurately. However, results for the growth of human cancers in patients are scarce. Our study mined a large dataset of 1133 brain metastases (BMs) with longitudinal imaging follow-up to find growth laws for untreated BMs and recurrent treated BMs. Untreated BMs showed high growth exponents, most likely related to the underlying evolutionary dynamics, with experimental tumors in mice resembling accurately the disease. Recurrent BMs growth exponents were smaller, most probably due to a reduction in tumor heterogeneity after treatment, which may limit the tumor evolutionary capabilities. In silico simulations using a stochastic discrete mesoscopic model with basic evolutionary dynamics led to results in line with the observed data.

Socioeconomic determinants of stay-at-home policies during the first COVID-19 wave

Introduction

The COVID-19 pandemic has had a significant impact on public health and social systems worldwide. This study aims to evaluate the efficacy of various policies and restrictions implemented by different countries to control the spread of the virus.

Methods

To achieve this objective, a compartmental model is used to quantify the "social permeability" of a population, which reflects the inability of individuals to remain in confinement and continue social mixing allowing the spread of the virus. The model is calibrated to fit and recreate the dynamics of the epidemic spreading of 42 countries, mainly taking into account reported deaths and mobility across the populations.

Results

The results indicate that low-income countries have a harder time slowing the advance of the pandemic, even if the virus did not initially propagate as fast as in wealthier countries, showing the disparities between countries in their ability to mitigate the spread of the disease and its impact on vulnerable populations.

Discussion

This research contributes to a better understanding of the socioeconomic and environmental factors that affect the spread of the virus and the need for equitable policy measures to address the disparities in the global response to the pandemic.

Implications of different membrane compartmentalization models in particle-based in silico studies

Studying membrane dynamics is important to understand the cellular response to environmental stimuli. A decisive spatial characteristic of the plasma membrane is its compartmental structure created by the actin-based membrane-skeleton (fences) and anchored transmembrane proteins (pickets). Particle-based reaction-diffusion simulation of the membrane offers a suitable temporal and spatial resolution to analyse its spatially heterogeneous and stochastic dynamics. Fences have been modelled via hop probabilities, potentials or explicit picket fences. Our study analyses the different approaches' constraints and their impact on simulation results and performance. Each of the methods comes with its own constraints; the picket fences require small timesteps, potential fences might induce a bias in diffusion in crowded systems, and probabilistic fences, in addition to carefully scaling the probability with the timesteps, induce higher computational costs for each propagation step.

Mutation rate, selection, and epistasis inferred from RNA virus haplotypes via neural posterior estimation

RNA viruses are particularly notorious for their high levels of genetic diversity, which is generated through the forces of mutation and natural selection. However, disentangling these two forces is a considerable challenge, and this may lead to widely divergent estimates of viral mutation rates, as well as difficulties in inferring the fitness effects of mutations. Here, we develop, test, and apply an approach aimed at inferring the mutation rate and key parameters that govern natural selection, from haplotype sequences covering full-length genomes of an evolving virus population. Our approach employs neural posterior estimation, a computational technique that applies simulation-based inference with neural networks to jointly infer multiple model parameters. We first tested our approach on synthetic data simulated using different mutation rates and selection parameters while accounting for sequencing errors. Reassuringly, the inferred parameter estimates were accurate and unbiased. We then applied our approach to haplotype sequencing data from a serial passaging experiment with the MS2 bacteriophage, a virus that parasites Escherichia coli. We estimated that the mutation rate of this phage is around 0.2 mutations per genome per replication cycle (95% highest density interval: 0.051-0.56). We validated this finding with two different approaches based on single-locus models that gave similar estimates but with much broader posterior distributions. Furthermore, we found evidence for reciprocal sign epistasis between four strongly beneficial mutations that all reside in an RNA stem loop that controls the expression of the viral lysis protein, responsible for lysing host cells and viral egress. We surmise that there is a fine balance between over- and underexpression of lysis that leads to this pattern of epistasis. To recap, we have developed an approach for joint inference of the mutation rate and selection parameters from full haplotype data with sequencing errors and used it to reveal features governing MS2 evolution.

Heterogeneity is a key factor describing the initial outbreak of COVID-19

Assessing the transmission potential of emerging infectious diseases, such as COVID-19, is crucial for implementing prompt and effective intervention policies. The basic reproduction number is widely used to measure the severity of the early stages of disease outbreaks. The basic reproduction number of standard ordinary differential equation models is computed for homogeneous contact patterns; however, realistic contact patterns are far from homogeneous, specifically during the early stages of disease transmission. Heterogeneity of contact patterns can lead to superspreading events that show a significantly high level of heterogeneity in generating secondary infections. This is primarily due to the large variance in the contact patterns of complex human behaviours. Hence, in this work, we investigate the impacts of heterogeneity in contact patterns on the basic reproduction number by developing two distinct model frameworks: 1) an SEIR-Erlang ordinary differential equation model and 2) an SEIR stochastic agent-based model. Furthermore, we estimated the transmission probability of both models in the context of COVID-19 in South Korea. Our results highlighted the importance of heterogeneity in contact patterns and indicated that there should be more information than one quantity (the basic reproduction number as the mean quantity), such as a degree-specific basic reproduction number in the distributional sense when the contact pattern is highly heterogeneous.

Inference of glioblastoma migration and proliferation rates using single time-point images

Cancer cell migration is a driving mechanism of invasion in solid malignant tumors. Anti-migratory treatments provide an alternative approach for managing disease progression. However, we currently lack scalable screening methods for identifying novel anti-migratory drugs. To this end, we develop a method that can estimate cell motility from single end-point images in vitro by estimating differences in the spatial distribution of cells and inferring proliferation and diffusion parameters using agent-based modeling and approximate Bayesian computation. To test the power of our method, we use it to investigate drug responses in a collection of 41 patient-derived glioblastoma cell cultures, identifying migration-associated pathways and drugs with potent anti-migratory effects. We validate our method and result in both in silico and in vitro using time-lapse imaging. Our proposed method applies to standard drug screen experiments, with no change needed, and emerges as a scalable approach to screen for anti-migratory drugs.

Modeling needs user modeling

Modeling has actively tried to take the human out of the loop, originally for objectivity and recently also for automation. We argue that an unnecessary side effect has been that modeling workflows and machine learning pipelines have become restricted to only well-specified problems. Putting the humans back into the models would enable modeling a broader set of problems, through iterative modeling processes in which AI can offer collaborative assistance. However, this requires advances in how we scope our modeling problems, and in the user models. In this perspective article, we characterize the required user models and the challenges ahead for realizing this vision, which would enable new interactive modeling workflows, and human-centric or human-compatible machine learning pipelines.

Revisiting the historical scenario of a disease dissemination using genetic data and Approximate Bayesian Computation methodology: The case of Pseudocercospora fijiensis invasion in Africa

The reconstruction of geographic and demographic scenarios of dissemination for invasive pathogens of crops is a key step toward improving the management of emerging infectious diseases. Nowadays, the reconstruction of biological invasions typically uses the information of both genetic and historical information to test for different hypotheses of colonization. The Approximate Bayesian Computation framework and its recent Random Forest development (ABC-RF) have been successfully used in evolutionary biology to decipher multiple histories of biological invasions. Yet, for some organisms, typically plant pathogens, historical data may not be reliable notably because of the difficulty to identify the organism and the delay between the introduction and the first mention. We investigated the history of the invasion of Africa by the fungal pathogen of banana Pseudocercospora fijiensis, by testing the historical hypothesis against other plausible hypotheses. We analyzed the genetic structure of eight populations from six eastern and western African countries, using 20 microsatellite markers and tested competing scenarios of population foundation using the ABC-RF methodology. We do find evidence for an invasion front consistent with the historical hypothesis, but also for the existence of another front never mentioned in historical records. We question the historical introduction point of the disease on the continent. Crucially, our results illustrate that even if ABC-RF inferences may sometimes fail to infer a single, well-supported scenario of invasion, they can be helpful in rejecting unlikely scenarios, which can prove much useful to shed light on disease dissemination routes.

Uncertainty quantification in mechanistic epidemic models via cross-entropy approximate Bayesian computation

This paper proposes a data-driven approximate Bayesian computation framework for parameter estimation and uncertainty quantification of epidemic models, which incorporates two novelties: (i) the identification of the initial conditions by using plausible dynamic states that are compatible with observational data; (ii) learning of an informative prior distribution for the model parameters via the cross-entropy method. The new methodology's effectiveness is illustrated with the aid of actual data from the COVID-19 epidemic in Rio de Janeiro city in Brazil, employing an ordinary differential equation-based model with a generalized SEIR mechanistic structure that includes time-dependent transmission rate, asymptomatics, and hospitalizations. A minimization problem with two cost terms (number of hospitalizations and deaths) is formulated, and twelve parameters are identified. The calibrated model provides a consistent description of the available data, able to extrapolate forecasts over a few weeks, making the proposed methodology very appealing for real-time epidemic modeling.

Estimation of a Within-Herd Transmission Rate for African Swine Fever in Vietnam

We describe results from a panel study in which pigs from a 17-sow African swine fever (ASF) positive herd in Thái Bình province, Vietnam, were followed over time to record the date of onset of ASF signs and the date of death from ASF. Our objectives were to (1) fit a susceptible-exposed-infectious-removed disease model to the data with transmission coefficients estimated using approximate Bayesian computation; (2) provide commentary on how a model of this type might be used to provide decision support for disease control authorities. For the outbreak in this herd, the median of the average latent period was 10 days (95% HPD (highest posterior density interval): 2 to 19 days), and the median of the average duration of infectiousness was 3 days (95% HPD: 2 to 4 days). The estimated median for the transmission coefficient was 3.3 (95% HPD: 0.4 to 8.9) infectious contacts per ASF-infectious pig per day. The estimated median for the basic reproductive number, R₀, was 10 (95% HPD: 1.1 to 30). Our estimates of the basic reproductive number R₀ were greater than estimates of R₀ for ASF reported previously. The results presented in this study may be used to estimate the number of pigs expected to be showing clinical signs at a given number of days following an estimated incursion date. This will allow sample size calculations, with or without adjustment to account for less than perfect sensitivity of clinical examination, to be used to determine the appropriate number of pigs to examine to detect at least one with the disease. A second use of the results of this study would be to inform the equation-based within-herd spread components of stochastic agent-based and hybrid simulation models of ASF.

Systematic comparison of modeling fidelity levels and parameter inference settings applied to negative feedback gene regulation

Quantitative stochastic models of gene regulatory networks are important tools for studying cellular regulation. Such models can be formulated at many different levels of fidelity. A practical challenge is to determine what model fidelity to use in order to get accurate and representative results. The choice is important, because models of successively higher fidelity come at a rapidly increasing computational cost. In some situations, the level of detail is clearly motivated by the question under study. In many situations however, many model options could qualitatively agree with available data, depending on the amount of data and the nature of the observations. Here, an important distinction is whether we are interested in inferring the true (but unknown) physical parameters of the model or if it is sufficient to be able to capture and explain available data. The situation becomes complicated from a computational perspective because inference needs to be approximate. Most often it is based on likelihood-free Approximate Bayesian Computation (ABC) and here determining which summary statistics to use, as well as how much data is needed to reach the desired level of accuracy, are difficult tasks. Ultimately, all of these aspects-the model fidelity, the available data, and the numerical choices for inference-interplay in a complex manner. In this paper we develop a computational pipeline designed to systematically evaluate inference accuracy for a wide range of true known parameters. We then use it to explore inference settings for negative feedback gene regulation. In particular, we compare a detailed spatial stochastic model, a coarse-grained compartment-based multiscale model, and the standard well-mixed model, across several data-scenarios and for multiple numerical options for parameter inference. Practically speaking, this pipeline can be used as a preliminary step to guide modelers prior to gathering experimental data. By training Gaussian processes to approximate the distance function values, we are able to substantially reduce the computational cost of running the pipeline.

Using mechanistic model-based inference to understand and project epidemic dynamics with time-varying contact and vaccination rates

Epidemiological models range in complexity from relatively simple statistical models that make minimal assumptions about the variables driving epidemic dynamics to more mechanistic models that include effects such as vaccine-derived and infection-derived immunity, population structure and heterogeneity. The former are often fitted to data in real-time and used for short-term forecasting, while the latter are more suitable for comparing longer-term scenarios under differing assumptions about control measures or other factors. Here, we present a mechanistic model of intermediate complexity that can be fitted to data in real-time but is also suitable for investigating longer-term dynamics. Our approach provides a bridge between primarily empirical approaches to forecasting and assumption-driven scenario models. The model was developed as a policy advice tool for New Zealand's 2021 outbreak of the Delta variant of SARS-CoV-2 and includes the effects of age structure, non-pharmaceutical interventions, and the ongoing vaccine rollout occurring during the time period studied. We use an approximate Bayesian computation approach to infer the time-varying transmission coefficient from real-time data on reported cases. We then compare projections of the model with future, out-of-sample data. We find that this approach produces a good fit with in-sample data and reasonable forward projections given the inherent limitations of predicting epidemic dynamics during periods of rapidly changing policy and behaviour. Results from the model helped inform the New Zealand Government's policy response throughout the outbreak.

Estimating the impact of COVID-19 vaccine allocation inequities: a modeling study

Access to COVID-19 vaccines on the global scale has been drastically impacted by structural socio-economic inequities. Here, we develop a data-driven, age-stratified epidemic model to evaluate the effects of COVID-19 vaccine inequities in twenty lower middle and low income countries (LMIC) sampled from all WHO regions. We focus on the first critical months of vaccine distribution and administration, exploring counterfactual scenarios where we assume the same per capita daily vaccination rate reported in selected high income countries. We estimate that, in this high vaccine availability scenario, more than 50% of deaths (min-max range: [56% - 99%]) that occurred in the analyzed countries could have been averted. We further consider a scenario where LMIC had similarly early access to vaccine doses as high income countries; even without increasing the number of doses, we estimate an important fraction of deaths (min-max range: [7% - 73%]) could have been averted. In the absence of equitable allocation, the model suggests that considerable additional non-pharmaceutical interventions would have been required to offset the lack of vaccines (min-max range: [15% - 75%]). Overall, our results quantify the negative impacts of vaccines inequities and call for amplified global efforts to provide better access to vaccine programs in low and lower middle income countries.

Impact and cost-effectiveness of non-governmental organizations on the HIV epidemic in Ukraine among MSM

OBJECTIVE

Non-governmental organizations (NGOs) in Ukraine have provided HIV testing, treatment, and condom distribution for MSM. HIV prevalence among MSM in Ukraine is 5.6%. We estimated the impact and cost-effectiveness of MSM-targeted NGO activities in Ukraine.

DESIGN

A mathematical model of HIV transmission among MSM was calibrated to data from Ukraine (2011-2018).

METHODS

The model, designed before the 2022 Russian invasion of Ukraine, evaluated the impact of 2018 status quo coverage levels of 28% of MSM being NGO clients over 2016-2020 and 2021-2030 compared with no NGO activities over these time periods. Impact was measured in HIV incidence and infections averted. We compared the costs and disability adjusted life years (DALYs) for the status quo and a counterfactual scenario (no NGOs 2016-2020, but with NGOs thereafter) until 2030 to estimate the mean incremental cost-effectiveness ratio (cost per DALY averted).

RESULTS

Without NGO activity over 2016-2020, the HIV incidence in 2021 would have been 44% (95% credibility interval: 36-59%) higher than with status quo levels of NGO activity, with 25% (21-30%) more incident infections occurring over 2016-2020. Continuing with status quo NGO coverage levels will decrease HIV incidence by 41% over 2021-2030, whereas it will increase by 79% (60-120%) with no NGOs over this period and 37% (30-51%) more HIV infections will occur. Compared with if NGO activities had ceased over 2016-2020 (but continued thereafter), the status quo scenario averts 14 918 DALYs over 2016-2030 with a mean incremental cost-effectiveness ratio of US$600.15 per DALY averted.

CONCLUSION

MSM-targeted NGOs in Ukraine have prevented considerable HIV infections and are highly cost-effective compared with a willingness-to-pay threshold of 50% of Ukraine's 2018 GDP (US$1548).

Mathematical analysis of phototransduction reaction parameters in rods and cones

Retinal photoreceptor cells, rods and cones, convert photons of light into chemical and electrical signals as the first step of the visual transduction cascade. Although the chemical processes in the phototransduction system are very similar to each other in these photoreceptors, the light sensitivity and time resolution of the photoresponse in rods are functionally different than those in the photoresponses of cones. To systematically investigate how photoresponses are divergently regulated in rods and cones, we have developed a detailed mathematical model on the basis of the Hamer model. The current model successfully reconstructed light intensity-, ATP- and GTP-dependent changes in concentrations of phosphorylated visual pigments (VPs), activated transducins (Tr*s) and phosphodiesterases (PDEs) in rods and cones. In comparison to rods, the lower light sensitivity of cones was attributed not only to the lower affinity of activated VPs for Trs but also to the faster desensitization of the VPs. The assumption of an intermediate inactive state, MIIⁱ, in the thermal decay of activated VPs was essential for inducing faster inactivation of VPs in rods, and possibly also in cones.

Probabilistic program inference in network-based epidemiological simulations

Accurate epidemiological models require parameter estimates that account for mobility patterns and social network structure. We demonstrate the effectiveness of probabilistic programming for parameter inference in these models. We consider an agent-based simulation that represents mobility networks as degree-corrected stochastic block models, whose parameters we estimate from cell phone co-location data. We then use probabilistic program inference methods to approximate the distribution over disease transmission parameters conditioned on reported cases and deaths. Our experiments demonstrate that the resulting models improve the quality of fit in multiple geographies relative to baselines that do not model network topology.

Estimation of age-stratified contact rates during the COVID-19 pandemic using a novel inference algorithm

Well parameterized epidemiological models including accurate representation of contacts are fundamental to controlling epidemics. However, age-stratified contacts are typically estimated from pre-pandemic/peace-time surveys, even though interventions and public response likely alter contacts. Here, we fit age-stratified models, including re-estimation of relative contact rates between age classes, to public data describing the 2020-2021 COVID-19 outbreak in England. This data includes age-stratified population size, cases, deaths, hospital admissions and results from the Coronavirus Infection Survey (almost 9000 observations in all). Fitting stochastic compartmental models to such detailed data is extremely challenging, especially considering the large number of model parameters being estimated (over 150). An efficient new inference algorithm ABC-MBP combining existing approximate Bayesian computation (ABC) methodology with model-based proposals (MBPs) is applied. Modified contact rates are inferred alongside time-varying reproduction numbers that quantify changes in overall transmission due to pandemic response, and age-stratified proportions of asymptomatic cases, hospitalization rates and deaths. These inferences are robust to a range of assumptions including the values of parameters that cannot be estimated from available data. ABC-MBP is shown to enable reliable joint analysis of complex epidemiological data yielding consistent parametrization of dynamic transmission models that can inform data-driven public health policy and interventions. This article is part of the theme issue 'Technical challenges of modelling real-life epidemics and examples of overcoming these'.

Efficient Bayesian inference for stochastic agent-based models

The modelling of many real-world problems relies on computationally heavy simulations of randomly interacting individuals or agents. However, the values of the parameters that underlie the interactions between agents are typically poorly known, and hence they need to be inferred from macroscopic observations of the system. Since statistical inference rests on repeated simulations to sample the parameter space, the high computational expense of these simulations can become a stumbling block. In this paper, we compare two ways to mitigate this issue in a Bayesian setting through the use of machine learning methods: One approach is to construct lightweight surrogate models to substitute the simulations used in inference. Alternatively, one might altogether circumvent the need for Bayesian sampling schemes and directly estimate the posterior distribution. We focus on stochastic simulations that track autonomous agents and present two case studies: tumour growths and the spread of infectious diseases. We demonstrate that good accuracy in inference can be achieved with a relatively small number of simulations, making our machine learning approaches orders of magnitude faster than classical simulation-based methods that rely on sampling the parameter space. However, we find that while some methods generally produce more robust results than others, no algorithm offers a one-size-fits-all solution when attempting to infer model parameters from observations. Instead, one must choose the inference technique with the specific real-world application in mind. The stochastic nature of the considered real-world phenomena poses an additional challenge that can become insurmountable for some approaches. Overall, we find machine learning approaches that create direct inference machines to be promising for real-world applications. We present our findings as general guidelines for modelling practitioners.

Efficient Bayesian inference for mechanistic modelling with high-throughput data

Bayesian methods are routinely used to combine experimental data with detailed mathematical models to obtain insights into physical phenomena. However, the computational cost of Bayesian computation with detailed models has been a notorious problem. Moreover, while high-throughput data presents opportunities to calibrate sophisticated models, comparing large amounts of data with model simulations quickly becomes computationally prohibitive. Inspired by the method of Stochastic Gradient Descent, we propose a minibatch approach to approximate Bayesian computation. Through a case study of a high-throughput imaging scratch assay experiment, we show that reliable inference can be performed at a fraction of the computational cost of a traditional Bayesian inference scheme. By applying a detailed mathematical model of single cell motility, proliferation and death to a data set of 118 gene knockdowns, we characterise functional subgroups of gene knockdowns, each displaying its own typical combination of local cell density-dependent and -independent motility and proliferation patterns. By comparing these patterns to experimental measurements of cell counts and wound closure, we find that density-dependent interactions play a crucial role in the process of wound healing.

During wound healing, cells work together to close a wound to restore tissue integrity. Thousands of different genes play a role in wound healing, and scratch assay experiments are routinely used to investigate the role of these genes by analysing how a wound closes when each of these is not expressed, i.e. knocked down. So far, the impact of knocking down genes on wound healing has been determined by comparing the size of the wound before and after a given time period, but these measurements do not elucidate the fine-scale mechanisms that determine how cells behave in the presence of their neighbours. By combining a detailed mathematical model with experimental imaging of wound healing, we identify how cells respond to and work together with their neighbours during wound healing. Applying this method to a large number of gene knockdowns, we identify three well-defined functional subgroups of knockdowns, each displaying its own typical behaviours of movement and proliferation to close the wound. These observations explain the role of each of the knockdowns on wound healing and further our understanding of cell-cell interactions in wound healing.

A Bayesian Approach to Unsupervised, Non-Intrusive Load Disaggregation

Estimating household energy use patterns and user consumption habits is a fundamental requirement for management and control techniques of demand response programs, leading to a growing interest in non-intrusive load disaggregation methods. In this work we propose a new methodology for disaggregating the electrical load of a household from low-frequency electrical consumption measurements obtained from a smart meter and contextual environmental information. The method proposed allows, with an unsupervised and non-intrusive approach, to separate loads into two components related to environmental conditions and occupants' habits. We use a Bayesian approach, in which disaggregation is achieved by exploiting actual electrical load information to update the a priori estimate of user consumption habits, to obtain a probabilistic forecast with hourly resolution of the two components. We obtain a remarkably good accuracy for a benchmark dataset, higher than that obtained with other unsupervised methods and comparable to the results of supervised algorithms based on deep learning. The proposed procedure is of great application interest in that, from the knowledge of the time series of electricity consumption alone, it enables the identification of households from which it is possible to extract flexibility in energy demand and to realize the prediction of the respective load components.