Geology topics

Daniel P. Walsh

Publications and source records attributed to Daniel P. Walsh.

At least 19 recordsLinked to original sources

USGS2026

Digging into soil: Effects of soil texture on RT-QuIC performance for environmental prion surveillance.

Chronic wasting disease (CWD) is a fatal neurodegenerative disease caused by infectious prions affecting wild and captive cervids. Transmission occurs directly between hosts or indirectly through contact with prion-contaminated environments. Soils, particularly those rich in clay, are hypothesized to enhance prion stability, retention, and bioavailability. Accurate detection of prions is therefore important for understanding environmental transmission risks. Real-time quaking-induced conversion (RT-QuIC) is a sensitive assay used to detect PrP CWD in tissue, excreta, and environmental materials. However, RT-QuIC performance across soil textures has not been evaluated. This study assessed RT-QuIC sensitivity and specificity using laboratory-prepared soils spiked with CWD-positive brain homogenate or water controls under a standardized extraction method. Conditional on the extraction method used, results suggest that RT-QuIC performance depends on soil texture, and thus, an optimal time-to-threshold (TTT) cutoff required to balance sensitivity and specificity will also vary with soil texture. RT-QuIC exhibited higher sensitivity and moderate specificity in soils with low clay (<20%) and moderate to high silt content, whereas high-clay soils (>20%) with low to moderate silt content (2%–60%) reduced both sensitivity and specificity, and required shorter TTT cutoffs. These findings highlight the importance of accounting for soil texture in environmental CWD surveillance.

Environmental Science and Technology↗

USGS2025

Development of high-throughput genomic resources to inform white-tailed deer population and disease management

White-tailed deer ( Odocoileus virginianus ) are the most abundant and widespread cervid in North America. Genetic data are used as a tool to monitor populations and make management decisions for this game species. However, the development and use of genomic tools that can generate a set of markers suitable for longitudinal genomic data collection, whether for management purposes or to study the demographic and evolutionary processes of widely distributed species, have been challenging. This is mainly due to the cost required to fully implement and interpret the data produced. Here, we generated whole genome resequencing data for 44 free-ranging deer from three regions in their central and eastern North American range and identified over 89 million single nucleotide polymorphisms (SNPs). We used a subset of these SNPs to develop two nested SNP tools, a high-density array (702,183 SNPs) and a medium-density array (72,723 SNPs) to support deer and chronic wasting disease (CWD) management and research. SNPs were selected to ensure an even distribution across scaffolds of the reference genome and include SNPs associated with CWD susceptibility. Using genotyping results for 469 deer from 15 states in the US and Mexico generated by the high-density array and 1335 deer from 18 states generated by the medium-density array, we assessed genotyping success across different populations and explored some insights into population structure. These genomic tools offer a standard set of markers that will enable researchers and managers to address important questions related to white-tailed deer and CWD management. Our SNP arrays also offer the opportunity to examine aspects of white-tailed deer ecology and evolutionary history that were previously difficult to address.

Molecular Ecology Resources↗

USGS2025

Strategic planning of prevention and surveillance for emerging diseases and invasive species

Emerging infectious diseases and biological invasions pose increasing threats to public and ecosystems health. Proactive measures—such as prevention and surveillance taken before initial detection of the pathogen or species—are essential to ensure minimal spread prior to first detection. We developed an optimization model to determine where, when, and how much effort should be allocated to prevention versus surveillance. The model accounts for imperfect detection, system dynamics, spatial heterogeneity in risk and costs and is scalable to large landscapes. We found that the most cost-effective strategy is to maintain the prevention and surveillance efforts at stable equilibrium for the majority of the time, with deviations occurring only initially to steer the system toward the equilibrium. The equilibrium effort is jointly determined by the introduction risk, management costs, and total budget. Application of this model to chronic wasting disease in New York State suggests that the optimal strategy could reduce the cumulative disease cases before initial detection by an average of 22% compared to current practice. The optimal surveillance strategy could detect the disease on average over 8 mo earlier than the current strategy.

PNAS↗

USGS2025

Animal trajectory imputation and uncertainty quantification via deep learning

Imputing missing data in animal trajectories is crucial for understanding animal movements during unobserved periods. However, the traditional methods, such as linear interpolation and the continuous-time correlated random walk model, are often inadequate to capture the complexity of animal movements. Here, we develop a deep learning approach to animal trajectory imputation by a conditional diffusion model. Unlike the traditional methods, our deep learning method uses observed data and external covariates to impute missing positions along an animal trajectory, capturing periodic patterns and the influence of covariates, which leads to more accurate imputations. In a case study of imputing deer trajectories, our method not only provides more accurate deterministic imputations than existing approaches but also achieves uncertainty quantification through probabilistic imputation.

Environmetrics↗

USGS2025

Doe diligence: A regional analysis of antlerless deer harvest regulations in the Midwestern United States of America.

Wildlife management in the United States of America (US) is primarily delegated to the individual states wherein state wildlife agencies manage wildlife populations to achieve multiple and sometimes conflicting objectives. White-tailed deer ( Odocoileus virginianus ) are an important species in the Midwestern US whose populations are primarily managed through recreational hunting. Managers aim to adjust populations by altering the harvest of antlerless (usually female) animals by changing the number of harvest permits available, hunting season lengths, or applying incentive programs like earn-a-buck, where a hunter must harvest an antlerless deer before they may harvest an antlered deer. We estimated the effect on antlerless deer harvest from changes in these regulations and changes in the number of licensed hunters across eight states in the Midwest. We used a Bayesian hierarchical model to estimate individual state and regional (i.e., across all states) effects. We found that increasing antlerless harvest permits increased antlerless harvest; however, this effect plateaued as the number of available permits increased. Providing unlimited harvest permits increased harvest, but the same increases were achieved by minimally increasing the number of limited harvest permits. Increasing the length of hunting season had a generally positive effect on antlerless harvest but the effect was non-linear and state dependent. The earn-a-buck incentive program resulted in the largest estimated increase in harvest. Finally, the number of licensed deer hunters in a state had a strong positive effect on the number of antlerless deer harvested. Our findings show that commonly applied changes in harvest regulations have a weak effect on the number of antlerless deer harvested, highlighting the challenges facing deer managers in the Midwestern US.

Illinois, Indiana, Iowa, Michigan, Minnesota, Miss↗

USGS2025

White‐tailed deer habitat use and implications for chronic wasting disease transmission

Animal space use, activity patterns, and habitat selection—and heterogeneity in these patterns—have important implications for where and when infectious diseases are transmitted. White-tailed deer ( Odocoileus virginianus ) are habitat generalists, with a high degree of heterogeneity in their movement ecology based on sex, age, season, and region. These heterogeneities have important implications for the transmission and management of chronic wasting disease (CWD), which is a deadly prion disease transmitted both directly and indirectly through the environment. As such, favored deer habitats may promote direct interactions between conspecifics or indirect spatial overlap and subsequent environmental transmission. However, little is known about how individual animal space use translates to actual spatial overlap between individuals, leaving uncertainty in how habitat shapes the risk of direct or environmental CWD transmission. In this study, we evaluated seasonal activity patterns, home ranges, and habitat selection for 596 white-tailed deer in southwest Wisconsin, USA, from 2017-2022. We also estimated seasonal encounter distributions—regions where a pair of deer were most likely to encounter each other—for all pairs of deer putatively in different social groups (between-group) in our study, and quantified seasonal variation in the habitat composition of these areas. We found that deer selection for crops, pasture, or grasslands was generally low, relative to forest, but was highest in the post-fawning (summer) and non-breeding (winter) seasons. We observed similar patterns for the habitat composition of encounter distributions, suggesting that crops, pasture, and grasslands may be attractive resources that facilitate between-group transmission. Site fidelity between years was generally high; combined with small female home ranges in the fawning season, this implies that females likely re-use the same small, high-quality fawning habitats from year to year. We found that attraction toward between-group individuals was low during the post-fawning season but high during the breeding (fall) and non-breeding seasons. These results suggest that space use and habitat selection could shape the risk of environmental transmission in the fawning and post-fawning seasons, social selection could favor direct transmission risk in the breeding season, and combined social and habitat selection may shape risk of both direct and environmental transmission during the non-breeding season. We provide a detailed picture of the physiological and social drivers of deer movement through the year, with implications for CWD transmission and management.

Iowa, Wisconsin↗

USGS2025

Learning complex spatial dynamics of wildlife diseases with machine learning-guided partial differential equations

Emerging wildlife pathogens often display geographic variability due to landscape heterogeneity. Modeling approaches capable of learning complex, non-linear spatial dynamics of diseases are needed to rigorously assess and mitigate the effects of pathogens on wildlife health and biodiversity. We propose a novel machine learning (ML)-guided approach that leverages prior physical knowledge of ecological systems, using partial differential equations. We present our approach, taking advantage of the universal function approximation property of neural networks for flexible representation of the underlying dynamics of the geographic spread and growth of wildlife diseases. We demonstrate the benefits of our approach by comparing its forecasting power with commonly used methods and highlighting the obtained insights on disease dynamics. Additionally, we show the theoretical guarantees for the approximation error of our model. We illustrate the implementation of our ML-guided approach using data from white-nose syndrome (WNS) outbreaks in bat populations across the US. WNS is an infectious fungal disease responsible for significant declines in bat populations. Our results on WNS are useful for disease surveillance and bat conservation efforts. Our methods can be broadly used to assess the effects of environmental and anthropogenic drivers impacting wildlife health and biodiversity.

Environmental Data Science↗

USGS2025

Characterization of the long-distance dispersal kernel of white-tailed deer and evaluating its impact on chronic wasting disease spread in Wisconsin

Chronic wasting disease (CWD) is a fatal neurodegenerative disease infecting cervids. It is highly contagious and caused by misfolded prions that propagate via templated conformational conversion of the cervid’s normal prion protein. Prevalence of CWD in free-ranging deer in North America is mostly low, but in some regions local prevalence has reached 80%. CWD prions can be transmitted via direct contact with infected individuals or indirectly through the environment. Infected individuals shed prions through feces, urine, saliva or carcasses, and prions have long environmental persistence. Long-distance dispersal of infected deer poses a significant risk for CWD spread. We propose an integrodifference equation (IDE) model to capture CWD dynamics and the consequences of long-distance dispersal behavior in white-tailed deer (WTD, Odocoileus virginianus ). A diffusion-settling model characterizes long-distance dispersal kernels, accommodating hypothetical dispersal behaviors through time-dependent settling rate functions. Three new closed-form dispersal kernels are approximated using Laplace’s method and parameterized with GPS location data collected from WTD in Wisconsin, USA. Settling rates reflecting ongoing sensitivity to stimuli which prompt deer to disperse from their natal home range give the most supported long-distance dispersal kernel. Impact of long-distance dispersal on CWD spread is quantified using the IDE model. At high population densities, long-distance dispersal can magnify CWD spread by a factor of four. At lower population densities single infected individuals cannot initiate an outbreak, but CWD may still spread due to the accumulation of environmental hazard from prions behind the wave of invasion, possibly presenting substantial management challenges.

Wisconsin↗

USGS2025

Spatiotemporal causal inference with mechanistic ecological models: Evaluating targeted culling on chronic wasting disease dynamics in cervids

Spatiotemporal causal inference methods are needed to detect the effect of interventions on indirectly measured epidemiological outcomes that go beyond studying spatiotemporal correlations. Chronic wasting disease (CWD) causes neurological degeneration and eventual death to white-tailed deer ( Odocoileus virginianus ) in Wisconsin. Targeted culling involves removing deer after traditional hunting seasons in areas with high CWD prevalence. The evaluation of the causal effects of targeted culling in the spread and growth of CWD is an important unresolved research and CWD management question that can guide surveillance efforts. Reaction–diffusion partial differential equations (PDEs) can be used to mechanistically model the underlying spatiotemporal dynamics of wildlife diseases, like CWD, allowing researchers to make inference about unobserved epidemiological quantities. These models indirectly regress spatiotemporal covariates on diffusion and growth rates parameterizing such PDEs, obtaining associational conclusions. In this work we develop an innovative method to obtain causal estimators for the effect of targeted culling interventions on CWD epidemiological processes using an inverse-probability-of-treatment-weighted technique by means of marginal structural models embedded in the PDE fitting process. Additionally we establish a novel scheme for sensitivity analysis under unmeasured confounder for testing the hypothesis of a significant causal effect in the indirectly measured epidemiological outcomes. Our methods can be broadly used to study the impact of spatiotemporal interventions and treatment exposures in the epidemiological evolution of infectious diseases that can help to inform future efforts to mitigate public health implications and wildlife disease burden.

Wisconsin↗

USGS2025

A system dynamics model to understand the integrated ecological and human dimension aspects of wildlife health and disease management

Chronic wasting disease (CWD) presents an ongoing challenge for the management of deer populations and sustaining harvest opportunities across North America. Existing disease models often fail to fully capture the complex interplay between disease dynamics, host ecology, and socio-economic factors. We developed a comprehensive system dynamics (SD) model that integrates demographic, epidemiological, ecological, and socio-economic processes within a single model to more fully characterize the complex network of causal feedbacks throughout the system. The model was calibrated using a Bayesian approach that incorporates prior knowledge to generate biologically interpretable outcomes, even with sparse data. For estimating the joint posterior distribution of model parameters, we leveraged time series of deer abundance, harvest composition, genetic profiles, CWD surveillance, and hunter demographics and behavior. Model outputs reproduced key system behaviors, including observed CWD prevalence trends, deer population dynamics, and hunter license purchasing patterns. Model predictions were most sensitive to parameters governing initial deer population size and recruitment. While model predictions generally aligned with observed data, discrepancies in early CWD detection and overestimation of the reactivation of long-inactive hunters reflect data limitations and modeling challenges. Key results suggest that indirect transmission is necessary to explain observed prevalence, that transmission is moderately density-dependent, and that observed population-level genetic shifts driven by CWD may play a role in transmission and progression. The SD modelling approach enabled estimation of difficult-to-measure parameters and identified potential leverage points for management—such as prioritizing increasing participation in antlerless harvest of existing hunters over the recruitment of new hunters. This integrated modeling approach offers a flexible foundation for adaptive wildlife disease management and emphasizes the value of unifying biological and human dimension processes to better inform effective, evidence-based policy.

BioRxiv↗

USGS2025

Fatal interactions: Pneumonia in bighorn lambs following experimental exposure to carriers of Mycoplasma ovipneumoniae

We hypothesized that bighorn sheep ewes with chronic nasal Mycoplasma ovipneumoniae carriage are the source of infection that results in fatal lamb pneumonia. We tested this hypothesis in captive bighorn ewes at two study facilities over a 5-year period, by identifying carrier ewes and then comparing lamb fates in groups that did (exposed pens) or did not (non-exposed pens) include one or more carrier ewes. Most (23 of 30) lambs born in exposed pens, but none of 11 lambs born in non-exposed pens, contracted fatal pneumonia. In addition, surviving lambs in exposed pens showed obvious signs of respiratory disease while lambs in non-exposed pens did not. In crossover experiments, individual non-carrier ewes had lambs that experienced fatal pneumonia in years when housed in exposed pens, but not in years when housed in non-exposed pens. The results of these studies clearly associate lamb pneumonia to exposure to M. ovipneumoniae carrier ewes, consistent with a necessary role for this agent in epizootic pneumonia of bighorn sheep. These data specifically highlight the role of chronic M. ovipneumoniae carriage by some bighorn ewes in the epidemiology of this population-limiting wildlife disease.

Idaho, Montana, Nevada, Oregon, South Dakota, Wash↗

USGS2024

Wildlife health capacity enhancement in Thailand through the World Organisation for Animal Health Twinning Program

There is an increasing need for robust wildlife health programs that provide surveillance and management for diseases in wildlife and wild aquatic populations to manage associated risks. This paper illustrates the value of a systematic method to enhancing wildlife health programs. The U.S. Geological Survey and Mahidol University, Faculty of Veterinary Science, Thailand National Wildlife Health Center formally twinned under the auspices of the World Organisation for Animal Health to enhance wildlife health capacity in Thailand and the Southeast Asia Region. We used a system-wide approach to holistically and interdependently enhance capacity. The project commenced with a wildlife health program needs assessment, and capacity enhancement focused on strengthening the general wildlife health surveillance network and improving wildlife health information management. Activities included partner surveys, interactive and didactic workshops, and individual personnel training. Topics included development of wildlife health information management systems, analysis of the current surveillance network, development of a Theory of Change for a strengthened surveillance network, planning workshops to create a wildlife health network, training on wildlife disease outbreak investigation and field sample collection, leading networks, and individual training on bioinformatics and laboratory techniques. Engagement of stakeholders at all levels, continuous communication throughout the project, use of both strategic planning tools and pedagogical methods, and using iterative and adaptive approaches, were key factors to the success of this project.

Frontiers in Veterinary Science↗

USGS2024

Epidemiological modeling of SARS-CoV-2 in white-tailed deer (Odocoileus virginianus) reveals conditions for introduction and widespread transmission

Emerging infectious diseases with zoonotic potential often have complex socioecological dynamics and limited ecological data, requiring integration of epidemiological modeling with surveillance. Although our understanding of SARS-CoV-2 has advanced considerably since its detection in late 2019, the factors influencing its introduction and transmission in wildlife hosts, particularly white-tailed deer ( Odocoileus virginianus ), remain poorly understood. We use a Susceptible-Infected-Recovered-Susceptible epidemiological model to investigate the spillover risk and transmission dynamics of SARS-CoV-2 in wild and captive white-tailed deer populations across various simulated scenarios. We found that captive scenarios pose a higher risk of SARS-CoV-2 introduction from humans into deer herds and subsequent transmission among deer, compared to wild herds. However, even in wild herds, the transmission risk is often substantial enough to sustain infections. Furthermore, we demonstrate that the strength of introduction from humans influences outbreak characteristics only to a certain extent. Transmission among deer was frequently sufficient for widespread outbreaks in deer populations, regardless of the initial level of introduction. We also explore the potential for fence line interactions between captive and wild deer to elevate outbreak metrics in wild herds that have the lowest risk of introduction and sustained transmission. Our results indicate that SARS-CoV-2 could be introduced and maintained in deer herds across a range of circumstances based on testing a range of introduction and transmission risks in various captive and wild scenarios. Our approach and findings will aid One Health strategies that mitigate persistent SARS-CoV-2 outbreaks in white-tailed deer populations and potential spillback to humans.

PLOS Computational Biology↗

USGS2024

Validation of a real-time quaking-induced conversion (RT-QuIC) assay protocol to detect chronic wasting disease using rectal mucosa of naturally infected, pre-clinical white-tailed deer (Odocoileus virginianus)

Chronic wasting disease (CWD) is a fatal prion disease of cervids spreading across North America. More effective mitigation efforts may require expansion of the available toolkit to include new methods that provide earlier antemortem detection, higher throughput, and less expense than current immunohistochemistry (IHC) methods. The rectal mucosa near the rectoanal junction is a site of early accumulation of CWD prions and is safely sampled in living animals by pinch biopsy. A fluorescence-based, 96-well format, protein-aggregation assay-the real-time quaking-induced conversion (RT-QuIC) assay-is capable of ultra-sensitive detection of CWD prions. Notably, the recombinant protein substrate is crucial to the assay's performance and is now commercially available. In this blinded independent study, the preclinical diagnostic performance of a standardized RT-QuIC protocol using a commercially sourced substrate (MNPROtein) and a laboratory-produced substrate was studied using mock biopsy samples of the rectal mucosa from 284 white-tailed deer ( Odocoileus virginianus ). The samples were from a frozen archive of intact rectoanal junctions collected at depopulations of farmed herds positive for CWD in the United States. All deer were pre-clinical at the time of depopulation and infection status was established from the regulatory record, which evaluated the medial retropharyngeal lymph nodes (MRPLNs) and obex by CWD-IHC. A pre-analytic sample precipitation step was found to enhance the protocol's detection limit. Performance metrics were influenced by the choice of RT-QuIC diagnostic cut points (minimum number of positive wells and assay time) and by deer attributes (preclinical infection stage and prion protein genotype). The peak overall diagnostic sensitivities of the protocol were similar for both substrates (MNPROtein, 76.8%; laboratory-produced, 73.2%), though each was achieved at different cut points. Preclinical infection stage and prion protein genotype at codon 96 (G = glycine, S = serine) were primary predictors of sensitivity. The diagnostic sensitivities in late preclinical infections (CWD-IHC positive MPRLNs and obex) were similar, ranging from 96% in GG96 deer to 80% in xS96 deer (x = G or S). In early preclinical infections (CWD-IHC positive MRPLNs only), the diagnostic sensitivity was 64-71% in GG96 deer but only 25% in xS96 deer. These results demonstrate that this standardized RT-QuIC protocol for rectal biopsy samples using a commercial source of substrate produced stratified diagnostic sensitivities similar to or greater than those reported for CWD-IHC but in less than 30 hours of assay time and in a 96-well format. Notably, the RT-QuIC protocol used herein represents a standardization of protocols from several previous studies. Alignment of the sensitivities across these studies suggests the diagnostic performance of the assay is robust given quality reagents, optimized diagnostic criteria, and experienced staff.

PLoS ONE↗

USGS2024

Prion seeding activity in plant tissues detected by RT-QuIC

Prion diseases such as scrapie, bovine spongiform encephalopathy (BSE), and chronic wasting disease (CWD) affect domesticated and wild herbivorous mammals. Animals afflicted with CWD, the transmissible spongiform encephalopathy of cervids (deer, elk, and moose), shed prions into the environment, where they may persist and remain infectious for years. These environmental prions may remain in soil, be transported in surface waters, or assimilated into plants. Environmental sampling is an emerging area of TSE research and can provide more information about prion fate and transport once shed by infected animals. In this study, we have developed the first published method for the extraction and detection of prions in plant tissue using the real-time quaking-induced conversion (RT-QuIC) assay. Incubation with a zwitterionic surfactant followed by precipitation with sodium phosphotungstate concentrates the prions within samples and allows for sensitive detection of prion seeding activity. Using this protocol, we demonstrate that prions can be detected within plant tissues and on plant surfaces using the RT-QuIC assay.

Pathogens↗

USGS2024

Detection of prions from spiked and free-ranging carnivore feces

Chronic wasting disease (CWD) is a highly contagious, fatal neurodegenerative disease caused by infectious prions (PrP CWD ) affecting wild and captive cervids. Although experimental feeding studies have demonstrated prions in feces of crows ( Corvus brachyrhynchos ), coyotes ( Canis latrans ), and cougars ( Puma concolor ), the role of scavengers and predators in CWD epidemiology remains poorly understood. Here we applied the real-time quaking-induced conversion (RT-QuIC) assay to detect PrP CWD in feces from cervid consumers, to advance surveillance approaches, which could be used to improve disease research and adaptive management of CWD. We assessed recovery and detection of PrP CWD by experimental spiking of PrP CWD into carnivore feces from 9 species sourced from CWD-free populations or captive facilities. We then applied this technique to detect PrP CWD from feces of predators and scavengers in free-ranging populations. Our results demonstrate that spiked PrP CWD is detectable from feces of free-ranging mammalian and avian carnivores using RT-QuIC. Results show that PrP CWD acquired in natural settings is detectable in feces from free-ranging carnivores, and that PrP CWD rates of detection in carnivore feces reflect relative prevalence estimates observed in the corresponding cervid populations. This study adapts an important diagnostic tool for CWD, allowing investigation of the epidemiology of CWD at the community-level.

Scientific Reports↗

USGS2024

Chronic wasting disease: State of the science

Chronic wasting disease (CWD) is a prion disease affecting cervid species, both free-ranging and captive populations. As the geographic range continues to expand and disease prevalence continues to increase, CWD will have an impact on cervid populations, local economies, and ecosystem health. Mitigation of this “wicked” disease will require input from many different stakeholders including hunters, landowners, research biologists, wildlife managers, and others, working together. The NC1209 (North American interdisciplinary chronic wasting disease research consortium) is composed of scientists from different disciplines involved with investigating and managing CWD. Leveraging this broad breadth of expertise, the Consortium has created a state-of-the-science review of five key aspects of CWD, including current diagnostic capabilities for detecting prions, requirements for validating these diagnostics, the role of environmental transmission in CWD dynamics, and potential zoonotic risks associated with CWD. The goal of this review is to increase stakeholders’, managers’, and decision-makers’ understanding of this disease informed by current scientific knowledge.

Pathogens↗

USGS2024

Spatio-temporal ecological models via physics-informed neural networks for studying chronic wasting disease

To mitigate the negative effects of emerging wildlife diseases in biodiversity and public health it is critical to accurately forecast pathogen dissemination while incorporating relevant spatio-temporal covariates. Forecasting spatio-temporal processes can often be improved by incorporating scientific knowledge about the dynamics of the process using physical models. Ecological diffusion equations are often used to model epidemiological processes of wildlife diseases where environmental factors play a role in disease spread. Physics-informed neural networks (PINN) are deep learning algorithms that constrain neural network predictions based on physical laws and therefore are powerful forecasting models useful even in cases of limited and imperfect training data. In this paper, we develop a novel ecological modeling tool using PINNs, which fits a feedforward neural network and simultaneously performs parameter identification in a partial differential equation (PDE) with varying coefficients. We demonstrate the applicability of our model by comparing it with the commonly used Bayesian stochastic partial differential equation method and traditional machine learning approaches, showing that our proposed model exhibits superior prediction and forecasting performance when modeling chronic wasting disease in deer in Wisconsin. Furthermore, our model provides the opportunity to obtain scientific insights into spatiotemporal covariates affecting spread and growth of diseases. This work contributes to future machine learning and statistical methodology development by studying spatio-temporal processes enhanced by prior physical knowledge.

Spatial Statistics↗