Pith. sign in

REVIEW 3 major objections 4 minor 108 references

A Bayesian Spatiotemporal Model to Estimate Disease Burden Using Hospital-Based Active Surveillance

T0 review · 3 major / 4 minor · reviewed 2026-08-02 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims Puerto Rico's true hospital-presenting leptospirosis rate is roughly four times the reported 8–9 per 100,000 — about 35 per 100,000 in 2022 and 2023 — once under-capture and imperfect tests are modeled.

desk verdict Worth engaging: a genuinely new way to combine misaligned active/passive surveillance, but the headline burden estimate rests on an informative prior that the data barely update. read the letter →

arxiv 2607.13185 v1 pith:B337SLKK submitted 2026-07-14 stat.AP stat.ME

Brent Strong , Claudia Muñoz-Zanzi , Caitlin Ward This is my paper
classification stat.APstat.ME MSC 62F1562M3062P10
keywords Bayesianspatiotemporalmodeldiseaseburdenleptospirosisactivesurveillancepassivecapture-recapturespatialmisalignmentPoisson-logistic
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper sets out to estimate the true burden of hospital-presenting leptospirosis in Puerto Rico when the only region-level data come from an under-counting passive surveillance system and the corrective active-surveillance data come from just four hospitals. Its central empirical claim is that the passive system captures only about one in five hospital-presenting cases, so the observed island-wide rate of 8–9 per 100,000 understates the true rate by roughly a factor of four: the model estimates 35 per 100,000 in both 2022 and 2023. Methodologically, it extends the Poisson-logistic model to reconcile hospital-level and region-level data by decomposing capture into a hospital-specific probability times a travel-time-based probability of choosing that hospital, and it incorporates imperfect diagnostic testing by marginalizing out individual disease status. The authors argue this matters for public health because the true rate, not the reported one, is what should drive resource allocation, testing guidance, and mitigation for a neglected disease.

What carries the argument

The engine is an extended Poisson-logistic model in which observed regional counts follow Poisson(μ Σ_h π^C_{h,s} π^P_h): the true disease rate is multiplied by a capture-weighted average over hospitals, where π^P_h is the probability a case at hospital h is recorded by passive surveillance and π^C_{h,s} is the probability a patient from region s chooses hospital h. Choice probabilities come from a gravity model of travel time and capacity, letting hospital-level capture estimates be transported to regions without active surveillance. Active-surveillance data enter through a capture-recapture multinomial over the four capture states; imperfect tests are handled by marginalizing out individua

What would settle it

Obtain hospital-level passive case counts, or expand active surveillance until there are dozens of passive–active overlaps rather than one, and re-estimate the model. If the capture probability came out near 0.35 instead of the prior-centered 0.22, the corrected island-wide rate would drop to roughly 22 per 100,000 — empirically distinguishable from 35. The paper's 7-, 14-, and 21-hospital simulations show how much surveillance coverage would be needed to settle the estimate.

Watch

Extended reading notes

Core claim

The paper claims Puerto Rico's true hospital-presenting leptospirosis rate is far higher than passive surveillance reports. Extending the Poisson-logistic capture-recapture framework — true counts drawn Poisson, observed counts binomial thinnings of them — to the case where active surveillance runs in only four hospitals while passive counts exist only as regional aggregates, the model estimates passive capture at ~22% of cases and an island-wide true rate near 35 per 100,000 in 2022 (95% PI 19–64) and 2023 (95% PI 20–65), versus observed 8 and 9. Regional estimates span 12 (Fajardo) to 60 (Caguas) per 100,000 in 2023; sensitivity analyses moving the diagnostic-test priors shift the island-w

Load-bearing premise

The load-bearing premise is the informative prior on the average passive surveillance capture probability, α0 ~ N(−1.4, 0.5²), which centers capture near 20%: with only one passive–active overlap among 16 positive tests, the posterior capture probability (median 0.22) essentially mirrors that prior, and the estimated true rate is close to the observed count divided by the prior-centered capture probability — if the prior is off, the burden estimate is off by the same factor.

Editorial extensions

If this is right

  • If the model is right, Puerto Rico's passive surveillance captures only about one in five hospital-presenting leptospirosis cases, so the reported island-wide rate of 8–9 per 100,000 understates the true burden by roughly a factor of four.
  • The framework solves the motivating spatial misalignment: hospital-level capture probabilities can be linked to region-level aggregate counts through the travel-time and capacity weighting, so active surveillance need not be repeated in every region.
  • Diagnostic test accuracy is consequential for burden estimates: varying the sensitivity and specificity priors moves the island-wide rate from the main estimate of 35 per 100,000 up to roughly 40 or down to about 27–28 per 100,000, though the conclusion of substantial under-estimation persists.
  • Simulations show the correction removes most of the bias: at a true 10% capture probability the naive bias is near −90%, while the model's estimates are biased by only about −10% even with four hospitals, improving with more surveilled hospitals and more informative priors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the posterior capture probability (median 0.22) nearly coincides with the center of the α0 prior, the headline figure of 35 per 100,000 is effectively the observed count divided by a prior-centered capture fraction; if the true capture probability were 0.30, the corrected island-wide rate would fall closer to 25–26 per 100,000, and if it were 0.15 the rate would rise toward 50. Treat the p
  • The same machinery transfers directly to other hospital-based sentinel-surveillance diseases such as dengue, brucellosis, or undifferentiated acute febrile illness, wherever travel-time-to-care data exist; the binding constraint in each new setting will be an externally validated prior on passive capture probability.
  • The gravity-model weighting is directly testable: if hospital-level passive counts for a single region were obtained, one could check whether the travel-time-weighted capture mixture reproduces the observed distribution of case counts across hospitals; a mismatch would signal misspecified γ, δ.
  • The paper's own simulations with 7, 14, and 21 hospitals show the prior dependence and bias shrink as surveillance coverage expands, implying that adding sentinel hospitals — rather than more years at the same four — is the fastest route to a data-driven burden estimate.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper develops a Bayesian spatiotemporal capture–recapture model for integrating hospital-based active surveillance with regionally aggregated passive surveillance counts. The key methodological contributions are: (i) extending the Poisson-logistic model to accommodate spatial misalignment between hospital-level capture and region-level true-count processes via gravity-model hospital choice probabilities; (ii) incorporating imperfect diagnostic testing by marginalizing over individual disease-status indicators using a recursive Poisson-binomial algorithm; and (iii) providing posterior predictive distributions for the true hospital-presenting disease rate. The model is evaluated in a simulation study and applied to leptospirosis surveillance data from Puerto Rico, where it estimates island-wide true rates of 35 per 100,000 in 2022 and 2023, compared with observed passive-surveillance rates of 8 and 9 per 100,000. The paper reports that this provides 'strong evidence' that passive surveillance substantially underestimates the true burden.

Significance. If the methodological framework is accepted, the paper makes a useful contribution to an important practical problem: correcting spatially misaligned passive surveillance counts when active surveillance is available only at a subset of hospitals. The derivation of the marginal Poisson form with hospital-choice weights is careful, and the recursive exact computation of the testing likelihood is a genuine technical advance over earlier individual-level approaches. The simulation study is extensive, with multiple capture probabilities, prior informativeness levels, hospital counts, and test-accuracy priors, and the code is promised publicly. However, the central empirical claim — that the Puerto Rico data provide 'strong evidence' of substantial under-estimation — is not supported by the data themselves, because the passive-capture probability α0 is barely updated by the active surveillance data. The strength of this claim rests almost entirely on an informative prior whose location is not varied in any sensitivity analysis.

major comments (3)
  1. [§5.2 and Table S3.2] The posterior median for logit⁻¹(α0) is 0.22 (95% CI 0.12–0.40), which is essentially identical to the N(−1.4, 0.5²) prior (median ≈ 0.20, 95% interval ≈ 0.08–0.40). With only one passive–active overlap (n_PA = 1 in Table 1), the capture–recapture likelihood is almost flat in π_P, so the data barely update α0. Consequently, the estimated island-wide rate of 35 per 100,000 is approximately the observed rate divided by the prior-centered capture probability, rather than being driven by the active surveillance data. The claim in §5.2 and §6 that the model provides 'strong evidence' of substantial under-estimation is therefore not warranted; the evidence is only as strong as the untested prior on α0.
  2. [§5.3 and §4.1] The sensitivity analyses in the data application vary only the priors on test sensitivity and specificity, not the prior on α0. The simulation study (Section 4.1) also always centers the prior on the true α0 value. Thus the paper provides no assessment of how the main conclusion would change under a plausible alternative prior for the passive capture probability. For example, if the true average capture probability were 0.33 instead of 0.22, the estimated rate would be roughly 24 per 100,000; if it were 0.40, the rate would be roughly 20 per 100,000. Such a sensitivity analysis, or a substantial softening of the real-data conclusions, is needed to support the central empirical claim.
  3. [§3.1.3 and §5.1] The hospital choice probabilities π_C_{h,s} are treated as known, with γ and δ fixed at (2,0) in the data application. The authors state in §3.1.3 that 'the sensitivity of the results to reasonable values for γ and δ will be limited,' but no sensitivity analysis is provided for these parameters in the real-data setting. Since the burden estimates depend on the weighted combination of hospital-level capture probabilities, a brief sensitivity check over moderate γ values (or a citation to a prior analysis that supports this insensitivity) would strengthen the paper.
minor comments (4)
  1. [§3.4] Typo: 'Markov chain Mone Carlo' should be 'Markov chain Monte Carlo'.
  2. [§5.3] In the second sensitivity analysis paragraph, 'and 36 to 41 per 100,000 in 2022 (95% CI=21, 83)' should refer to 2023 (the first mention of 2022 is correct for that sentence).
  3. [Figure 6] The caption states 'by health region in 2023' but the text in §5.2 reports both 2022 and 2023 rate summaries. Clarify whether the figure shows only 2023 or both years.
  4. [§4.1] The discussion of UAFI generation says 'Thus, we expect the rates of additional UAFI cases to be much higher than that of leptospirosis.' It would help to state explicitly how this rate (600 per 100,000) relates to the leptospirosis rate used for the true counts, since the ratio affects the number of tested patients and hence the information in the active surveillance sample.

Circularity Check

0 steps flagged · score 1.0 of 10

No constructional circularity; the burden estimate is prior-sensitive but not definitionally forced by the model equations.

full rationale

The paper's derivation chain is mathematically self-contained. The marginal Poisson form for y (Eq. 3.2 and Supplement S1.1) follows from Poisson/multinomial/binomial thinning, and the posterior predictive for z−y (Eq. 3.12) follows from the same thinning given posterior draws. The capture-recapture likelihood (Eq. 3.5) and the marginalization over n^D and n^A (Eq. 3.10) are standard and correctly derived. No equation reduces to itself by construction, and no fitted parameter is renamed as a prediction: the active surveillance data (Table 1) are used to estimate capture probabilities, and the 2022–2023 aggregate counts are separate data used for the rate estimate. The central claim that observed rates under-estimate burden is not equivalent to the model's inputs because the posterior for logit^-1(α0) is slightly updated from the prior (0.22 vs ~0.20 prior median) and the prediction intervals reflect posterior uncertainty. The paper explicitly acknowledges that informative priors are necessary and that results are not entirely data-driven (Section 6), which is a transparent limitation rather than a circular step. The only self-citation, Ward, Brown and Oleson (2023), is used as a building block for the testing-data likelihood, but the paper derives and marginalizes the model itself; the citation is not load-bearing for the main empirical conclusion. The prior-sensitivity concern raised by the skew is real but falls under correctness/robustness, not circularity, because no formula equates the output to the prior by construction.

Assumptions & free parameters 6 free parameters · 9 assumptions · 0 invented entities

The burden estimates rest on nine stated background assumptions, several domain-specific to surveillance (independence, perfect passive specificity, no time trends) and several choices made by the authors (fixed gravity parameters, informative priors). The most consequential is the informative α0 prior, which dominates the real-data correction. No new physical entities are postulated.

free parameters (6)
  • gamma (distance decay) = 2
    Fixed in simulation and real analysis; not estimated. Controls gravity-model weights πᴼ_h,s; authors assert sensitivity is limited but do not test it in the data application (Sec. 3.1.3, 5.1).
  • delta (capacity exponent) = 1 (simulation), 0 (data)
    Fixed; set to 0 in the real analysis because hospital capacity data are incomplete. Directly changes hospital choice weights and hence the burden correction (Sec. 3.1.3, 5.1).
  • rho (GP range) = 0.06
    Fixed spatial correlation range for the hospital capture-probability GP in both simulation and data; not estimated; affects borrowing of strength across hospitals (Sec. 4.1, S3.2).
  • alpha0 prior mean and SD = N(-1.4, 0.5^2)
    Informativeness of this prior is required for identifiability; the posterior of the average capture probability almost coincides with the prior (0.22 vs ~0.20 mean), so the 4–5x correction is largely prior-determined (Sec. 5.1, Table S3.2).
  • piA prior = Beta(5,20)
    Prior on active surveillance enrollment probability chosen from study enrollment; posterior 0.14; affects decomposition of the capture-recapture likelihood (Sec. S3.2).
  • test sensitivity/specificity priors = IgM sens Beta(8.5,1.5), PCR sens Beta(8,2), IgM spec Beta(8.5,1.5), PCR spec Beta(9.5,0.5)
    Informative priors necessary for identifiability; sensitivity analyses varying these move the island rate between 27 and 40 per 100,000 (Sec. 5.3, Table S3.3).
assumptions (9)
  • standard math Poisson thinning: binomial thinning of a Poisson count yields a Poisson mean multiplied by the thinning probability; sums of independent Poissons are Poisson.
    Used to derive marginal distribution of y_s,t and z−y in S1.1 (Ross propositions 5.2, 5.4).
  • standard math GEV discrete-choice model yields multinomial logit choice probabilities.
    Justifies the gravity-like form of hospital choice probabilities in S1.2 (Train 2009).
  • standard math Barlow–Heidtmann recursion computes Poisson-binomial pmf exactly.
    Basis for tractable marginalization over latent disease indicators in S1.3.
  • domain assumption Active and passive surveillance operate independently conditional on disease presence at the hospital level.
    Needed for the multinomial capture-recapture likelihood in Eq. 3.5; if the systems are dependent (e.g., passive detection triggers active testing), estimates break. Stated in Sec. 3.2.
  • domain assumption Passive surveillance has perfect specificity; all passively captured cases are true positives.
    Assumed in Sec. 3.3 because accounting for passive false positives would require extra data; false positives would inflate capture counts and bias burden downward.
  • domain assumption Test results are conditionally independent given disease status; tested patients within a hospital-region-time bin are exchangeable.
    Used in the testing-data likelihood S1.3 and the uniform prior over disease-status vectors; violation (e.g., correlated tests) would bias sensitivity/specificity and n_A.
  • ad hoc to paper Hospital choice probabilities πᴼ are fixed and known from external travel-time and capacity data; γ and δ are fixed with no uncertainty.
    The real-data analysis sets γ=2, δ=0 because capacity data are incomplete; sensitivity is asserted, not demonstrated (Sec. 3.1.3, 5.1).
  • domain assumption Passive capture probabilities in 2022–2023 equal those estimated from the active-surveillance period (2019–2021); no time trends.
    Active data were collected July 2019–May 2021, but the correction is applied to 2022–23 regional counts; stated for simulations in Sec. 4.2, implicit in Sec. 5.
  • ad hoc to paper Informative priors on α0, πA, and sensitivity/specificity are sufficient for identifiability.
    The model is non-identifiable from the likelihood alone; priors are necessary and partly determine the numerical burden estimate (Sec. 3.2, 5.1, Discussion).

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Bayesian Spatiotemporal Model to Estimate Disease Burden Using Hospital-Based Active Surveillance." pith.science (2026). https://pith.science/paper/B337SLKK

@misc{pith2026260713185,
  author       = {Pith},
  title        = {Pith review of: A Bayesian Spatiotemporal Model to Estimate Disease Burden Using Hospital-Based Active Surveillance},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/B337SLKK}},
  note         = {Machine review of arXiv:2607.13185}
}
read the original abstract

Passive surveillance systems, in which data routinely collected by medical facilities are used to monitor the caseload of infectious diseases, are relatively straightforward to implement but often result in underestimation of the burden of disease due to under-diagnosis and imperfect testing. Targeted active surveillance can be used to correct these case counts to better reflect the true burden of disease. However, when the active surveillance effort is performed at a subset of hospitals and passive surveillance data is reported at an aggregated regional level, the resulting spatial misalignment must be reconciled to estimate the true rate of hospital-presenting disease at the spatial region level. Motivated by a recent active surveillance project for leptospirosis in four Puerto Rican hospitals, we address this challenge and develop a novel Bayesian spatio-temporal framework to better reflect the true number of hospital-presenting individuals with the disease. In particular, our method extends the Poisson-logistic framework to incorporate spatial heterogeneity in the probability of presenting to the hospitals across the study region. Our framework also accounts for imperfect diagnostic testing within the active surveillance data, addressing a common challenge for infectious diseases, particularly for neglected ones like leptospirosis. The model is assessed via simulation under various scenarios and then applied to the motivating leptospirosis data. Our approach offers a comprehensive framework for integrating spatially misaligned passive and active surveillance data, enabling better estimation of true disease burden.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

108 extracted references · 5 canonical work pages

  1. [1]

    Estimating health facility-level catchment populations using routine surveillance data and a

    Millar, Justin and Arambepola, Rohan and Cameron, Ewan and Hamainza, Busiku and Silumbe, Kafula and Miller, John and Bennett, Adam and Slater, Hannah , month = feb, year =. Estimating health facility-level catchment populations using routine surveillance data and a. doi:10.1101/2025.02.13.25322240 , abstract =

    work page doi:10.1101/2025.02.13.25322240 2025

  2. [2]

    International Journal of Health Geographics , author =

    Gravity models for potential spatial healthcare access measurement: a systematic methodological review , volume =. International Journal of Health Geographics , author =. 2023 , note =. doi:10.1186/s12942-023-00358-z , abstract =

    work page doi:10.1186/s12942-023-00358-z 2023

  3. [3]

    2018 , note =

    Malaria Journal , author =. 2018 , note =. doi:10.1186/s12936-018-2500-5 , abstract =

    work page doi:10.1186/s12936-018-2500-5 2018

  4. [4]

    Nature Medicine , author =

    Global maps of travel time to healthcare facilities , volume =. Nature Medicine , author =. 2020 , note =. doi:10.1038/s41591-020-1059-1 , abstract =

    doi:10.1038/s41591-020-1059-1 2020

  5. [5]

    Scientific Reports , author =

    A regional analysis of healthcare utilization trends during consecutive disasters in. Scientific Reports , author =. 2025 , note =. doi:10.1038/s41598-025-89983-1 , abstract =

    work page doi:10.1038/s41598-025-89983-1 2025

  6. [6]

    , year =

    Train, Kenneth E. , year =. Discrete. doi:10.1017/CBO9780511805271 , file =

    doi:10.1017/cbo9780511805271

  7. [7]

    https://arxiv.org/pdf/0903.5480 , url =

    arXiv

  8. [8]

    The Annals of Statistics , author =

    The pseudo-marginal approach for efficient. The Annals of Statistics , author =. 2009 , note =. doi:10.1214/07-AOS574 , abstract =

    doi:10.1214/07-aos574 2009

Show all 108 references
  1. [9]

    Genetics , author =

    Estimation of. Genetics , author =. 2003 , pages =. doi:10.1093/genetics/164.3.1139 , abstract =

    2003 doi

  2. [10]

    Biometrics , volume=

    An individual level infectious disease model in the presence of uncertainty from multiple, imperfect diagnostic tests , author=. Biometrics , volume=. 2023 , publisher=

    2023

  3. [11]

    Computational Statistics & Data Analysis , volume=

    On computing the distribution function for the Poisson binomial distribution , author=. Computational Statistics & Data Analysis , volume=. 2013 , publisher=

    2013

  4. [12]

    Barlow, R. E. and Heidtmann, K. D. , journal=. Computing k-out-of-n System Reliability , year=

  5. [13]

    , author=

    The No-U-Turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo. , author=. J. Mach. Learn. Res. , volume=

  6. [14]

    The annals of statistics , volume=

    Slice sampling , author=. The annals of statistics , volume=. 2003 , publisher=

    2003

  7. [15]

    Biometrics , volume=

    A hierarchical model for analyzing multisite individual-level disease surveillance data from multiple systems , author=. Biometrics , volume=. 2023 , publisher=

    2023

  8. [16]

    2005 , publisher=

    Gaussian Markov random fields: theory and applications , author=. 2005 , publisher=

    2005

  9. [17]

    Journal of Tropical Medicine , volume=

    Diagnosis and Clinical Description of Leptospirosis Cases Identified Through Hospital-Based Active Surveillance in Puerto Rico, 2019--2021 , author=. Journal of Tropical Medicine , volume=. 2025 , publisher=

    2019

  10. [18]

    Journal of the American statistical Association , volume=

    Strictly proper scoring rules, prediction, and estimation , author=. Journal of the American statistical Association , volume=. 2007 , publisher=

    2007

  11. [19]

    Science of The Total Environment , volume=

    Rainfall-driven resuspension of pathogenic Leptospira in a leptospirosis hotspot , author=. Science of The Total Environment , volume=. 2024 , publisher=

    2024

  12. [20]

    Spatial and spatio-temporal epidemiology , volume=

    A spatial hierarchical model for integrating and bias-correcting data from passive and active disease surveillance systems , author=. Spatial and spatio-temporal epidemiology , volume=. 2020 , publisher=

    2020

  13. [21]

    Pathogens , volume=

    Leptospirosis and the environment: a review and future directions , author=. Pathogens , volume=. 2023 , publisher=

    2023

  14. [22]

    PLoS neglected tropical diseases , volume=

    Global morbidity and mortality of leptospirosis: a systematic review , author=. PLoS neglected tropical diseases , volume=. 2015 , publisher=

    2015

  15. [23]

    Summary of notifiable infectious diseases and conditions—United States, 2015 , author=. MMWR. Morbidity and mortality weekly report , volume=

    2015

  16. [24]

    Ramos, Luis Javier Santiago and L. An. GeoFocus. International Review of Geographical Information Science and Technology , number=

  17. [25]

    2003 , publisher=

    Hierarchical modeling and analysis for spatial data , author=. 2003 , publisher=

    2003

  18. [26]

    Statistics and its Interface , volume=

    Estimating the sizes of populations at risk of HIV infection from multiple data sources using a Bayesian hierarchical model , author=. Statistics and its Interface , volume=

  19. [27]

    Annals of Epidemiology , volume=

    Modeling of under-detection of cases in disease surveillance , author=. Annals of Epidemiology , volume=. 2005 , publisher=

    2005

  20. [28]

    Statistics in medicine , volume=

    A random-censoring Poisson model for underreported data , author=. Statistics in medicine , volume=. 2017 , publisher=

    2017

  21. [29]

    Journal of Computational and Graphical Statistics , volume=

    Programming with models: writing statistical algorithms for general model structures with NIMBLE , author=. Journal of Computational and Graphical Statistics , volume=. 2017 , publisher=

    2017

  22. [30]

    Journal of the American Statistical Association , year=

    A hierarchical framework for correcting under-reporting in count data , author=. Journal of the American Statistical Association , year=

  23. [31]

    , title =

    Winkelmann, Rainer and Zimmermann, Klaus F. , title =. 1993 , type =

    1993

  24. [32]

    The Annals of Applied Statistics , volume=

    A Bayesian model of underreporting for sexual assault on college campuses , author=. The Annals of Applied Statistics , volume=. 2024 , publisher=

    2024

  25. [33]

    Bayesian Analysis , volume=

    Bias correction in clustered underreported data , author=. Bayesian Analysis , volume=. 2022 , publisher=

    2022

  26. [34]

    Statistical Modelling , volume=

    Sparse Bayesian modelling of underreported count data , author=. Statistical Modelling , volume=. 2016 , publisher=

    2016

  27. [35]

    Veterinary research , volume=

    Evaluation and optimization of surveillance systems for rare and emerging infectious diseases , author=. Veterinary research , volume=

  28. [36]

    Wildlife monographs , number=

    Statistical inference from capture data on closed animal populations , author=. Wildlife monographs , number=. 1978 , publisher=

    1978

  29. [37]

    Biometrics , volume=

    N-mixture models for estimating population size from spatially replicated counts , author=. Biometrics , volume=. 2004 , publisher=

    2004

  30. [38]

    medRxiv , pages=

    Estimating health facility-level catchment populations using routine surveillance data and a Bayesian gravity model , author=. medRxiv , pages=. 2025 , publisher=

    2025

  31. [39]

    Social science & medicine , volume=

    Measuring potential physical accessibility to general practitioners in rural areas: a method and case study , author=. Social science & medicine , volume=. 1982 , publisher=

    1982

  32. [40]

    Proceedings of the royal society of london

    A contribution to the mathematical theory of epidemics , author=. Proceedings of the royal society of london. Series A, Containing papers of a mathematical and physical character , volume=. 1927 , publisher=

    1927

  33. [41]

    The Annals of Applied Statistics , volume=

    Phenomenological forecasting of disease incidence using heteroskedastic Gaussian processes: A dengue case study , author=. The Annals of Applied Statistics , volume=

  34. [42]

    PLOS Computational Biology , volume=

    Informing policy via dynamic models: Cholera in Haiti , author=. PLOS Computational Biology , volume=. 2024 , publisher=

    2024

  35. [43]

    Malaria in Northwest India: Data analysis via partially observed stochastic differential equation models driven by L

    Bhadra, Anindya and Ionides, Edward L and Laneri, Karina and Pascual, Mercedes and Bouma, Menno and Dhiman, Ramesh C , journal=. Malaria in Northwest India: Data analysis via partially observed stochastic differential equation models driven by L. 2011 , publisher=

    2011

  36. [44]

    Statistica Sinica , volume=

    Inference for individual-level models of infectious diseases in large populations , author=. Statistica Sinica , volume=

  37. [45]

    2002 , publisher =

    Foot and Mouth Disease 2001: Lessons to be Learned Inquiry Report , author =. 2002 , publisher =

    2001

  38. [46]

    Biometrics , volume=

    Statistical inference in a stochastic epidemic SEIR model with control intervention: Ebola as a case study , author=. Biometrics , volume=. 2006 , publisher=

    2006

  39. [47]

    Infectious Disease Modelling , volume=

    Bayesian modeling of dynamic behavioral change during an epidemic , author=. Infectious Disease Modelling , volume=. 2023 , publisher=

    2023

  40. [48]

    Clinical microbiology reviews , volume=

    Melioidosis: epidemiology, pathophysiology, and management , author=. Clinical microbiology reviews , volume=. 2005 , publisher=

    2005

  41. [49]

    Clinical Infectious Diseases , volume=

    Coccidioidomycosis , author=. Clinical Infectious Diseases , volume=. 2005 , publisher=

    2005

  42. [50]

    Statistical Science , pages=

    A general framework for the parametrization of hierarchical models , author=. Statistical Science , pages=. 2007 , publisher=

    2007

  43. [51]

    Biometrics , volume=

    A linear noise approximation for stochastic epidemic models fit to partially observed incidence counts , author=. Biometrics , volume=. 2022 , publisher=

    2022

  44. [52]

    2014 , publisher=

    Introduction to probability models , author=. 2014 , publisher=

    2014

  45. [53]

    1995 , publisher=

    Bayesian data analysis , author=. 1995 , publisher=

    1995

  46. [54]

    Chipman and Edward I

    Hugh A. Chipman and Edward I. George and Robert E. McCulloch , journal =

  47. [55]

    The Journal of Machine Learning Research , volume=

    Variational inference for latent variables and uncertain inputs in Gaussian processes , author=. The Journal of Machine Learning Research , volume=. 2016 , publisher=

    2016

  48. [56]

    arXiv preprint arXiv:2510.10268 , year=

    Neural variational inference for cutting feedback during uncertainty propagation , author=. arXiv preprint arXiv:2510.10268 , year=

  49. [57]

    Proceedings of the thirteenth international conference on artificial intelligence and statistics , pages=

    Elliptical slice sampling , author=. Proceedings of the thirteenth international conference on artificial intelligence and statistics , pages=. 2010 , organization=

    2010

  50. [58]

    Statistics and Computing , volume=

    Cuts in Bayesian graphical models , author=. Statistics and Computing , volume=. 2015 , publisher=

    2015

  51. [59]

    Journal of the American statistical Association , volume=

    Variational inference: A review for statisticians , author=. Journal of the American statistical Association , volume=. 2017 , publisher=

    2017

  52. [60]

    Statistical Science , volume=

    Variational inference for cutting feedback in misspecified models , author=. Statistical Science , volume=. 2023 , publisher=

    2023

  53. [61]

    Artificial intelligence and statistics , pages=

    Deep gaussian processes , author=. Artificial intelligence and statistics , pages=. 2013 , organization=

    2013

  54. [62]

    Journal of Applied Meteorology , volume=

    Estimates of root-mean-square random error for finite samples of estimated precipitation , author=. Journal of Applied Meteorology , volume=

  55. [63]

    British Actuarial Journal , volume=

    Loss modelling from first principles , author=. British Actuarial Journal , volume=. 2024 , publisher=

    2024

  56. [64]

    PloS one , volume=

    Diagnosis of leptospirosis: comparison between microscopic agglutination test, IgM-ELISA and IgM rapid immunochromatography test , author=. PloS one , volume=. 2015 , publisher=

    2015

  57. [65]

    BMC infectious diseases , volume=

    Diagnosis of human leptospirosis: systematic review and meta-analysis of the diagnostic accuracy of the Leptospira microscopic agglutination test, PCR targeting Lfb1, and IgM ELISA to Leptospira fainei serovar Hurstbridge , author=. BMC infectious diseases , volume=. 2024 , publisher=

    2024

  58. [66]

    Microbiology Spectrum , volume=

    Urine PCR testing as an effective method for early diagnosis of leptospirosis , author=. Microbiology Spectrum , volume=. 2025 , publisher=

    2025

  59. [67]

    2019 , publisher =

    Hospitals of. 2019 , publisher =

    2019

  60. [68]

    Artificial intelligence and statistics , pages=

    Handling sparsity via the horseshoe , author=. Artificial intelligence and statistics , pages=. 2009 , organization=

    2009

  61. [69]

    Sparsity information and regularization in the horseshoe and other shrinkage priors , author=

  62. [70]

    Bayesian analysis , volume=

    Locally adaptive smoothing with Markov random fields and shrinkage priors , author=. Bayesian analysis , volume=

  63. [71]

    , title =

    Lin, Ying and Mitchell, Kenneth E. , title =. 19th Conference on Hydrology , year =

  64. [72]

    PLOS Global Public Health , volume=

    Quantifying the relationship between climatic indicators and leptospirosis incidence in Fiji: A modelling study , author=. PLOS Global Public Health , volume=. 2023 , publisher=

    2023

  65. [73]

    Earth system science data , volume=

    Improving intelligent dasymetric mapping population density estimates at 30 m resolution for the conterminous United States by excluding uninhabited areas , author=. Earth system science data , volume=. 2022 , publisher=

    2022

  66. [74]

    Biosensors , volume=

    Point of care diagnostics in resource-limited settings: A review of the present and future of PoC in its most needed environment , author=. Biosensors , volume=. 2020 , publisher=

    2020

  67. [75]

    New England Journal of Medicine , volume=

    Syphilis complicating pregnancy and congenital syphilis , author=. New England Journal of Medicine , volume=. 2024 , publisher=

    2024

  68. [76]

    Scenario Projections of Congenital Syphilis Cases in California, 2023--2030 , author=

    P-1394. Scenario Projections of Congenital Syphilis Cases in California, 2023--2030 , author=. Open Forum Infectious Diseases , volume=. 2025 , organization=

    2023

  69. [77]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Particle markov chain monte carlo methods , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2010 , publisher=

    2010

  70. [78]

    Sexual health , volume=

    Syphilis transmission: a review of the current evidence , author=. Sexual health , volume=. 2015 , publisher=

    2015

  71. [79]

    Nature climate change , volume=

    Less than 2 C warming by 2100 unlikely , author=. Nature climate change , volume=. 2017 , publisher=

    2017

  72. [80]

    Biostatistics , volume=

    Health-exposure modeling and the ecological fallacy , author=. Biostatistics , volume=. 2006 , publisher=

    2006

  73. [81]

    Current opinion in virology , volume=

    Transmission routes of respiratory viruses among humans , author=. Current opinion in virology , volume=. 2018 , publisher=

    2018

  74. [82]

    The Lancet Planetary Health , volume=

    Ecological and socioeconomic factors associated with the human burden of environmentally mediated pathogens: a global analysis , author=. The Lancet Planetary Health , volume=. 2022 , publisher=

    2022

  75. [83]

    The annals of applied statistics , volume=

    Predictive modeling of cholera outbreaks in Bangladesh , author=. The annals of applied statistics , volume=

  76. [84]

    International Journal of Biometeorology , volume=

    Prediction of leptospirosis outbreaks by hydroclimatic covariates: a comparative study of statistical models , author=. International Journal of Biometeorology , volume=. 2022 , publisher=

    2022

  77. [85]

    PLoS One , volume=

    Time series models for prediction of leptospirosis in different climate zones in Sri Lanka , author=. PLoS One , volume=. 2021 , publisher=

    2021

  78. [86]

    Scientific data , volume=

    High resolution global gridded data for use in population studies , author=. Scientific data , volume=. 2017 , publisher=

    2017

  79. [87]

    2020 , institution =

    GPM Integrated Multi-satellitE Retrievals for GPM (IMERG) Algorithm Theoretical Basis Document (ATBD) Version 06 , author =. 2020 , institution =

    2020

  80. [88]

    American journal of epidemiology , volume=

    A new framework and software to estimate time-varying reproduction numbers during epidemics , author=. American journal of epidemiology , volume=. 2013 , publisher=

    2013

  81. [89]

    2009 , publisher=

    Probabilistic graphical models: principles and techniques , author=. 2009 , publisher=

    2009

  82. [90]

    2017 , publisher=

    Elements of causal inference: foundations and learning algorithms , author=. 2017 , publisher=

    2017

  83. [91]

    International encyclopedia of public health , pages=

    Infectious disease surveillance , author=. International encyclopedia of public health , pages=

  84. [92]

    2021 , note =

    Max Roser and Hannah Ritchie and Fiona Spooner , title =. 2021 , note =

    2021

  85. [93]

    Frontiers in Tropical Diseases , volume=

    Neglected infectious diseases in the Americas: current situation and perspectives for the control and elimination by 2030 , author=. Frontiers in Tropical Diseases , volume=. 2024 , publisher=

  86. [94]

    Clinical Medicine , volume=

    Leptospirosis: clinical aspects , author=. Clinical Medicine , volume=. 2022 , publisher=

    2022

  87. [95]

    2025 , note =

    Poverty Status in the Past 12 Months: 2024 American Community Survey 1-Year Estimates , howpublished =. 2025 , note =

    2024

  88. [96]

    Cureus , volume=

    Congenital syphilis epidemiology, prevention, and management in the United States: a 2022 update , author=. Cureus , volume=

    2022

  89. [97]

    Frontiers in medicine , volume=

    Global, regional, and national burden of syphilis, 1990--2021 and predictions by Bayesian age-period-cohort analysis: a systematic analysis for the global burden of disease study 2021 , author=. Frontiers in medicine , volume=. 2024 , publisher=

    1990

  90. [98]

    American Journal of Public Health , volume=

    Congenital Syphilis: From Near-Elimination to Crisis , author=. American Journal of Public Health , volume=. 2026 , publisher=

    2026

  91. [99]

    Statistical science , volume=

    Inference from iterative simulation using multiple sequences , author=. Statistical science , volume=. 1992 , publisher=

    1992

  92. [100]

    , author=

    Active surveillance and risk factors for leptospirosis in Hawaii. , author=

  93. [101]

    Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics , pages =

    XBART: Accelerated Bayesian Additive Regression Trees , author =. Proceedings of the Twenty-Second International Conference on Artificial Intelligence and Statistics , pages =. 2019 , editor =

    2019

  94. [102]

    Chipman and Edward I

    Hugh A. Chipman and Edward I. George and Robert E. McCulloch , doi =. The Annals of Applied Statistics , keywords =. 2010 , bdsk-url-1 =

    2010

  95. [103]

    Annals of statistics , pages=

    Greedy function approximation: a gradient boosting machine , author=. Annals of statistics , pages=. 2001 , publisher=

    2001

  96. [104]

    Journal of statistical software , volume=

    Stan: A probabilistic programming language , author=. Journal of statistical software , volume=

  97. [105]

    Journal of the American Statistical Association , volume=

    Log-linear Bayesian additive regression trees for multinomial logistic and count regression models , author=. Journal of the American Statistical Association , volume=. 2021 , publisher=

    2021

  98. [106]

    Journal of the American Statistical Association , volume=

    Generalized Bayesian additive regression trees models: Beyond conditional conjugacy , author=. Journal of the American Statistical Association , volume=. 2025 , publisher=

    2025

  99. [107]

    Biometrika , volume=

    Reversible jump Markov chain Monte Carlo computation and Bayesian model determination , author=. Biometrika , volume=. 1995 , publisher=

    1995

  100. [108]

    2026 , howpublished =

    Strong, Brent , title =. 2026 , howpublished =

    2026

Pith tools

Reviewed August 2, 2026 · model on record in the stance chip above.