Pith. sign in

REVIEW 4 major objections 6 minor 17 references

Linking climate and dengue in the Philippines using a two-stage Bayesian spatio-temporal model

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Temperature raises dengue risk in the Philippines up to a heat threshold, while rainfall's effect reverses by region — positive in the dry-season west, negative in the uniformly wet east — per a two-stage Bayesian model that propagates…

desk verdict The paper's rainfall-by-climate-type interaction loses significance under the resampling method the authors themselves prefer, yet the abstract presents it as established; the temperature curve also peaks near 20°C, making 'positively associated' a stretch for the Philippines. read the letter →

arxiv 2506.22334 v1 pith:JPQ5HJO5 submitted 2025-06-27 stat.AP

Stephen Jun Villejo , Sara Martino , Janine Illian This is my paper · ORCID
classification stat.AP MSC 62P1062M3062F15
keywords denguePhilippinestwo-stageBayesianmodellingINLAspatialepidemiologyrainfall-dengueassociationnon-lineartemperatureeffectuncertaintypropagation
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

Temperature drives dengue risk in the Philippines upward up to a heat threshold, beyond which very hot conditions push risk back down, the paper argues. It also claims the rainfall–dengue association is not one sign but two, split by climate regime: in the western Philippines, where dry and wet seasons are pronounced, rainfall is positively associated with dengue; in the east, where rain is abundant all year, rainfall is negatively associated. These claims come from a two-stage Bayesian spatio-temporal model that fuses weather-station and forecast-model climate data into province-level exposures, then regresses monthly dengue counts on those exposures while carrying the climate-model uncertainty through posterior sampling and model averaging. A reader should care because the direction of the rainfall effect determines whether wet-season forecasts signal outbreak risk or relief, and because the temperature curve implies that the effect of warming on dengue is not monotone.

What carries the argument

The argument is carried by a two-stage spatial-misalignment pipeline. The first stage is a data fusion model that treats weather-station measurements and Global Spectral Model forecasts as noisy versions of one latent spatio-temporal climate process, represented with SPDE/INLA Matérn fields. Province-level climate exposures are then computed by the block-average formula (Equation 3), an unweighted mean of predicted climate values over grid points inside each province. The second stage is a Poisson disease-mapping model for monthly dengue counts whose random effects combine a scaled BYM2 spatial component, a random-walk-2 time component, and a Type II space–time interaction, all defined over an iCAR graph made proper for the archipelago's disconnected islands. Uncertainty from the first stage is propagated by drawing posterior samples of the climate process, refitting the health model on each sample, and averaging the resulting posteriors (Algorithm 2).

What would settle it

Exclude the COVID-19-affected counts of 2020 and re-estimate the health model with the resampling uncertainty propagation, on both the data-fusion and stations-only climate inputs. If the negative temperature-squared coefficient or the east/west rainfall interaction loses its sign or its credible interval under either input, the central climate–dengue associations are artifacts of the epidemic and reporting-collapse years rather than stable climate effects.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the climate–dengue link in the Philippines is both non-linear in temperature and spatially conditional for rainfall. The fitted health model finds a positive temperature coefficient and a negative temperature-squared coefficient, so transmission risk rises with warmth but falls under extremely hot conditions. For rainfall, the paper splits the country by climate type: in the western section, with a pronounced dry and wet season, rainfall is positively associated with dengue; in the eastern section, wet all year round, rainfall is negatively associated with dengue, consistent with steady rain flushing mosquito breeding sites rather than creating them. Relative humidity shows the same east/west contrast. The paper further reports that after these climate effects and covariates, a substantial amount of structured spatial and temporal variation in dengue remains unexplained, and that the posterior resampling approach generally widens parameter uncertainty and pulls estimates toward the null compared with the plug-in approach.

Load-bearing premise

The load-bearing premise is that the unweighted province-level means of the first-stage climate predictions are unbiased measures of true climate exposure, and that the 2019–2020 estimation window — which contains a record epidemic and the COVID-19 reporting collapse — is representative; if the underlying climate surfaces are biased or the window is unrepresentative, every climate–dengue coefficient inherits the error.

Editorial extensions

If this is right

  • Public-health agencies can translate rainfall forecasts into dengue early warnings only after conditioning on climate type: wetter months signal higher risk in the western Philippines and lower risk in the eastern Philippines.
  • Projected warming will not raise dengue burden monotonically: the quadratic temperature effect implies risk peaks at an intermediate temperature and declines under extreme heat.
  • The sizeable structured variation left after climate adjustment means climate-based early-warning systems should be paired with local vector and reporting indicators rather than used alone.
  • The two-stage pipeline with posterior-sampling uncertainty propagation is a transferable template for linking point-referenced climate or pollution data to areal health outcomes in data-sparse archipelagos.

Reading between the lines

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

  • A sharper mechanistic test would replace total monthly rainfall with a wet-day-frequency or consecutive-dry-days index at province level: the flushing story in the east and the sporadic-breeding story in the west make opposite predictions about which rainfall metric drives dengue, a contrast total rainfall cannot resolve.
  • Because the covid indicator is non-significant while the random-walk time effect absorbs the 2020 decline, the model identifies the pandemic's effect only through its timing; dropping 2020 entirely would show how much of the temperature and rainfall coefficients depends on that single anomalous year.
  • The systematic attenuation of resampled posterior means toward the null suggests that earlier single-stage Philippine climate–dengue studies that treat predicted climate covariates as known likely understate uncertainty; this caution generalizes beyond dengue to any areal-health analysis fed by predicted exposures.
  • Population-weighted block averages would be a natural variant of Equation (3), since dengue burden concentrates in dense urban areas where unweighted grid means may misrepresent the exposure that actually drives transmission; whether the east/west contrast survives that reweighting is an open check.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. This paper proposes a two-stage Bayesian spatio-temporal model linking monthly provincial dengue counts in the Philippines (2016–2020) to climate variables. In the first stage, climate surfaces are obtained from previously published stations-only and data-fusion models; in the second stage, a Poisson disease-mapping model with temperature, rainfall, relative humidity, climate type, and spatio-temporal random effects is fitted with INLA. Uncertainty in climate exposures is propagated by comparing a plug-in approach with a posterior-resampling approach. The reported findings are a non-linear positive temperature effect that turns negative at high temperatures, a rainfall–dengue association that differs between eastern and western climate zones, and substantial residual spatio-temporal variation.

Significance. The paper addresses a relevant public-health question and demonstrates a computationally practical INLA-based two-stage framework with explicit handling of spatial misalignment and disconnected spatial graphs. The transparent presentation of both plug-in and resampling results, including the sensitivity of the rainfall interaction, is a strength. If the rainfall-by-climate-type finding were robust, the paper would make a notable empirical contribution; however, the paper's own uncertainty-propagating analysis does not support that headline claim. The more defensible contributions are the temperature non-linearity and the relative-humidity interaction, which are significant under both methods, and the methodological template.

major comments (4)
  1. [Abstract; Section 6.1; Table 2; Section 7] The abstract and Section 7 state as an established result that rainfall is negatively associated with dengue in uniform-rainfall areas and positively associated in areas with pronounced dry/wet seasons. This claim rests on γ5 (log Rain × ClimateType) in Table 2. Under the plug-in method the 90% credible interval is (-0.1786, -0.0075), but under the resampling method—the approach the authors advocate for propagating first-stage uncertainty—the 90% interval is (-0.1561, 0.0185), which includes zero. Section 6.1 explicitly says the resampling approach 'causing it to be no longer significant.' The abstract and conclusions should be revised to present the rainfall interaction as suggestive only, or the analysis should provide a principled reason why the plug-in result should be preferred.
  2. [Section 2; Section 4.1; Section 6] The time window of the main analysis is not stated clearly. Section 2 says the dengue data are monthly counts from 2016 to 2020, while the data-fusion climate predictions used for the main results cover only 2019–2020. Section 6 presents the data-fusion-based results as the main results but never specifies T for this model. If the main health model is fit on 2019–2020 only, those two years contain a major national dengue epidemic and the COVID-19-related reporting changes, and the conclusions in Section 7 about general climate–dengue associations are not supported by the main analysis. Please state the actual time period used in each model and discuss the representativeness of this window, or make the longer stations-only analysis (currently in the Supplementary Material) the primary analysis.
  3. [Equation (11); Section 5.2] There is a scaling error in the data-fusion block-average formula. The weights w_ℓ in Equation (10) are normalized so that Σ_ℓ w_ℓ = 1, yet Equation (11) multiplies the weighted sum by an additional factor 1/L, so the block average is (1/L) times the intended weighted average. The same extraneous 1/L appears in the resampling expression below Equation (11). This changes the scale of all climate exposures entering the health model and should be corrected or explicitly justified.
  4. [Section 5.2; Algorithm 2; Equation (9)] The uncertainty-propagation step is described as 'Bayesian model averaging,' but Equation (9) averages posterior distributions over samples from a single first-stage model; this is posterior averaging over imputations, not BMA over models. The manuscript should clarify the terminology and justify the equal-weight averaging of INLA posterior densities. It should also describe how joint posterior samples of the first-stage latent field and hyperparameters are generated (e.g., via inla.posterior.sample) and how many samples J are used, since the resampling credible intervals in Tables 2 and 4 depend on this implementation.
minor comments (6)
  1. [Equation (12)] The notation is garbled; the terms involving log Rain should be written as γ3 log(Rain(B_i,t)) + γ4 ClimateType(B_i,t) + γ5 log(Rain(B_i,t)) × ClimateType(B_i,t), with matching parentheses.
  2. [Figure 6 caption] Panel (c) is labelled 'γ3, log Rain × ClimateType' but the parameter is γ5.
  3. [Section 7] The statement that 'the posterior uncertainties were stable after around 12 resamples' is not supported by any formal convergence check; please provide a plot or numeric criterion.
  4. [Section 5.2] The term 'Bayesian model averaging' is used in two senses—for averaging over α1 values in Equation (10) and for averaging over posterior samples in Equation (9). Please distinguish these with different terms.
  5. [Section 6.1] The statement about the stations-only model ('the coefficient of log rainfall is significant') should be accompanied by the relevant supplementary table number so the reader can verify it.
  6. [Section 4.2; Equation (3)] The unweighted mean over prediction points ignores any population or area weighting; the authors should discuss the potential ecological-bias implications of this choice.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the two-stage climate-to-dengue chain is not self-defined, although the rainfall-dengue headline is weakened by the authors' own resampling-based credible interval.

full rationale

The derivation chain is linear and independent: first-stage climate models from Villejo et al. (2025) produce point predictions, Equation (3) forms unweighted block averages, and these averages enter the dengue GLMM (Equations (1) and (12)) as fixed covariates with dengue counts as the outcome. The first-stage model is fitted to weather-station and GSM data only and never sees dengue counts, so no health-model coefficient is defined by construction from a fitted constant. The headline east-west rainfall difference is the posterior estimate of the interaction coefficient gamma5; it is an estimated parameter, not a renamed input. The paper itself reports that under the resampling method the 90% credible interval for gamma5 is (-0.1561, 0.0185), which contains zero (Section 6.1, Table 2); this makes the abstract's rainfall claim overstated, but overstatement is a correctness and interpretation issue, not circularity. The citation to Villejo et al. (2025) is a self-citation, but it is a separate published climate model with its own leave-group-out cross-validation and does not include the dengue outcome, so it constitutes independent support rather than a load-bearing self-citation. No self-definitional reduction, fitted-parameter-renamed-as-prediction, imported uniqueness theorem, or ansatz-smuggled-via-citation pattern was found.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

No new physical entities are introduced. The model relies on standard statistical assumptions and on climate predictions produced in the authors' earlier work. The free parameters listed are modeling choices that shape the results rather than quantities fitted to make a derivation work.

free parameters (3)
  • ClimateType east/west split
    A hand-chosen binary covariate dividing the Philippines into eastern and western climate zones; it defines the interaction through which the rainfall-dengue relationship is claimed to vary.
  • Quadratic temperature functional form
    Temperature and temperature-squared are entered as linear predictors (Equation 12); the 'extreme heat reduces risk' conclusion is a direct consequence of choosing this form.
  • Number of posterior resamples J = 12-15
    The resampling method averages over J posterior samples; the paper reports stability after about 12 resamples, but the choice is computational and affects the posterior standard deviations.
assumptions (5)
  • domain assumption Poisson distribution for monthly dengue counts with log link and internal-standardization offset
    Equation (1) assumes counts follow a Poisson with mean equal to expected cases times risk; internal standardization uses the same national data to compute expected cases, which is a standard but not assumption-free approach.
  • domain assumption First-stage climate model predictions are unbiased
    The health model uses block means of predictions from Villejo et al. (2025); if these predictions are biased, the climate-dengue coefficients are biased.
  • domain assumption 2019-2020 data fusion window is representative
    The main analysis uses the years covered by the data fusion model, which include the 2019 epidemic and the 2020 COVID-19 reporting decline; representativeness is assumed.
  • standard math iCAR scaling on disconnected graphs
    Section 4.4 applies Freni-Sterrantino et al. (2018) scaling to make precisions comparable across disconnected island graphs; this is a standard methodological tool.
  • standard math INLA approximation is accurate
    All inference uses the integrated nested Laplace approximation, which is approximate for latent Gaussian models; the paper relies on its known accuracy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Linking climate and dengue in the Philippines using a two-stage Bayesian spatio-temporal model." pith.science (2026). https://pith.science/paper/JPQ5HJO5

@misc{pith2026250622334,
  author       = {Pith},
  title        = {Pith review of: Linking climate and dengue in the Philippines using a two-stage Bayesian spatio-temporal model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JPQ5HJO5}},
  note         = {Machine review of arXiv:2506.22334}
}
read the original abstract

Dengue is an infectious disease which poses significant socioeconomic and disease burden in many tropical and subtropical regions of the world. This work aims to provide additional insight into the association between dengue and climate in the Philippines. We employ a two-stage modelling framework: the first stage fits climate models, while the second stage fits a health model that uses the climate predictions from the first stage as inputs. We postulate a Bayesian spatio-temporal model and use the integrated nested Laplace approximation (INLA) approach for inference. To account for the uncertainty in the climate models, we perform posterior sampling and then perform Bayesian model averaging to compute the final posterior estimates of second-stage model parameters. The results indicate that temperature is positively associated with dengue, although extremely hot conditions tend to have a negative effect. Moreover, the relationship between rainfall and dengue varies in space. In areas with uniform amounts of rainfall all year round, rainfall is negatively associated with dengue. In contrast, in regions with pronounced dry and wet season, rainfall shows a positive association with dengue. Finally, there remains unexplained structured variation in space and time after accounting for the impact of climate variables and other covariates.

Figures

Figures reproduced from arXiv: 2506.22334 by the authors.

Figure 1
Figure 1. Time series plot of the number of dengue cases in the Philippines from January 2016 to January 2021 introduce higher biases (Lawson et al., 2016). In climate prediction, commonly used supplementary data sources include global circulation models (GCMs) and numerical weather prediction models (Naish et al., 2014). The process of integrating multiple data sources to improve predictions is known as data fusion or data a… view at source ↗
Figure 2
Figure 2. Plot of total dengue cases yearly (2016–2020) in the Philippines 3 [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Plot of standardized incidence ratios (SIR) of dengue in the Philippines from August 2019 to November 2019 2019. In August 2019, the Philippines declared a national dengue alert and epidemic due to a surge in cases and deaths (BBC, 2019). During 2020, coinciding with the onset of the COVID-19 pandemic, the number of reported cases is very low. This global phenomenon (WHO, 2023a) was attributed to both reduced mobili… view at source ↗
Figures from the paper (17 more)
Figure 4
Figure 4. Figure 4: Climate data sources for Philippines: (a) 57 weather synoptic stations (b) Global Spectral Model (GSM), a numerical weather prediction model maintained by the Japan Meteorological Agency Temperature affects the extrinsic incubation period of the pathogen, as well as th…
Figure 5
Figure 5. Figure 5: (a) prediction grid, (b) 19 disconnected graphs for the iCAR model (out of the 19 graphs, 12 are singletons / isolated islands), (c) mesh used for the SPDE approximation 4.3 Spatio-temporal effects φ(Bi , t) We assume the following form for the spatio-temporal effects:…
Figure 6
Figure 6. Figure 6: Plot showing the posterior means and 90% credible intervals for the following parameters: (a) γ1 (b) γ2 (c) γ5; for the model with temperature and log rainfall as climate covariates. The first vertical line shows the estimates for the plug-in method, while the rest of …
Figure 7
Figure 7. Figure 7: Comparison of (a) posterior mean and (b) posterior standard deviation of the space effects ψ(Bi) between the plug-in method and resampling method, for the dengue model with temperature and log rainfall as climate covariate (a) Temperature and log rainfall as covariates…
Figure 8
Figure 8. Figure 8: Plot of the estimated structured time effects ν(t) with the 95% credible intervals between the plug-in method and resampling method: (a) temperature and log rain as climate covariates (b) relative humidity as covariate [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Estimated space-time interaction effect υ(Bi , t) for five provinces. Four of them are contiguous provinces which exhibit the same temporal structure pre-pandemic, and which also agrees with the trend in the SIRs. The fifth province (located in the north) has a decreas…
Figure 10
Figure 10. Figure 10: Comparison of classical SIR estimates and model-based SIR estimates from the health model with temperature and log rainfall as climate covariates: (a) plug-in method (b) resampling method The posterior means of ψ(Bi) are provided in Figure 16a in the Appendix, which s…
Figure 11
Figure 11. Figure 11: c shows the probability that the dengue risks λ(Bi , t) exceed 1, i.e., P  λ(Bi , t) > 1  for the same months. Note that most of the areas with an estimated probability of exceedence equal to 1 are the same areas badly hit by dengue during the dengue epidemic in the…
Figure 12
Figure 12. Figure 12: Comparison of the posterior standard deviations in the estimated risks ˆ λ(Bi , t) between three approaches: classical approach based on the asymptotic (Gaussian) distribution of the SIR, model-based estimates from the plug-in approach, model-based estimates from the …
Figure 13
Figure 13. Figure 13: Pairwise correlation among the block-level estimates of the climate variables: temperature, relative humidity, and log rainfall Plug-in method Resampling method Parameter Mean SD P2.5% P97.5% Mean SD P2.5% P97.5% σ 2 ν , RW2 time 0.0084 0.0091 0.0001 0.0326 0.0726 0.1…
Figure 14
Figure 14. Figure 14: Plots showing the posterior means and 90% credible intervals of the fixed effects (except γ1, γ2, and γ5) for the dengue model with temperature and log rainfall as covariates. The first vertical line shows the estimates for the plug-in method, while the rest of the li…
Figure 15
Figure 15. Figure 15: Plots showing the posterior means and 90% credible intervals of the fixed effects for the dengue model with relative humidity as climate covariate. The first vertical line shows the estimates for the plug-in method, while the rest of the lines show the estimates for t…
Figure 16
Figure 16. Figure 16: Comparison of (a) posterior mean and (b) posterior standard deviation of the space effects ψ(Bi) between the plug-in method and resampling method, for the model with relative humidity as climate covariate (a) Plug-in method (b) Resampling method [PITH_FULL_IMAGE:figu…
Figure 17
Figure 17. Figure 17: Comparison of classical SIR estimates and model-based SIR estimates from the health model with relative humidity as climate covariate: (a) plug-in method (b) resampling method [PITH_FULL_IMAGE:figures/full_fig_p029_17.png]
Figure 18
Figure 18. Figure 18: Model-based estimates of dengue risks, λˆ(Bi , t), from August 2019 to November 2019, for both plug-in method and resampling method on the dengue model with temperature and log rainfall as climate covariates 29 [PITH_FULL_IMAGE:figures/full_fig_p029_18.png]
Figure 19
Figure 19. Figure 19: Posterior uncertainty of model-based estimates of dengue risks from August 2019 to November 2019, for both plug-in method and resampling on the dengue model with temperature and log rainfall as climate covariates 30 [PITH_FULL_IMAGE:figures/full_fig_p030_19.png]
Figure 20
Figure 20. Figure 20: Probability of exceedence, i.e., P [PITH_FULL_IMAGE:figures/full_fig_p031_20.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 16 canonical work pages

  1. [1]

    Abdullah, N. A. M. H., Dom, N. C., Salleh, S. A., Salim, H., and Precha, N. (2022). The association between dengue case and climate: A systematic review and meta-analysis. One Health, page 100452. Adin, A., Krainski, E., Lenzi, A., Liu, Z., Mart´ ınez-Minaya, J., and Rue, H. (2023). Automatic cross- validation in structured models: Is it time to leave out...

    work page Pith review arXiv 2022

  2. [2]

    The SPDE representation of the observation models in Equation (21) are latent Gaussian models; thus, inference can be performed using INLA

    This implies that the first-stage observation models are specified as: w1t =β01 +Ztβ + B1 ˜ξt +σ2 e1 Inw1 w2t = B2 ˜α0t +α1 β01 +Z1tβ + B1 ˜ξt +σ2 e2 Inw2 , (21) where B1 and B2 are appropriate projection matrices, which map the values of the fields from the mesh to the spatial location of the observed data w1t and w2t, respectively. The SPDE representati...

    work page 2025

  3. [6]

    Comparison of hyperparameter estimates between the plug-in method and the resampling method for the dengue model with relative humidity as the climate covariate 26 A preprint - August 24, 2025 (a) γ0, Intercept (b) γ3, log Rain (c) γ4, Climate Type (d) γ6, covid (e) γ7, log Population Density Figure

    work page 2025

  4. [7]

    Moreover, ω1(s,t ) is a Gaussian innovation process with the Mat´ ern covariance function

    +ω1(s,t ), |ϕ1|< 1, (14) where β0 is an intercept, and z(s,t ) is a set of known covariates, typically topographical features such as elevation, with linear effect β, and ξ(s,t ) is a spatio-temporal random effect, which evolves in time as an autoregressive (AR) process of order 1, where ϕ1 is the AR parameter. Moreover, ω1(s,t ) is a Gaussian innovation ...

    work page 2011

  5. [8]

    +ω2(gj,t ), (18) where α0(gj,t ) is a spatially-structured additive bias which varies in time following an AR1 model, where |ϕ2| < 1 is the AR parameter. The term ω2(gj,t ) is another time-independent Gaussian process with the Mat´ ern covariance function, similar to Equations (15) and (16), and with its own set of parameters, namely, σ2 2 and κ2. The par...

    work page 2025

  6. [10]

    For the data fusion model, we have three additional parameters: σ2,ρ 2 andϕ2, which are the parameters of the Mat´ ern fieldω2(s,t )

    = 0.5. For the data fusion model, we have three additional parameters: σ2,ρ 2 andϕ2, which are the parameters of the Mat´ ern fieldω2(s,t ). The prior choice for these models are discussed in Villejo et al. (2025). 8.3 Predictor expressions The following are the predictor expressions for the climate models: Temperature(s,t ) = β0 +β1 log Elevation(s,t ) +...

    work page 2025

  7. [14]

    28 A preprint - August 24, 2025 (a) Posterior mean (b) Posterior standard deviation Figure

    work page 2025

  8. [15]

    27 A preprint - August 24, 2025 (a) γ0, Intercept (b) γ1, Relative humidity (c) γ2, Climate Type (d) γ3, RH × Climate Type (e) γ4, covid (f) γ5, log Population Density Figure

    work page 2025

Show all 17 references
  1. [18]

    Model-based estimates of dengue risks, ˆλ(Bi,t ), from August 2019 to November 2019, for both plug-in method and resampling method on the dengue model with temperature and log rainfall as climate covariates 29 A preprint - August 24, 2025   γ1x1 γ2x2 γ3x3 γ4x4 γ5x5 γ6x6 ...

    2019

  2. [19]

    Posterior uncertainty of model-based estimates of dengue risks from August 2019 to November 2019, for both plug-in method and resampling on the dengue model with temperature and log rainfall as climate covariates 30 A preprint - August 24, 2025 Figure

    2019

  3. [20]

    Probability of exceedence, i.e., P λ(Bi,t )> 1 from August 2019 to November 2019, for both plug-in method and resampling method on the dengue model with temperature and log rainfall as climate covariates 31

    2019

  4. [24]

    Climate of the philippines

    21 A preprint - August 24, 2025 PAGASA (2023). Climate of the philippines. https://www.pagasa.dost.gov.ph/information/ climate-philippines. Accessed: 2023-06-12. Patz, J. A., McGeehin, M. A., Bernard, S. M., Ebi, K. L., Epstein, P. R., Grambsch, A., Gubler, D. J., Reither, P.,...

    2023

  5. [77]

    R., Bolneo, A

    Edillo, F., Ymbong, R. R., Bolneo, A. A., Hernandez, R. J., Fuentes, B. L., Cortes, G., Cabrera, J., Lazaro, J. E., and Sakuntabhai, A. (2022). Temperature, season, and latitude influence development-related phenotypes of philippine aedes aegypti (linnaeus): Implications for d...

    2022 arXiv

  6. [300]

    The value 300 km is one-third of the maximum distance in the spatial domain, while the value of σ10 is the empirical standard deviation of the data

    = 0 .5, where σ1 and ρ1 are the marginal standard deviation and range parameter of the Mat´ ern fieldω1(s,t ). The value 300 km is one-third of the maximum distance in the spatial domain, while the value of σ10 is the empirical standard deviation of the data. For the AR parame...

    2025

  7. [734]

    Seposo, X., Valenzuela, S., and Apostol, G. L. (2023). Socio-economic factors and its influence on the association between temperature and dengue incidence in 61 provinces of the philippines, 2010–2019. PLOS Neglected Tropical Diseases, 17(10):e0011700. Seposo, X., Valenzuela,...

    2023

  8. [999]

    Su, G. L. S. (2008). Correlation of climatic factors and dengue incidence in metro manila, philippines. AMBIO: A Journal of the Human Environment , 37(4):292–294. Subido, M. E. and Aniversario, I. S. (2022). A correlation study between dengue incidence and climatological facto...

    2008

  9. [2472]

    Marigmen, J

    AIP Publishing. Marigmen, J. L. D. C. and Addawe, R. C. (2022b). Forecasting and on the influence of climatic factors on rising dengue incidence in baguio city, philippines. Journal of Computational Innovation and Analytics (JCIA), 1(01):43–68. McMichael, A. J. (2003). Climate...

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.