**6. Conclusion**

Environmental health research is becoming a cornerstone for supporting evidence-based (informed) decision making in healthcare services and management. Providing evidence through robust and relevant epidemiologic studies in environmental health research may be improved through an adequate utilization of statistics methods. In this chapter, we reviewed the Generalized Additive Models and the most used estimating methods and presented their advantages and limits. Knowing this, researchers should take into account these aspects when it is time to define exposures and outcomes, to map spatial variations, to design epidemiologic studies' conceptual frameworks and to select suitable estimating models. These critical aspects are of central importance for developing clinical and public health decision making to reduce the burden of environment impacts on individual and population health. Moreover, using accurate and relevant methods, i.e. GAM, in environmental epidemiology studies is a cornerstone for developing effective actions that may help save cost and improve decision making performance.

Improvements will be seen also in clinical practices through a better understanding and the integration in medical decisional algorithms of the effects of long term exposition to specific environmental factors. These effects are translated into risks of occurrence and prognosis of sensitive diseases (i.e. breast cancer, lung cancers). Spreading GAM method utilization in environmental epidemiology through a clinical perspective is highly recommended to develop effective decisional tools that may greatly improve personalized medicine. Moreover, GAM method may help to better manage follow-up of patients exposed to long term medications and reduce side-effects and complications. This review highlights the utility of Generalized Additive Model (GAM) for risk assessment (such as breast cancer) related to environmental factors and explored the use of the GAM for risk assessment in the presence of multiple non-linear effects. The selection and the estimation of the parameters and non-linear functions (B-Splines and P-splines) are essential for an adequate estimate of the risk. Next research should explore how GAM models may help the development of relevant risk assessment tools that may be integrated in personalized medical decision making algorithms. The GAM will allow the integration of environmental factors and others health determinants in clinical algorithms that may help improve the personalization of healthcare delivery. These algorithms will be implemented in public health programs (i.e. personalization of breast cancer screening based on women individual risk) and clinical algorithms (i.e. for patients with a diagnosis of breast cancer the personalization of followup will be based on the surveillance of relevant factors such as the biomarkers, the clinical signs and the exposition to environmental factors).

This chapter presented the potential of the Generalized Additive Model (GAM) for environmental studies. Generalized additive models (GAMs) are a generalization of generalized linear models (GLMs) and constitute a powerful technique to capture nonlinear relationships between explanatory variables and a response variable. Selection of the best parameter estimation methods, control for confounding variables and concurvity aims to reduce bias and improve the use of the GAM model. Moreover, when using the GAM model in environmental health, and for an adequate interpretation of the outputs, socio-economic and demographic parameters should be considered.

### **7. Acknowledgement**

The authors would like to thank the CIHR Team on Familial breast cancer at Université Laval (QC) leaded by Dr Jacques Simard; and also the Consortium national de formation en santé-Université de Moncton (NB) for the financial support they provided to prepare and publish this chapter.

### **8. References**

94 Novel Approaches and Their Applications in Risk Assessment

effect of a variable depends on the value observed for the other one. A form of interaction often found in bibliography is the modification of the effect. The modification of the effect happens when the statistical measure of the association between the explicative variable *X*<sup>1</sup> and the response variable *Y* depends on the level of another variable *X*2, known as the effect modifier. The extent of the relationship depending on the value of the effect modifier contributes to the improvement of the model fit. In the field of environmental health, this allows us to identify the most vulnerable groups to a particular condition (Wood, 2006;

Environmental health research is becoming a cornerstone for supporting evidence-based (informed) decision making in healthcare services and management. Providing evidence through robust and relevant epidemiologic studies in environmental health research may be improved through an adequate utilization of statistics methods. In this chapter, we reviewed the Generalized Additive Models and the most used estimating methods and presented their advantages and limits. Knowing this, researchers should take into account these aspects when it is time to define exposures and outcomes, to map spatial variations, to design epidemiologic studies' conceptual frameworks and to select suitable estimating models. These critical aspects are of central importance for developing clinical and public health decision making to reduce the burden of environment impacts on individual and population health. Moreover, using accurate and relevant methods, i.e. GAM, in environmental epidemiology studies is a cornerstone for developing effective actions that

Improvements will be seen also in clinical practices through a better understanding and the integration in medical decisional algorithms of the effects of long term exposition to specific environmental factors. These effects are translated into risks of occurrence and prognosis of sensitive diseases (i.e. breast cancer, lung cancers). Spreading GAM method utilization in environmental epidemiology through a clinical perspective is highly recommended to develop effective decisional tools that may greatly improve personalized medicine. Moreover, GAM method may help to better manage follow-up of patients exposed to long term medications and reduce side-effects and complications. This review highlights the utility of Generalized Additive Model (GAM) for risk assessment (such as breast cancer) related to environmental factors and explored the use of the GAM for risk assessment in the presence of multiple non-linear effects. The selection and the estimation of the parameters and non-linear functions (B-Splines and P-splines) are essential for an adequate estimate of the risk. Next research should explore how GAM models may help the development of relevant risk assessment tools that may be integrated in personalized medical decision making algorithms. The GAM will allow the integration of environmental factors and others health determinants in clinical algorithms that may help improve the personalization of healthcare delivery. These algorithms will be implemented in public health programs (i.e. personalization of breast cancer screening based on women individual risk) and clinical algorithms (i.e. for patients with a diagnosis of breast cancer the personalization of followup will be based on the surveillance of relevant factors such as the biomarkers, the clinical

may help save cost and improve decision making performance.

signs and the exposition to environmental factors).

Bates and Maechler, 2009).

**6. Conclusion** 


Generalized Additive Models in Environmental Health: A Literature Review 97

Lall R., Ito K. and G.D. Thurston (2011). Distributed Lag Analyses of Daily Hospital

Lang, S. and A. Brezger (2004): Bayesian P-Splines. Journal of Computational and Graphical

Liang K.Y. and S. L. Zeger (1986). Longitudinal data analysis using general linear models.

Mackenbach, J., A. Kunst, and C. Looman (1992). Seasonal variation in mortality in The Netherlands. Journal of Epidemiology and Community Health 46, 261–265. Marx B.D. and P. H. C. Eilers (1998). Direct generalized additive modeling with penalized

Michelozzi P et al. (2007) Assessment and prevention of acute health effects of weather

Peng R.D., F. Dominici and A. L. Thomas (2006). Model choice in time series studies of air

Peng R.D., Chang H.H., Bell M.L., McDermott A., Zeger S.L., Samet J.M. and F. Dominici

Ramsay T.O., R.T. Burnett and D. Krewski (2003). Exploring bias in generalized additive

Ruppert, D. (2002). Selecting the number of knots for penalized splines. *Journal of* 

Schwartz J., J. M. Samet and J. A. Patz (2004). Hospital admissions for heart disease: the effects of temperature and humidity. *Epidemiology*, 15(6), pp. 755-761. Terzi Y. and M. A. Cengiz (2009). Using of generalized additive model for model selection in

Vieira V., Webster Thomas., Weinberg J., Aschengrau A. and Ozonoff D. (2005) Spatial

Wood, S. N. (2006). Generalized Additive Models, An introduction with R. Chapman &

Wong CM, Thach TQ, Chau PY, Chan EK, Chung RY, Ou CQ, Yang L, Peiris JS, Thomas

*Computational and Graphical Statistics,*11(4), pp 735-757.

conditions in Europe, the PHEWE project: background, objectives, design. Environ

(2008) Coarse Particulate Matter Air Pollution and Hospital Admissions for Cardiovascular and Respiratory Diseases Among Medicare Patients

models for spatial air pollution data. *Environmental Health Perspectives*, 111(10),

multiple poisson regression for air pollution data. Scientific Research and Essay

analysis of lung, colorectal, and breast cancer on Cape Cod: An application of generalized additive models to case-control data. *Environmental Health: A Global* 

GN, Lam TH, Wong TW, Hedley AJ; HEI Health Review Committee. Part 4. Interaction between air pollution and respiratory viruses: time-series study of daily mortality and hospital admissions in Hong Kong. *Resp Rep Health Eff Inst.* 2010

likelihood. *Computational Statistics & Data Analysis*, 28, pp.193-209. McCullagh P. and J. A. Nedler (1989). Genralised linear models. *Chapman&Hall.* 

pollution and mortality. *J. R. Statist. Soc. A*, 169, *Part 2*, pp. 179-203.

*Perspect.* April; 119(4): 455–460.

Statistics, 13, 183-212.

*Biometrika*, 73, 13-22.

Health; 6:12.

1283-1288.

*JAMA*;299(18):2172-2179.

Vol.4 (9), pp. 867-871.

*Access Science Source.* 4:11

Hall/ CRC.

Nov;(154):283-362.

Admissions and Source-Apportioned Fine Particle Air Pollution. *Environ Health* 


*http://etd.library.pitt.edu/ETD/available/etd-12022004-103805/unrestricted/ShuiHe.pdf* 

Katsouyanni K, Samet JM, Anderson HR, Atkinson R, Le Tertre A, Medina S, Samoli E, Touloumi G, Burnett RT, Krewski D, Ramsay T, Dominici F, Peng RD, Schwartz J, Zanobetti A; HEI Health Review Committee. Air pollution and health: a European and North American approach (APHENA). *Res Rep Health Eff Inst*. 2009 Oct;(142):5-90.

Budtz-Jorgensen E., N. Keiding, P. Grandjean and P. Weiher (2007). Confounder selection in

Dominici F., A. McDermott and T.J. Hastie (2003). Issues in Semi-parametric Regression

Doyon B., D. Bélanger and P. Gosselin (2006). Effets du climat sur la mortalité au Québec

Eilers P. H. C. and B. D. Marx (1996). Flexible Smoothing with B-splines and Penalties.

El Adlouni, S., Favre, A.C. and Bobée, B. (2006). Comparison of methodologies to assess the

Fahrmeir L. and S. Lang (2001). Bayesian inference for generalized additive mixed models based on Markov random field priors. *Appl. Statist.*, 50, Part 2, pp. 201-220. Figueiras A., J. Roca-Pardiñas and C. Cadarso-Suárez (2005). A bootstrap method to avoid

pollution. Journal of Epidemiology and Community Health. 59, pp. 881-884. Gehrmanu U., B. Hellriegel, A. Neiss and L. Fahrmeir (2003). Analysis of the time to

Hastie T., L. Sleeper and R. Tibshirani (1992). Flexible covariate effects in the proportional hazard model. Breast Cancer Research and *Treatment*, 22, pp. 241-250. Hastie T. and R. Tibshirani (1990). Generalized Additive Models. *Chapman & Hall*, New-

Hastie T. and R. Tibshirani (1993). Varying-coefficient models. *Journal of the Royal Statistical* 

Harnack, L., G. Block, et al. (1997). "Association of Cancer Prevention-Related Nutrition

Hatzakis, A., Katsouyanni, K., Kalandidi, A., Day, N. and Trichopoulos, D. (1986) Shortterm effects of air pollution on mortality in Athens. Int. J. Epidemiol., 15, 73-81. He S.(2004). Generalized additive models for data with concurvity: statistical issues and a

Knowledge, Beliefs, and Attitudes to Cancer Prevention Dietary Behavior." *Journal* 

*http://etd.library.pitt.edu/ETD/available/etd-12022004-103805/unrestricted/ShuiHe.pdf*  Katsouyanni K, Samet JM, Anderson HR, Atkinson R, Le Tertre A, Medina S, Samoli E,

Touloumi G, Burnett RT, Krewski D, Ramsay T, Dominici F, Peng RD, Schwartz J, Zanobetti A; HEI Health Review Committee. Air pollution and health: a European and North American approach (APHENA). *Res Rep Health Eff Inst*. 2009

exposure. Ann. Epidemiol. 17(1), pp 27-35.

Statistical Science, Vol. 11, No. 2, pp. 89-121.

models. Discussion paper 354-SFB 386- LMU Munich.

*of the American Dietetic Association* 97(9): 957-965.

novel fitting approach. Rapport de recherche.

*of Epidemiology*. 156(3), pp 193-203.

*Rapport de recherche*, INSPQ.

Data Analysis 50(10): 2685-2701.

York.

*Society B*, 55, pp. 757-796.

Oct;(142):5-90.

environmental epidemiology: Assessment of health effects of prenatal mercury

with Applications in Time Series Models for Air pollution and Mortality. Rapport de recherche. http://biosun01.biostat.jhsph.edu/~fdominic/trevorpaper.pdf Dominici F., McDermott A., Zeger S. L. and J.M. Samet (2002). On the use of generalized

additive models in time-series studies of air pollution and health. *American Journal* 

méridional de 1981 à 1999 et simulations pour des scénarios climatiques futurs.

convergence of Markov Chain Monte Carlo methods. Computational Statistics and

the effect of concurvity in generalized additive models in time series of air

sustained progression in Multiple Sclerosis using generalized linear and additive


**6** 

*US Forest Service* 

*USA* 

**The Science and Opportunity** 

**of Wildfire Risk Assessment** 

Mark A. Finney, Dave E. Calkin and Nicole M. Vaillant

Wildfire management within the United States continues to increase in complexity, as the converging drivers of (1) increased development into fire-prone areas, (2) accumulated fuels from historic management practices, and (3) climate change potentially magnify threats to social and ecological values (Bruins et al., 2010; Gude et al., 2008; Littell et al., 2009). The need for wildfire risk assessment tools continues to grow, as land management agencies attempt to map wildfire risk and develop strategies for mitigation. Developing and employing wildfire risk assessment models can aid management decision-making, and can facilitate prioritization of investments in mitigating losses and restoring fire on fire prone landscapes. Further, assessment models can be used for monitoring trends in wildfire risk

The term risk is generally used to measure the chance of loss, as determined from estimates of likelihood and magnitude of particular outcomes. Probabilistic approaches to risk assessment estimate the expected value of the conditional probability of the even occurring and the consequence of the event given that it has occurred. Risk assessments are conducted when predicted outcomes are uncertain, but possible outcomes can be described and their likelihoods can be estimated. Wildfire risk assessment entails projecting wildfire extent and

We begin by introducing a conceptual model of wildfire management (Figure 1) that considers the major drivers of wildfire risk and strategic options for mitigation. Ignition processes influence the spatiotemporal pattern of wildfire occurrence (natural and humancaused), and strategic prevention efforts can reduce the number of wildfires and associated damage (Prestemon et al., 2010). Given an ignition that escapes suppression, fuel, weather, and topography jointly drive wildfire behavior. Of these, only fuel conditions (loading, structure, continuity) can be altered to induce desirable changes in fire behavior (Agee & Skinner 2005). Suppression efforts are intended to slow the growth of active wildfires and reduce the chance of loss. Collectively these factors influence wildfire extent and intensity, which in turn determine the consequences (detrimental and beneficial) to social and ecological values. Wildfire losses can also be prevented or reduced by activities that lessen

intensity, and the consequences of fires interacting with values-at-risk.

**1. Introduction** 

over space and across time.

Matthew P. Thompson, Alan A. Ager,

