Conferências
Conferência 1 (C1) – Conferência de Abertura – 23/10 (09:00 – 10:00)
Complex survival analysis with INLA
INLA is an approximate Bayesian inference method that can fit many different statistical models including survival models. In this talk I will present two projects in which INLA was used for survival analysis. The first shows how to fit a spatial survival model based on areal data or point-indexed data. The second project employs a quantile joint model to study adherence in the midst of outliers.
Conferência 2 (C2) – 23/10 (10:30 – 11:30)
Bridging the Gap: Reliability Innovations in Brazil Connecting Academia and Industry
Palestrante: Francisco Louzada – Universidade de São Paulo (ICMC-USP)
In today’s world, our reliance on mechanical and electronic devices is growing exponentially. From intelligent sensors and artificial intelligence to agricultural, financial, and medical robots, these technologies have become integral to our daily lives. However, despite their efficiency, these devices are not immune to failure. As a result, statistical reliability analysis has emerged as a crucial tool in the innovation process, ensuring the dependability and longevity of these technologies. This talk presents several reliability innovation projects that showcase the efforts being made in Brazil to bridge the gap between academia and various industrial sectors, including oil and gas, agriculture, medicine, and finance. By establishing a connection between theoretical research and practical applications, these projects aim to enhance the reliability and performance of critical equipment and systems. The focus of this talk is on the development of reliability models for a range of applications. These include oil well construction equipment, which plays a vital role in the energy sector; bucket tracking equipment, essential for efficient material handling; and agricultural machinery, crucial for maintaining food production. Additionally, the talk explores communication modeling for mobile phones, a technology that has revolutionized personal and business communication. Through these projects, we demonstrate how statistical reliability analysis can be leveraged to improve the design, manufacturing, and maintenance of complex systems. By collaborating with industry partners, academia can contribute to the development of more reliable and resilient technologies, ultimately benefiting society as a whole. The talk concludes by emphasizing the importance of continued collaboration between academia and industry in the field of reliability engineering. By working together, we can drive innovation, improve safety, and ensure the long-term performance of the devices and systems upon which we increasingly depend.
Conferência 3 (C3) – 23/10 (14:00 – 15:00)
Log-symmetric models with cure fraction with application to leprosy reactions data
Palestrante: Francisco Medeiros – Universidade Federal do Rio Grande do Norte (UFRN)
In this paper, we propose a log-symmetric survival model with cure fraction, considering that the distributions of lifetimes for susceptible individuals belong to the log-symmetric class of distributions. This class has continuous, strictly positive, and asymmetric distributions, including the log-normal, log-$t$-Student, Birnbaum-Saunders, log-logistic I, log-logistic II, log-normal-contaminated, log-exponential-power, and log-slash distributions. The log-symmetric class is quite flexible and allows for including bimodal distributions and outliers. This includes explanatory variables through the parameter associated with the cure fraction. We evaluate the performance of the proposed model through extensive simulation studies and consider a real data application to evaluate the effect of factors on the immunity to leprosy reactions in patients with Hansen’s disease.
Conferência 4 (C4) – 24/10 (10:30 – 11:30)
Survival Data Simulation With the R Package rsurv
Palestrante: Fábio Nogueira Demarqui – Universidade Federal de Minas Gerais (UFMG)
In this talk we introduce the R package rsurv, aimed for general survival data simulation purposes. The package, which is available on CRAN at https://CRAN.R-project. org/package=rsurv, is built under a new approach to simulate survival data that depends deeply on the use of dplyr verbs. The proposed package allows the simulation of survival data from a wide range of regression models, including accelerated failure time (AFT), proportional hazards (PH), proportional odds (PO), accelerated hazard (AH), Yang and Prentice (YP), and extended hazard (EH) models. The package rsurv also stands out by its ability to generate survival data from an unlimited number of baseline distributions provided that an implementation of the quantile function of the chosen baseline distribution is available in R. Another nice feature of the package rsurv lies in the fact that linear predictors are specified via a formula-based approach, facilitating the inclusion of categorical variables and interaction terms. The functions implemented in the package rsurv can also be employed to simulate survival data with more complex structures, such as survival data with different types of censoring mechanisms, survival data with cure fraction, survival data with random effects (frailties), multivariate survival data, and competing risks survival data.
Conferência 5 (C5) – 24/10 (15:00 – 16:00)
Bayesian solution to the monotone likelihood in the standard mixture cure model
Palestrante: Vinícius D. Mayrink – Universidade Federal de Minas Gerais (UFMG)
An advantage of the standard mixture cure model over a usual survival model is how it accounts for population heterogeneity. It allows a joint estimation for the distribution related to the susceptible and non-susceptible subjects. The estimation algorithm may provide coefficients when the likelihood cannot be maximized. This phenomenon is known as Monotone Likelihood (ML), common in survival and logistic regressions. The ML appears in situations with a small sample size, many censored times, and many binary or unbalanced covariates. Particularly, it occurs when all uncensored cases correspond to one level of a binary covariate. The existing frequentist solution is an adaptation of the Firth correction, originally proposed to reduce the bias of maximum likelihood estimates. It prevents estimates by penalizing the likelihood, with the penalty interpreted as the Bayesian Jeffreys prior. This study considers the penalized likelihood of the standard mixture cure model with different penalties (Bayesian priors). A Monte Carlo simulation study indicates good inference results, especially for balanced data sets. Finally, a real application involving melanoma data illustrates the approach.
This is a joint work with Frederico M. Almeida and Enrico A. Colosimo.
Conferência 6 (C6) - Conferência de Encerramento – 25/10 (11:30 – 12:30)
Semiparametric joint modeling of competing risks survival and longitudinal data*
Palestrante: Gisela Tunes – Universidade de São Paulo (IME-USP)
In many observational studies, patients are followed until death or another outcome of interest is observed and several clinical variables are also measured during follow up. When longitudinal data is available as well as survival, it may be of interest to understand the effect of prognostic variables in both longitudinal and survival outcomes. In this work, a joint model for competing risks survival data and longitudinal data is discussed. A semiparametric approach is presented, along with simulation results and real data application.
* Joint work with Renato Santos da Silva
