Upcoming GRBIO seminars

Organizers: Ferran Reverter and Jordi Cortés

Click on the seminar to see more info.

Online. If you want to attend, please contact with 

Statistical approaches to correct for baselines in clinical trials

When adjusting a model, it is possible to adopt different strategies regarding the use pf baseline. It can be included in the model as a covariate or as part of the dependent variable.  In this work, two different methods, with different approaches to baseline, constrained longitudinal data analysis (cLDA) and analysis of covariance (ANCOVA) were applied to a real data set from a clinical trial and simulated data. It is aimed to study the behaviour of the methods under different conditions and try to figure out what would be the best approach.

Biosketch

Matilde Francisco is graduated in Biochemistry at Universidade do Porto and did a master's degree in Biostatistics at the Universidade de Lisboa with the master's thesis developed at Universitat Politècnica de Catalunya (UPC). Currently, she is working as a research technician at UPC. 

Online. If you want to attend, please contact with 

Generación de datos multiómicos sintéticos con Generative Adversarial Networks

En este trabajo exploramos la generación de datos multiómicos sintéticos usando GANs, VAEs y un modelo de GANFORMER. Además, se explora la conversión entre datos ómicos usando GANs y Transformers. 

Biosketch

Posición Actual: En búsqueda de posición para realizar tesis doctoral. Educación: Ingenieria Superior Informática, Facultat Informática Barcelona (2007); Máster Bioinformática y Bioestadística, UOC (2024). Experiencia Profesional: Ingeniero de Calidad de Software(2009-2015); Consultor Sistemas SAP (2008); Analista/Programador (2006-2008); Técnico informático (2001-2005).

Hybrid format (Online & Face to face in C5202 room of Campus Nord). If you want to attend online, please contact with

Data Integration in Statistical Inference

In comparative effectiveness research (CER) for rare types of cancer, it is desirable to combine multiple sources of data, e.g., the primary cohort data containing detailed tumor profiles together with aggregate information derived from cancer registry databases. Such integration of data may improve statistical efficiency in CER, but also pose statistical challenges for incomparability between different sources of dat. We develop the adaptive estimation procedures, which used the combined information to determine the degree of information borrowing from the aggregate data of the external resource. We apply the proposed method to evaluate the long-term effect of several commonly used treatments for inflammatory breast cancer by tumor subtypes, while combining the inflammatory breast cancer patient cohort at MD Anderson and external data.

Biosketch

Yu Shen, Ph.D., Professor and Interim Chair, Department of Biostatistics. Dr. Yu Shen is a Professor of Biostatistics at the University of Texas M.D. Anderson (MDA) Cancer Center, where she holds the Conversation with a Living Legend Professorship. She obtained her PhD of biostatistics from University of Washington in 1994 and has been a faculty member of MDA since 1995. Her research spans the areas of novel biostatistics methodology research, health service research in cancer early detection, and personalized cancer treatment. Her primary interest is to develop state-of-the-art statistical methods and models to address important clinical research questions. She has developed statistical methods in the areas of data integration, modeling the natural history of cancer, adaptive clinical trial designs, and modeling survival data subject to biased sampling, with continuous NIH grant support for her statistical method development and health economic research. In her service to the profession, Dr. Shen has held leadership roles in the American Statistical Association and other statistical societies, including Program Chair for the Lifetime Data Analysis Interest Group and Secretary/Treasurer of the Biometrics Section, and ENAR Program Committee. She has also served as an Associate Editor for four major biostatistics and clinical trial journals. Her contributions to the field have earned her recognition, including the Cancer Research and Prevention Foundation award, finalist for the Julie and Ben Roger Award for Excellence, and elected Fellow of the American Statistical Association (2014).

Online. If you want to attend, please contact with 

Quality reporting in scientific articles in the sports field

This seminar primarily addressed the importance of reporting quality in sports-based studies with ordinal and time-to-event outcomes, and those with data-driven discoveries, in particular, based on clustering techniques. Emphasis was placed on enhancing the transparency and rigor of research in the sports domain, with a particular focus on these specific types of approaches. The discussion delved into strategies to improve the clarity and comprehensibility of findings in scientific articles within the sports sciences, ultimately contributing to the overall advancement of research quality in this field.

Biosketch

Martí Casals holds a PhD in Statistics from the University of Barcelona. Currently, he is associate professor of statistics at the Faculty of Medicine of the UVic-UCC and of Sport Analytics at the National Institute of Physical Education of Catalonia (INEFC). Marti’s research lies in the fields of sports analytics and statistical thinking.
Daniel Fernández holds a PhD in Statistics from the Victoria University of Wellington. Since June 2021 he is Serra Hunter Lecturer Professor at the Statistics and Operations Research department (UPC). Daniel’s research lies in the fields of clustering and ordinal data.
Jordi Cortés holds a PhD in Statistics from the UPC. Since September 2022 he is Lecturer Professor at the Statistics and Operations Research department (UPC). Jordi’s research lies in the fields of statistics for clinical trials and survival analysis.

Online. If you want to attend, please contact with 

 

Online. If you want to attend, please contact with 

Journal club: Sample size and predictive performance of machine learning methods with survival data: A simulation study" de Gabriele Infante, Rosalba Miceli, Federico Ambrogi

Prediction models are increasingly developed and used in diagnostic and prognostic studies, where the use of machine learning (ML) methods is becoming more and more popular over traditional regression techniques. For survival outcomes the Cox proportional hazards model is generally used and it has been proven to achieve good prediction performances with few strong covariates. The possibility to improve the model performance by including nonlinearities, covariate interactions and time-varying effects while controlling for overfitting must be carefully considered during the model building phase. On the other hand, ML techniques are able to learn complexities from data at the cost of hyper-parameter tuning and interpretability. One aspect of special interest is the sample size needed for developing a survival prediction model. While there is guidance when using traditional statistical models, the same does not apply when using ML techniques. This work develops a time-to-event simulation framework to evaluate performances of Cox regression compared, among others, to tuned random survival forest, gradient boosting, and neural networks at varying sample sizes. Simulations were based on replications of subjects from publicly available databases, where event times were simulated according to a Cox model with nonlinearities on continuous variables and time-varying effects and on the SEER registry data.

Link al paper: https://onlinelibrary.wiley.com/doi/10.1002/sim.9931 

Biosketch

PhD in Statistics and Operations Research (2014, UPC) developing the thesis on statistical methodologies applied to pharmacoeconomic problems. Degree in Statistical Sciences and Techniques (2005, UPC) and with an MSc in Biostatistics from the University of Hasselt (2016). I am a lecturer at the Universitat Politècnica de Catalunya and a collaborator at the Universitat Oberta de Catalunya. With more than 20 years of experience as a biostatistician developing and applying methods of experimental design and data analysis in clinical trials and observational studies, mostly in the field of HIV/AIDS. I do my research on correlated data analysis; missing data and statistical methods applied to clinical trials. I co-authored a total of 88 published articles, with 2183 citations and a h-index of 27; researcher in 8 competitive research projects and 4 in outreach and dissemination grants. Member of GRBio(consolidated research group). Member of the board of the Societat Catalana d’Estadistica (SoCE) since 2016. Social media office at the International Biometric Society since 2022.

Hybrid format (Online & Face to face in C5202 room of Campus Nord). If you want to attend online, please contact with 

The risk of injuries in NBA and players’ vulnerability

Injuries commonly occur in sports, and their prevention is of utmost importance because of economic implications and their psychological impact on athletes. This seminar deals with the methods that allow to assess the risk NBA players have of getting injured, distinguishing the first injury and the following relapses. The first step involved the construction of a unique data set obtained after a non-trivial harmonization and merging of several data sources. The final dataset contained information about all the injuries occurred from the 2010-2011 season to the 2019-2020 season, the minutes played by each player (retrieved from play-by-play data), the players characteristics (age, role, BMI) and injuries-related features (e.g. number of rest days).

Two separate analyses have been carried out in the survival analysis framework with two different aims:

  • to evaluate the risk of getting injured (all the players have been considered);
  • to examine the risk of injury recurrence (all the players who got injured at least once were included in the dataset).

In particular, the Cox regression model with the frailty has been used. Using the analysed data, the role of the frailty has been further investigated to clarify its meaning. Moreover, we defined a new variable called Weakness to model the momentary vulnerability of a player caused by the occurrence of multiple injuries in close succession. The idea is that injuries that occur after few play minutes increase the feebleness of a player. Results suggest that the player’s role and the Body Mass Index have an important effect on the risk of injury and that the players’ contingent vulnerability, together with the length of rest affect the risk of suffering a new injury in a short period of time.

Biosketch

Ambra Macis, PhD in Analytics for Economics and Management (scientific area: Statistics). Is an Assistant Professor of Statistics at the University of Brescia. Her scientific research deals with the Statistical Sciences from both a methodological and applied point of view. She authored/co-authored several scientific papers in international journals and participated to national/international conferences. She teaches undergraduate and graduate courses in the field of Statistics. She is a member of the BDsports project (bdsports.unibs.it) and is Associate Editor of the Journal of Sport Analytics, responsible to manage the papers dealing with basketball.