Upcoming Lecture: 27 May 2026
Event Information
- Date: 27 May 2026
- Time: 7:00 a.m. EST | 11:00 a.m. UTC | 9:00 9.m. AEDT
- Title: "From Clinical Data to Life-History Trajectories: The Synergy of Multistate Models and Collaborative Science"
Abstract
Traditional survival analysis often oversimplifies the complexity of patient clinical trajectories. In this lecture, we explore how the DIVINE project, a prospective multicenter cohort study of hospitalized adults with COVID-19, served as the foundation for converting fragmented clinical records into coherent life-history trajectories. By using Multistate Models (MSM), we offer a detailed map of disease progression that goes beyond the limitations of classical methods.
We begin by addressing the initial challenges of the pandemic: characterizing early clinical pathways through the development of MSMpred, a user-friendly tool for fitting models and predicting individual patient evolution. We also discuss the estimation of incubation time using novel survey data, which offered critical insights for determining quarantine lengths and refining pandemic simulation models.
A central theme of this talk is the critical re-evaluation of the first-order Markov assumption. We introduce a formal second-order framework designed to capture clinical memory, where a patient’s prognosis is explicitly linked to their preceding State. By extending the Chapman-Kolmogorov equations and introducing two novel non-parametric estimators: the Conditional (CPE) and the Relative (RPE) Probability Estimators, we demonstrate how this past history dependence enhances the prediction of future transition probabilities.
This leads to the second part of the lecture: Beyond the Initial Scope: Emerging Challenges in DIVINE, where we address the theoretical foundations required to compare both approaches. Our analysis of asymptotic efficiency proves that the RPE is a superior estimator for clinical inference than the CPE.
Finally, we discuss extended applications, including Discriminative Analysis for competing risks and the modeling of reinfection trajectories. This ongoing collaborative work with a Basque Research Team highlights the utility of multistate models in analyzing complex, real-world systematic registries. By providing a comprehensive description of disease progression and recovery cycles, this framework serves as a backbone for evidence-based medicine, offering a robust approach to managing chronic conditions and recurring clinical events beyond the immediate scope of infectious outbreaks.
Biography
 |
Guadalupe Gómez-Melis is a mathematician by training and a biostatistician by vocation, dedicated to developing statistical methods with significant scientific and social impact. She earned her PhD at Columbia University under the supervision of Professor John Van Ryzin, specializing in survival analysis, a field that remains a cornerstone of her research. After serving as an Assistant Professor at Ohio State University, she joined the Universitat Politècnica de Catalunya (UPC), where she has been a Full Professor since 2005.
Her international trajectory includes fellowships at Harvard University, collaborating with Stephen W. Lagakos on HIV/AIDS research, as well as research stays at Oxford University’s Clinical Research Unit (Vietnam) and the MD Anderson Cancer Center. Recently, she was elected to the Royal Academy of Sciences and Arts of Barcelona (RACAB).
Professor Gómez-Melis has a long-standing history of leadership within the International Biometric Society (IBS). Beyond her current role on the Executive Board, her service includes contributions to the Awards Committee, Council, Strategic Planning Committee, Advisory Committee and Mentoring Committee. Her leadership extends to the Catalan Council of Statistics, where she served as President. In recognition of her significant impact on the field, she was honored with the Marvin Zelen Memorial Lecture Award and has been named an IBS Distinguished Lecturer.
|
The IBS Distinguished Lecture Series is open to all IBS members worldwide.
Future Sessions
Do you have suggestions for an upcoming session? Let us know! Please email your recommendations to the IBO staff. The IBS Education Committee will carefully review all submissions, contact potential speakers to confirm their availability, and announce the selected presenters on the IBS website. We appreciate your input!