Re: AUEB Stats Seminars 19/11/2021: Estimation of optimal individualized treatment rules for multistate disease processes by Giorgos Bakoyannis (Indiana University School of Medicine)
Thu 18 Nov 2021 - 23:08
The video of the talk is now available at the Youtube channel of AUEB Stats.
AUEB Stats Seminars 19/11/2021: Estimation of optimal individualized treatment rules for multistate disease processes by Giorgos Bakoyannis (Indiana University School of Medicine)
Wed 20 Oct 2021 - 18:17
Presenter: Giorgos Bakoyannis
Assistant Professor, School of Public Health, Indiana University School of Medicine, Indianapolis, Indiana.
Date: Friday 19/11/2021, 16:00
Title: Estimation of optimal individualized treatment rules for multistate disease processes
Abstract: Clinical trials and observational studies evaluating treatments frequently involve multistate disease processes. An example is cancer trials where patient event history involves states such as tumor response, disease progression, and death. The standard practice in such studies is to only evaluate the overall efficacy or effectiveness of a new treatment. However, there is typically high patient heterogeneity associated with most chronic diseases and, thus, a drug that works for one patient may not be effective for another. The modern precision medicine paradigm acknowledges this heterogeneity and aims to develop and deliver therapeutics that are tailored to the individual patient. This effort is expected to lead to better health outcomes for the patients, thereby improving public health. However, there is an important gap in the statistical literature focused on precision medicine. Specifically, there are no methodologies that can be used to estimate optimal individualized treatment rules in the context of multistate disease processes. In this work, we address this significant gap by proposing a methodology that utilizes support vector machines. The methodology provides treatment rules that can be substantially more effective compared to traditional one-size-fits-all treatment rules, and does not impose restrictive, and often unrealistic, parametric assumptions. The theoretical properties of the proposed estimator, including Fisher consistency and asymptotic normality of the estimated value function, are rigorously established using empirical process theory. Simulation studies show that the proposed methodology works well in realistic settings. The methodology is illustrated using real data from a randomized controlled trial on the treatment of metastatic squamous-cell carcinoma of the head and neck. This analysis shows that optimal individualized treatment rules can lead to significantly better health outcomes compared to one-size-fits-all rules.
- AUEB Stats Seminars 26/11/2021: A new framework of semi-Markov processes for parameter estimation and Reliability Analysis by Andreas Makrides (University of the Aegean)
- AUEB Stats Seminars 24/3/2021: Improved estimation of partially-specified models by I. Kosmidis (Univ. of Warwick)
- AUEB Stats Seminars 4/3/2021: Practical Distributionally Robust Markov Decision Processes using Relative Entropy by William Greenall (UCL)
- AUEB Stats Seminars 24/3/2021: Improved estimation of partially-specified models by I. Kosmidis (Univ. of Warwick) από grstats
- AUEB Stats Seminars 10/6/2021: Monte-Carlo Statistical Methods for Parameter Estimation of the GreenLab Plant Growth Model by S. Trevezas
Permissions in this forum:You cannot reply to topics in this forum