Registration for a live webinar on 'Gamma-delta T cells for immunotherapy of cancer' is now open.See webinar details
Phase II clinical trials - Bayesian methods
Published on July 31, 2018 23 min
A selection of talks on Methods
Artificial intelligence in medicine: history & state of the art
- Prof. John Fox
- University of Oxford, UK
Understanding statistics in epidemics and pandemics: lessons learned from COVID-19
- Prof. Sarah Ransdell
- Nova Southeastern University, USA
International biobanking: overview of key practices and policies
- Dr. Jim Vaught
- Editor in Chief, Biopreservation & Biobanking, USA
Hi, this is Fei Ye, from Vanderbilt University Medical Center. In the second part of my lecture, I will be focusing on using the Bayesian methods to design phase II clinical trials.
As I mentioned at the previous lecture, adaptive designs have emerged rapidly during the last two decades. But mostly with Frequentist approach, often as an effective way to speed up the evaluation process because of the flexibility and the more intuitive way of interpreting the results with Bayesian methods. Many Bayesian designs have also been proposed and conducted for phase II clinical trials. However, it has been reported that they are still poorly used in practice mainly because many clinicians do not fully understand the Bayesian methods. So, many statisticians, as well as clinicians, are putting a lot of effort into promoting all phases of clinical trials to use Bayesian methods. First, I'm going to explain the differences between a Bayesian and a Frequentist approach. Then, I'm going to talk about Bayes' theorem, the impact of prior distribution and then I will give an example of Bayesian trial, and the last I will introduce the very new Bayesian basket design.
From a Frequentist perspective, we draw conclusions from sample data, from data we collect in the current study with the emphasis on the frequency or proportion of the data. For example, we can calculate the response rate, we can calculate the adverse event rate from the current trial. This is the inference framework in which the well-established methodologies of statistical hypothesis testing including p-values, including confidence intervals are based upon. With a Frequentist approach we write the probability P of an uncertain event A, P of A, which is defined by the frequency of that event based on the data we observed during the trial. In other words, based on previous observations. For example, in the United States, 48.8 percent of all babies born are girls; and suppose that we are interested in the event A, which is a randomly selected baby is a girl. According to the Frequentist approach, this P of A equals point four eight eight.