Prediction-driven decision rules, randomized control trial design, and survival analysis

in doctoral studies, Adam Brand in Department of Medical Epidemiology and Biostatistics Develop prediction-driven decision rules, and identify randomized controlled trials (RCTs) that can be used to evaluate prediction-driven decision rules, and perform survival analyses. His studies include five projects that might seem disjointed at first glance, but when you take a closer look you’ll see that they followed a natural thought process where you kept asking the question “What happens next?”

In his first project, Adam developed new prediction-based decision rules for treating HIV patients. In the second project, an appropriate randomized controlled design was selected to test the effectiveness of decision rules driven by predictors of disease. In Projects 3-5, examine how to analyze the predicted survival data from a prediction-driven RCT design. Adam developed new applications of the R statistical software for RMST-based statistical inference methods and compared these methods with existing standard methods for testing differences in survival between two groups in sequential real-world RCT cohorts.

Can you tell us something about the most important findings of your dissertation?

– During the course of the work, we discovered some shortcomings in the current research, and we thought it was important to fill in where knowledge was missing and correct errors. I believe that we have developed promising prediction-based decision rules for HIV treatment, and that we have developed a deep understanding of how to accurately test these rules, and hopefully through this we can contribute to statistical knowledge.

How do you think research should continue in the future in this area?

I would like to see a validated randomized trial conducted in the resource-limited areas targeted by these pooled testing methods. If two trials show that the two methods provide increased efficacy without risking patient safety, the methods should be implemented and hopefully enable more HIV-infected patients to receive effective and regular testing for HIV treatment.

thesis

“Prediction-driven decision rules, RCT design, and survival analysis.“

Adam Brand. Karolinska Institutet (2023), ISBN: 978-91-8016-941-7