Activity Number:
|
343
- Longitudinal Analysis, Clinical Trial Design, and Other Topics in Biopharmaceutical Statistics
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 5, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Biopharmaceutical Section
|
Abstract #312228
|
|
Title:
|
Predicted Response Rate Differences of Antidepressants at End Point Among Patients Not Meeting Study Early Response
|
Author(s):
|
Marc Sobel* and Ibrahim Turkoz and Xiwu Lin and Ella Daly
|
Companies:
|
temple university and Janssen Research and Development, LLC and Janssen Research and Development, LLC and Janssen Scientific Affairs LLC
|
Keywords:
|
clinical trials;
logistic regression;
machine learning;
dynamic models;
model comparison
|
Abstract:
|
Esketamine nasal spray (ESK) plus a newly initiated antidepressant reduces depression symptoms in patients with treatment-resistant depression. This analysis informs clinicians regarding how long to continue ESK in patients without an early response. Response rates at day 28 were determined among those without prior responses. A two-part goal is considered: 1. Employing a variety of statistical and machine learning-based models, we estimate the conditional probabilities of a response for each treatment group at day 28 given no prior early response. Our estimates are compared with the corresponding empirical conditional probabilities to determine their viability for predicting future responses. 2. Clinical practice suggests that multiple factors influence response, therefore the importance of covariates are evaluated across models. Conditional probability computations are carried out using the covariates deemed to be important. We compare logistic regression models, dynamic logistic models, Multiple Adaptive Regression Spline models, and Neural Autologistic Distribution Estimators. The importance of covariates are compared using error rates based on Breiman’s formulation.
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2020 program
|