Using EEG Signals in Determining Respondents on Placebo

• Grant number: 24/16744-3
• Support Opportunities: Scholarships in Brazil - Master
• Start date: December 01, 2024
• End date: November 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Nancy Lopes Garcia
• Grantee: Nicoli Prosperi Pereira
• Host Institution: Instituto de Matemática, Estatística e Computação Científica (IMECC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated research grant: 23/13453-5 - Stochastic systems modeling, AP.TEM

Abstract

The data to be analyzed comes from a clinical trial to control/treat the effect of sertraline. The data set consists of 96 patients subject to severe depression, randomized between sertraline and placebo. For each individual, several scalar covariates are measured, as well as resting EEG signals on a total of 14 electrodes. The response for each individual is the difference in the Hamilton scale (HAM-D) before the start of the study (baseline) and one week after starting treatment or placebo. The objective is to identify early responders to treatment (indicative of a placebo effect as it is believed that the response to the medication only takes effect after two weeks). In the literature, a hierarchical probit model was proposed with binary unobserved variables indicating the subgroup (early responders or not) depending on the EEG signals and other covariates of interest by studying the EEG data as a matrix covariates (14x45) or using a linear model. generalized incorporating the EEG curves into the mixing probability through a weight function for each EEG which belongs to a finite-dimensional space generated by spline functions. The objective of this work is to use wavelet expansion.