Personalized treatment-response trajectories: Errors-in-variables, interpretability, and causality

Thumbnail Image

Files

master_Zhang_Guangyi_2019.pdf (640.56 KB)

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Master's thesis
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2019-06-17

Department

Major/Subject

Machine Learning, Data Science and Artificial Intelligence

Mcode

SCI3044

Degree programme

Master’s Programme in Computer, Communication and Information Sciences

Language

en

Pages

40+0

Series

Abstract

One fundamental problem in many applications is to estimate treatment-response trajectories given multidimensional treatment variables. However, in reality, the estimation suffers severely from measurement error both in treatment timing and covariates, for example when the treatment data are self-reported by users. We introduce a novel data-driven method to tackle this challenging problem, which models personalized treatment-response trajectories as a sum of a parametric response function, based on restored true treatment timing and covariates and sharing information across individuals under a hierarchical structure, and a counterfactual trend fitted by a sparse Gaussian Process. In a real-life dataset where the impact of diet on continuous blood glucose is estimated, our model achieves a superior performance in estimation accuracy and prediction.

Description

Supervisor

Marttinen, Pekka

Thesis advisor

Marttinen, Pekka

Keywords

bayesian methods, errors-in-variables, causality, hierarchical modeling

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-201906233984

Collections

[dipl] Perustieteiden korkeakoulu / SCI