Imputation and predictive modeling with biomedical multi-scale data

Placeholder Show Content


In recent years, large amounts of biomedical data are being collected, which allow researchers to discover new knowledge about diseases and to create methods to improve health care. However, when working with the emerging biomedical data, one may encounter problems of incomplete data, small sample data, or multi-scale data, each of which presents a unique challenge for data analysis. In this thesis, we present three pieces of work addressing these challenges. In the first part, we describe a deep learning model for efficient data imputation to handle the problem of incomplete data. When dealing with certain missing-not-at-random cases, an effective and practical shift-correction implementation is developed to improve imputation accuracy. Furthermore, we look into the mechanism that contributes to the new framework's superior performance. In the second part, we describe how we use meta-learning to alleviate the problem of small-sample data in the context of survival prediction. The meta-learning framework shows advantages over competing methods for the case of limited training samples from the target disease. In addition, the model is interpretable and allows us to uncover biological pathways related to the disease outcome. In the final part, we describe different data fusion strategies that integrate data modalities on different scales for improved results in survival prediction. The fusion strategies are evaluated with brain tumor cohorts across different age groups and tumor subtypes. We further examine whether the model learnt from one brain tumor cohort can potentially provide useful information for other cohorts with different disease characteristics.


Type of resource text
Form electronic resource; remote; computer; online resource
Extent 1 online resource.
Place California
Place [Stanford, California]
Publisher [Stanford University]
Copyright date 2020; ©2020
Publication date 2020; 2020
Issuance monographic
Language English


Author Qiu, Yeping
Degree supervisor Gevaert, Olivier Michel Simonne
Thesis advisor Gevaert, Olivier Michel Simonne
Thesis advisor Howe, Roger Thomas
Thesis advisor Wootters, Mary
Degree committee member Howe, Roger Thomas
Degree committee member Wootters, Mary
Associated with Stanford University, Department of Electrical Engineering


Genre Theses
Genre Text

Bibliographic information

Statement of responsibility Yeping Lina Qiu.
Note Submitted to the Department of Electrical Engineering.
Thesis Thesis Ph.D. Stanford University 2020.
Location electronic resource

Access conditions

© 2020 by Yeping Qiu
This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).

Also listed in

Loading usage metrics...