Modern Statistical Models and Artificial Intelligence for Early Detection of Disease with Longitudinal Biomarker
Principal Investigator
Name
Tim Tobias Arndt
Degrees
M.Sc.
Institution
Universität Augsburg
Position Title
Doctorate Student
Email
About this CDAS Project
Study
PLCO
(Learn more about this study)
Project ID
PLCO-857
Initial CDAS Request Approval
Nov 15, 2021
Title
Modern Statistical Models and Artificial Intelligence for Early Detection of Disease with Longitudinal Biomarker
Summary
An early disease detection increases the probability of a healing, facilitates the treatment and reduces the impact on the patient. Hence, models need to be developed which are capable to predict the risk of a patient. In previous work, shared random effects models as well as pattern mixture model were introduced [1]. In our study, we would like to extend these existing methods by incorporating more complex statistical models and AI-based methods to provide a more flexible approach with a better approximation. The PLCO data for ovarian cancer with the CA-125 biomarker should be used for practical application and a comparison with other methods.
[1] Han, Y, Albert, PS, Berg, CD, Wentzensen, N, Katki, HA, Liu, D. Statistical approaches using longitudinal biomarkers for disease early detection: A comparison of methodologies. Statistics in Medicine. 2020; 39: 4405– 4420. https://doi.org/10.1002/sim.8731
[1] Han, Y, Albert, PS, Berg, CD, Wentzensen, N, Katki, HA, Liu, D. Statistical approaches using longitudinal biomarkers for disease early detection: A comparison of methodologies. Statistics in Medicine. 2020; 39: 4405– 4420. https://doi.org/10.1002/sim.8731
Aims
- Extension of existing models with more complex statistical models and AI-based methods
- Application on PLCO data for ovarian cancer
- Comparison with other models
Collaborators
Stefan Schiele, M.Sc., Augsburg University
Elena Ney, Augsburg University
Prof. Dr. Gernot Müller, Augsburg University