Abstract: Disease registries are organized systems that use observational study methods to collect data for a particular disease or condition.
Measures from clinical assessments are collected in a disease registry for tracking the progression of the target disease. Due to the nature of observational study, measures collected in disease registry can be affected by factors not associated with disease progression. In addition, for clinical assessments, threshold values based on experiences of domain experts are used to determine the onset of new symptoms. Data-driven methods are needed to objectively determine the threshold values, or to support existing thresholds.
A patient registry may include one or more comparison groups of control participants. In this study, we propose a data-driven method for adjusting the effects of non-disease-related factors and generating robust clinical measures and symptom onset indicators based on raw measures from a disease registry that comes with a control cohort. We demonstrated the proposed method on observational Huntington's Disease datasets and discussed the properties of the robust measures in details.
Learning Objective 1: A data driven procedure for adjusting the effects of non-disease-related factors, and for generating robust clinical measures based on existing clinical measures collected in disease registries with case-control designs.
Zhaonan Sun (Presenter)
IBM T. J. Watson Research Center
Ying Li, IBM T. J. Watson Research Center
Soumya Ghosh, IBM T. J. Watson Research Center
Yu Cheng, IBM T. J. Watson Research Center
Amrita Mohan, CHDI Management/CHDI Foundation
Cristina Sampaio, CHDI Management/CHDI Foundation
Jianying Hu, IBM T. J. Watson Research Center