Files and links (1)
Thesis_David-Erick v51.07 MB
Embargoed Access, Embargo ends: 01/31/2027
Dissertation
Assessment of dynamic treatment regimens effects on longitudinal outcomes through clustering and causal inference
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Autumn 2024
DOI: 10.25820/etd.007580
Abstract
This dissertation presents methodologies for summarizing treatment dose trajectories, clustering longitudinal data, and evaluating the impact of dynamic treatment regimens on clinical outcomes over time. First, we propose a method to compare disease progression across treatment dose trajectories by clustering participants with similar multi-treatment dose patterns. In this approach we use a marginal mean regression model to assess clinical outcomes over time, accounting for potential biases through inverse probability weighting within a quadratic inference function framework. Our application focuses on clinical outcomes for Parkinson’s disease patients from the Parkinson's Progression Markers Initiative (PPMI), where we track disorder progression across treatment clusters.
In a complementary approach, we introduce Recursive clustering by heterogeneity of repeated measures (R-CHRM), a hierarchical clustering technique for longitudinal data, enabling identification and tracking of evolving clusters as new data points arise. This method provides a framework for detecting heterogeneous group trajectories early, facilitating timely interventions or dose adjustments in biomedical studies. Applied to the PPMI data, R-CHRM demonstrates robust handling of complex multivariate longitudinal datasets, supporting efficient cluster tracking and revealing heterogeneity in patient responses.
Finally, we explore dynamic treatment regimens through a causal inference framework, addressing the complexities of fluctuating dosages, multiple medication types, and confounding factors in observational settings. By using R-CHRM and inverse probability-weighted regression, we assess the effects of treatment regimen clusters on multiple longitudinal outcomes in Parkinson’s disease patients. Collectively, these methodologies offer comprehensive tools for advancing personalized medicine and improving clinical outcome evaluations in complex, time-varying treatment settings.
Details
- Title: Subtitle
- Assessment of dynamic treatment regimens effects on longitudinal outcomes through clustering and causal inference
- Creators
- David-Erick Lafontant
- Contributors
- Hyunkeun Ryan Cho (Advisor)Brian Smith (Committee Member)Christopher Coffey (Committee Member)Emily Roberts (Committee Member)Joseph Cavanaugh (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Biostatistics
- Date degree season
- Autumn 2024
- DOI
- 10.25820/etd.007580
- Publisher
- University of Iowa
- Number of pages
- ix, 84 pages
- Copyright
- Copyright 2024 David-Erick Lafontant
- Language
- English
- Date submitted
- 11/27/2024
- Description illustrations
- illustrations, graphs, tables
- Description bibliographic
- Includes bibliographical references (pages 66-73).
- Public Abstract (ETD)
- In biomedical studies, understanding the effects of varied treatment regimens on patient outcomes over time is critical, particularly in complex conditions like Parkinson’s disease (PD), which im- pacts daily life through both motor and non-motor symptoms. Observational studies, such as the Parkinson’s Progression Markers Initiative (PPMI), face unique challenges due to the variability in treatment types, timing of initiation, and changes in dosage levels among participants. Com- mon PD treatments, including levodopa and dopamine agonists, differ in how and when they are administered and adjusted, making it difficult to assess their long-term impacts using standard methods. This dissertation introduces methods to summarize these treatment dosages over time in participants with PD. More specifically, we developed clustering techniques to group participants with similar treatment trajectories, offering clearer insights into how specific treatment regimens affect health outcomes. By applying these methods to data from the PPMI study, we were able to identify impactful treatment groups and enhance our ability to determine which patterns led to improved cognitive and motor functions over time. This research highlights the potential for re- fined treatment evaluation and deeper insights into patient health outcomes in complex treatment environments.
- Academic Unit
- Biostatistics
- Record Identifier
- 9984774665302771
Metrics
18 Record Views