BayesPocket: Bayesian Causal Inference for Periodontal Diseases in
Longitudinal Studies
Implements the Mixed Treatment-State Causal Model (MTSCM), a Bayesian
framework for estimating causal effects of clinical interventions on bounded
continuous outcomes in longitudinal observational studies with irregular visits.
The methodology is specifically designed for periodontal disease research, where
discrete treatments and continuous disease states (e.g., proportion of periodontal
pockets exceeding 3 mm) reciprocally influence one another under dynamic feedback.
The package integrates a double-censored Tobit likelihood to handle boundary mass
at zero and one, subject-specific random effects to capture within-subject
correlation, and flexible tree-based ensemble priors (standard BART and Soft BART)
to model complex nonlinear interactions without parametric restrictions. Causal
identification is established under the potential outcomes framework via the
G-computation formula, with key estimands including the Mixed Average Potential
Outcome (MAPO) and the Mixed Probability of Disease Resolution (MPDR). The
package provides functions for model fitting, posterior inference, and causal estimand
estimation.
| Version: |
0.1.0 |
| Depends: |
R (≥ 3.5) |
| Imports: |
stats (≥ 4.4.2), GIGrvg (≥ 0.8), truncnorm (≥ 1.0-9), progress (≥ 1.2.3), stochtree (≥ 0.1.1), SoftBart (≥ 1.0.3), parallel (≥ 4.4.2), pbmcapply (≥ 1.5.1) |
| Published: |
2026-05-13 |
| DOI: |
10.32614/CRAN.package.BayesPocket (may not be active yet) |
| Author: |
Qingyang Liu 
[aut, cre],
Debdeep Pati [aut],
Yang Ni [aut],
Dipankar Bandyopadhyay [aut] |
| Maintainer: |
Qingyang Liu <rh8liuqy at gmail.com> |
| License: |
GPL (≥ 3) |
| NeedsCompilation: |
no |
| CRAN checks: |
BayesPocket results |
Documentation:
Downloads:
Linking:
Please use the canonical form
https://CRAN.R-project.org/package=BayesPocket
to link to this page.