Abstract
Longitudinal trials involving surgical interventions commonly have subject-specific intervention times, due to constraints on the availability of surgeons and operating theatres. Moreover, the intervention often effects a discontinuous change in the mean response. We propose a nonparametric estimator for the mean response profile of longitudinal data with staggered intervention times and a discontinuity at the times of intervention, as an exploratory tool to assist the formulation of a suitable parametric model. We use an adaptation of the standard generalized additive model algorithm for estimation, with smoothing constants chosen by a cross-validation criterion. We illustrate the method using longitudinal data from a trial to assess the effect of lung resection surgery in the treatment of emphysema patients.
Original language | English (US) |
---|---|
Pages (from-to) | 479-485 |
Number of pages | 7 |
Journal | Biostatistics |
Volume | 6 |
Issue number | 3 |
DOIs | |
State | Published - Jul 2005 |
Externally published | Yes |
Keywords
- Back-fitting algorithm
- Cross-validation
- Exploratory analysis
- Longitudinal trials
- Lung resection surgery
- Nonparametric estimator
ASJC Scopus subject areas
- Medicine(all)
- Statistics and Probability
- Statistics, Probability and Uncertainty