EUVIMEDEUVIMED entdecken
Crossref

An EM algorithm for estimating phase-type distributions with left-truncated and right-censored data

Hiroyuki Okamura, Junjun Zheng, Tadashi Dohi

Journal of Applied Probability · 2026-09 · 63 (3) · 1062-1081

DOI: 10.1017/jpr.2025.10068 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Applied Probability, online erschienen am 24.03.2026. Verfasst von Hiroyuki Okamura und 2 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 21 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper introduces a novel expectation-maximization (EM) algorithm for estimating general phase-type (PH) distributions from left-truncated and right-censored (LTRC) data, a common challenge in survival analysis. The proposed algorithm is highly efficient with computational complexity that scales with the number of nonzero elements in the generator matrix. This feature makes the estimation of high-dimensional, sparse PH models computationally tractable and enables the practical use of the computationally intensive extended information criterion for model selection. Numerical experiments demonstrate its significant speed advantage over a modern benchmark and the applicability of PH models to complex lifetime data.

Zitieren

APA

Okamura, H., Zheng, J., & Dohi, T. (2026). An EM algorithm for estimating phase-type distributions with left-truncated and right-censored data. Journal of Applied Probability, 63(3), 1062-1081. https://doi.org/10.1017/jpr.2025.10068

Vancouver

Okamura H, Zheng J, Dohi T. An EM algorithm for estimating phase-type distributions with left-truncated and right-censored data. Journal of Applied Probability. 2026;63(3):1062-1081. doi:10.1017/jpr.2025.10068

BibTeX
@article{Okamura2026,
  author = {Okamura, Hiroyuki and Zheng, Junjun and Dohi, Tadashi},
  title = {An EM algorithm for estimating phase-type distributions with left-truncated and right-censored data},
  journal = {Journal of Applied Probability},
  year = {2026},
  volume = {63},
  number = {3},
  pages = {1062--1081},
  doi = {10.1017/jpr.2025.10068}
}
Kostenlos starten →