It works for the fixed change-point Cox model $$\Lambda(t|Z, Z_2)=\Lambda_0(t)\exp(Z^\top\beta+\tilde{Z}^\top\gamma I(Z_2>\zeta))$$ under current status data (case-I interval censored data), where $t\in [0, M]$, $Z\in\mathbb{R}^p$, $\tilde{Z}=(1, Z^\top)^\top$, $Z_2\in\mathbb{R}$, $\Lambda_0(\cdot)$ denotes the baseline cumulative hazard and $I(\cdot)$ represents the indicator. A detailed discussion can be found in the article "Estimation of the Change Point Cox Proportional Hazards Model Based on Case I Interval-censored Data" (doi: 10.1007/s00180-025-01677-4).

I recommend considering the Python version first, as the R version was uploaded last year and may cause confusion due to its very low quality. Some $\textbf{VERY BAD}$ things happened when I wrote code in ``R_version" during 2024, which made me unwilling to update them, and now I have finally decided to rewrite it with Python. Actually, this outdated technique should be eliminated. These files are just kept as a souvenir.

$\textbf{Notification: }$ this project is protected by the $\textbf{Copyright Law of the P.R.C}$ (link: https://www.gov.cn/guoqing/2021-10/29/content_5647633.htm) and the $\textbf{Copyright Ordinance (Cap. 528)}$ (link: https://www.elegislation.gov.hk/hk/cap528). Any unauthorized copying, modification, or distribution will result in legal consequences, and the author reserves all rights to enforce its intellectual property.

Copyright © 2025 Qiyue Huang, created in Hong Kong, China. All rights reserved.

21/08/2025, Hung Hom, Kowloon, Hong Kong, China.