State occupation probabilities in non-Markov models (Q2304846)

Let \(U=U(t)\), \(t\in[0,\infty)\) be a cadlag process with the state space \(\{1,\ldots, d\}\). Let \(\overrightarrow{p}(t)=(p_1(t), \ldots, p_d(t)\) be the vector of the state occupation probabilities \(p_i(t)=\mathbb{P}(U(t)=i)\) and \(P(s,t)=\{P_{ik}(s,t)\}\) be a transition \(d\times d\) matrix with conditional probabilities \(P_{ik}(s,t)=\mathbb{P}(U(t)=k\,|\,U(s)=i)\). The author of the paper establishes the consistency of the Aalen-Johansen estimator [\textit{O. O. Aalen} and \textit{S. Johansen}, Scand. J. Stat., Theory Appl. 5, No. 3, 141--150 (1978; Zbl 0383.62058)] for the state occupation probabilities by using a new method. This new method is based on a simple identity for the state probabilities and on properties of additive and multiplicative transforms of the special interval functions.