1: \begin{abstract}

2:   We consider a unified framework of sequential change-point detection

3:   and hypothesis testing modeled by means of hidden Markov chains.

4:   One observes a sequence of random variables whose distributions are

5:   functionals of a hidden Markov chain. The objective is to detect

6:   quickly the event that the hidden Markov chain leaves a certain set

7:   of states, and to identify accurately the class of states into which

8:   it is absorbed.  We propose computationally tractable sequential

9:   detection and identification strategies and obtain sufficient

10:   conditions for the asymptotic optimality in two Bayesian

11:   formulations.  Numerical examples are provided to confirm the

12:   asymptotic optimality and to examine the rate of convergence.

13: 

14: %eitehr Bayes-optimal or optimal in a Bayesian fixed-error formulation, as the unit detection delay cost or the misdiagnosis and false alarm probabilities go to zero, respectively.  

15: %

16: %and show their asymptotic optimality under two

17: %Bayesian formulations.  (We verify the results numerically using an

18: %example where the observation process is normally distributed.)

19: %

20: %

21: %We develop an asymptotic theory of the hidden Markov model of the change diagnosis problem.  

22: %

23: %Suppose one observes a sequence of random variables whose distributions are functionals of an underlying Markov chain 

24: %

25: %

26: %We develop an asymptotic theory of the hidden Markov model of the joint problem of change-point detection and sequential hypothesis testing

27: %

28: %change diagnosis problem

29: %modeled by \cite{Dayanik2009}.  Based on a sequence of

30: %observations, one needs to detect an unobservable change at the

31: %earliest and identify its change type accurately.  The observation

32: %process is modeled in terms of some functional of an underlying hidden

33: %Markov chain.  

34: \end{abstract}