Astronomers are often interested in classifying the states of active coronae as either flaring or quiescent. Because these states are not directly observable, they must be inferred only from the characteristics of emitted photons captured by detectors. It would be useful to develop a model that can estimate parameters related to the timing of a star's flaring activity, and thus help to elucidate the underlying physical processes that drive stellar flares. We take a state-space modelling approach to this problem by considering the underlying physical process within the star as a latent Markov chain on a continuous state-space; the distribution of the observed photon counts at any time depends implicitly on the state of the underlying chain at that same time. The state predictions from this model are then dichotomized to produce flaring classifications. We demonstrate these techniques using data from the active dMe flare star EV Lac.