This paper presents a holistic method for health monitoring of systems under uncertainty. The ambition is to propose a method that can be applied indifferently to systems usually monitored by methods specific to the dynamics of these systems. Here, we want to be able to monitor continuous dynamic systems, discrete event systems or hybrid systems using the same formalism and the same methodology.To do so, the formalism of Heterogeneous Petri Nets (HtPN) is introduced. Then, the article focuses on an advanced diagnostic method. This method tracks the system health state taking into account uncertainties related to modeling and observations through the construction of an HtPN-based diagnoser. Two application cases illustrate the efficiency of the proposed diagnostic method: one is purely discrete whilst the other is hybrid.