Abstract
The successful application of modern machine learning for time series classification is often hampered by limitations in the quality and quantity of available training data. To overcome these limitations, domain knowledge can be leveraged in the form of parameterised mechanistic dynamical models, whereby time series observations may be represented as instances of a predefined class of dynamical systems. Provided the dynamical models are interpretable in terms of domain-specific variables and their dynamic interaction, the learning process becomes interpretable as well and enables the modeler to handle sparsely and irregularly sampled data naturally. However, the internal processes of a dynamical model are often only partially observed (PO). This can lead to ambiguity regarding which particular model realization best explains a given time series observation. This problem is well-known in the literature, and a dynamical model with this issue is referred to as structurally unidentifiable. Training a classifier that ignores knowledge about a structurally unidentifiable dynamical model can negatively influence classification performance. To address this issue, we employ structural identifiability (SI) analysis to explicitly relate parameter configurations that are associated with identical system outputs. Using the derived relations in classifier training, we demonstrate that this method significantly improves the classifier's ability to generalize to unseen data on a number of example models from the biomedical domain. This effect is especially pronounced when the number of training instances is limited. Our results demonstrate the importance of SI, a topic that has received relatively little attention from the machine learning community.