Abstract
Many rare diseases offer limited established treatment options, leadingpatients to switch therapies when new medications emerge. To analyze the impactof such treatment switches within the low sample size limitations of raredisease trials, it is important to use all available data sources. This,however, is complicated when usage of measurement instruments change during theobservation period, for example when instruments are adapted to specific ageranges. The resulting disjoint longitudinal data trajectories, complicate theapplication of traditional modeling approaches like mixed-effects regression.We tackle this by mapping observations of each instrument to a alignedlow-dimensional temporal trajectory, enabling longitudinal modeling acrossinstruments. Specifically, we employ a set of variational autoencoderarchitectures to embed item values into a shared latent space for each timepoint. Temporal disease dynamics and treatment switch effects are then capturedthrough a mixed-effects regression model applied to latent representations. Toenable statistical inference, we present a novel statistical testing approachthat accounts for the joint parameter estimation of mixed-effects regressionand variational autoencoders. The methodology is applied to quantify the impactof treatment switches for patients with spinal muscular atrophy. Here, ourapproach aligns motor performance items from different measurement instrumentsfor mixed-effects regression and maps estimated effects back to the observeditem level to quantify the treatment switch effect. Our approach allows formodel selection as well as for assessing effects of treatment switching. Theresults highlight the potential of modeling in joint latent representations foraddressing small data challenges.