Statistical learning and patient trajectories in healthcare analytics
Healthcare analytics helps improving the treatment quality for patients suffering
from various illnesses. In this regard, one commonly collects patient-related infor-
mation, often about their demography and prior illnesses, in order to predict the
outcome of treatments. We demonstrate this by showing how patient charactistics
can forecast the severity of low back pain. In a next step, we follow an innovative
approach and exploit the prognostic potential of patient trajectories. These stem
from weekly surveys collected throughout a year. By employing a Markov model,
we can then gain a detailed understanding of how pain intensity evolves over time.
This immediately leads to our vision of helping patients with choosing tailored
treatments and the optimal timing thereof.