Knowledge-based Patient Data Generation
Even if Electronic Health Record provide researchers with large amount of real-world clinical data, several problems like privacy issues, technical interoperability, and need for specific contexts, make simulated data still valuable for researchers who want to quickly test their tools or validate what-if scenarios. The clinical knowledge of Computer Interpretable Guidelines can be combined with probabilistic models in order to simulate the execution of a medical treatment and generate realistic patient data.
The candidate, according to the assigned scope (SE/PR/Bak/DA, see table below), has to perform the following activities:
State-of-The-Art Report: write an annotated survey of existing probabilistic approaches for modelling clinical guidelines as well as methods for guideline simulation and data generation.
Basic features: given a specific Computer Interpretable Guideline formalized in Asbru, develop a guideline simulator for generating data for patients and treatments.
Advanced features: extend the probabilistic model and introduce advanced capabilities to simulate data for particular patients within a given cohort or medical context.
Scientific summary: summarize the scientific contribution of the work, with reflections on the lesson learned and future advancement in this research field.
|Writing State-of-The-Art Report||√||√||√|
|Coding basic features||√||√||√|
|Designing/coding adv. features||√||√|
|Writing scientific summary||√||