Resumen
The paper proposes a unified declarative approach to developing and integrating medical information systems. Two main usage scenarios for data access are considered ? retrieving patient-based medical history for complex diagnostic scenarios, and retrieving research-based historical data for generalized statistical analysis. The concept of metamodel is introduced as a descriptive layer used to inter-integrate data access, expert system development and evaluation, and user interface for interacting with stored data and predictive models. Proposed metamodel structure can be used in order to generalize and simplify data access and validation independently of specific database or storage solution, and provide a common inter-application communication API. It also can be used to aggregate and translate individual entity information to datasets for developing and verifying supervised machine learning models and verifying rule-based inference models, allowing to create and analyze various types of decision-making systems based on provided data. Finally, using the metamodel for medical information systems development also allows to procedurally generate corresponding form-based and list-based views to rapidly prototype user interfaces based on common controls, for any given record data structure.