In contrast to already established tools for health technology assessments, Prospective Health Technology Assessment (ProHTA) uses healthcare hybrid simulation techniques to make early predictions about the effects of innovations. In such studies, it is necessary to consider problems at high abstraction levels, as well as at detailed microscopic levels. According to the diversity of expected output metrics and the lack of medical evidence, the situation can be compared to other large scale and complex simulation problems. In such cases well-known modeling methods are often not sufficient to use. Therefore, advanced modeling and simulation techniques have to be developed and properly applied. This paper describes a methodical and practical approach of hybrid model creation using the simulation tool AnyLogic. We focus on general modeling aspects and on advanced techniques using a Level-Based Architecture that help to develop large scale hybrid simulation models. An implementation of a stroke therapy use-case and its simulation results will be discussed. Finally, some practical ideas for validation will be outlined, as we experienced during the stroke use-case development.