Epidemiological surveillance systems evolve with time, depending on the context and the data already collected. Then, the software used must evolve in order to meet requirements. However, introducing new requirements in order to update the software takes time, is expensive and may lead to the problem of software regression. The problem of failed software developed for epidemiological surveillance are often the result of an unsystematic transfer of business requirements to the implementation. This problem can be avoided if the system is established using a well-defined framework/architecture permitting the rapid development/update of the surveillance software. Empirical research shows on the one hand that Model Driven Techniques such as Model Driven Architecture (MDA) are more effective than code-centric approaches for the development and the maintenance of software. On the other hand, Agile Processes such as Scrum are more effective than Structured Processes when requirements are subject to frequent change. Researchers demonstrated that developers of medical software such as epidemiological surveillance software are experiencing difficulties when following Structured Processes and code-centric approaches. The main goal of this empirical study was to apply the combination of Scrum and Model Driven Architecture for the development of epidemiological surveillance of tuberculosis. During this research, we found the approach ease of use and very useful when the MDA tool can generate the complete source code. It has had positive effects on programmer productivity and satisfaction, cost-effectiveness, timelines and customer satisfaction. In addition, we learned that to involve non-informatic experts in the development/update, the modeling user interface must be as simple as possible.