A model-driven approach for constructing ambient assisted-living multi-agent systems customized for Parkinson patients

The Parkinson disease affects some people, especially in the last years of their lives. Ambient assisted living systems can support them, especially in the middle stages of the disease. However, these systems usually need to be customized for each Parkinson patient. In this context, the current work follows the model-driven engineering principles to achieve this customized development. It represents each patient with a model. This is transformed into an agent-based model, from which a skeleton of programming code is generated. A case study illustrates this approach. Moreover, 24 engineers expert in model-driven engineering, multi-agent systems and/or health experienced the current approach alongside the three most similar works, by implementing actual systems. Some of these systems were tested by Parkinson patients. The results showed that (1) the current approach reduced the development time, (2) the developed system satisfied a higher percentage of the requirements established for certain Parkinson patients, (3) the usability increased, (4) the performance of the systems improved taking response time into account, and (5) the developers considered that the underlying metamodel is more appropriate for the current goal.