Developing artificial intelligence based assistive systems to aid Persons with Dementia (PwD) requires large amounts of training data. However, data collection poses ethical, legal, economic, and logistic issues. Synthetic data generation tools, in this regard, provide a potential solution. However, we believe that already available such tools do not adequately reflect cognitive deficiencies in behavior simulation. To counter these issues we propose a simulation model (SimDem ) that primarily focuses on cognitive impairments suffered by PwD and can be easily configured and adapted by the users to model and evaluate assistive solutions.