The adoption of the Internet of Things (IoT) drastically witnesses an increase in different domains, and contributes to the fast digitalization of the universe. Henceforth, next generation of IoT-based systems are set to become more complex to design and manage. Collecting real-time IoT generated data unleashes a new wave of opportunities for business to take more precise and accurate decisions at the right time. However, a set of challenges including the design complexity of IoT-based systems and the management of the ensuing heterogeneous big data as well as the system scalability; need to be addressed for the development of flexible smart IoT-based systems. Consequently, we proposed a set of design patterns that diminish the system design complexity through selecting the appropriate/combination of patterns based on the system requirements. These patterns identify four maturity levels for the design and development of smart IoT-based systems. In this paper, we are mainly dealing with the system design complexity to manage the context changeability at runtime. Thus, we delineate the autonomic cognitive management pattern, which is most mature level. Based on the autonomic computing, this pattern identifies a combination of management processes able to continuously detect and manage the context changes. These processes are coordinated based on cognitive mechanisms that allow the system perceiving and understanding the meaning of the received data to take business decisions, as well as to dynamically discover new processes meeting the requirements evolution at runtime. We demonstrated the use of the proposed pattern with a use case from the healthcare domain, more precisely the patient comorbidity management based on wearables.