The Internet of Things (IoT), acknowledged as a new paradigm, plays a crucial role in various healthcare domains, such as remote monitoring of vital signs and physical activity, early disease detection, and disease risk prevention. Accordingly, the design of an integrated health system built upon IoT presents a significant challenge and depends heavily on the creation of a layered architecture model. The primary aim of this study is to conduct an in-depth review of relevant studies related to IoT-based architectures within the healthcare domain between the period 2020 and 2023. We utilized three prominent scientific databases: Scopus, WoS, and IEEE, and employed Zotero and NVIVO for the extraction and analysis of references. We identify 34 academic studies meeting the specific inclusion and exclusion criteria requirements. Moreover, this study conducted a comparative analysis between IoT and AI-based systems and frameworks, with a particular focus on the types of sensors utilized, the communication protocol employed, the machine learning (ML) or deep learning (DL) algorithms adopted, and the computational unites used (Cloud, Fog, or Edge). To conclude, we analyzed the most frequently employed layers among the various proposals for system architectures. This in-depth examination of architectural layers has led to the identification of five essential layers that consistently appear in most structures: physical, network, access, cloud, and application. These foundational layers play a central role in shaping the effectiveness of IoT architecture design for healthcare applications.