Healthcare 4.0 is a new concept that originates from hospitals’ evolution due to technological advances in medical activities. Nowadays, more and more doctors and healthcare administrators require real-time data analysis from sensors and surgery monitoring. In such settings, having real-time information may represent the difference between death or life. Currently, the analysis of data from medical settings takes place reactively. Actions to tackle problems with the patients’ health are only taken when critical situations take place. With the arrival of the Internet of Things (IoT) in these environments, the data revolution can allow medical processes to generate many sorts of real-time data automatically. Specialized applications rely on centralized systems to transform medical data into precise feedback so that actuators, whether humans or not, can take the right actions. Although positive, centralized systems suffer from scalability problems. As the connected number of sensors and applications increase, the system might fail due to too many connections. Therefore, quality of service (QoS) is essential because, without it, the applications’ results become unreliable. Although several approaches currently provide QoS for healthcare applications, it is still challenging to produce real-time data from sensors. More specifically, current studies present architecture models that employ different strategies to optimize the time to deliver data from sensors to the final users. Although presenting valuable contributions, they are restrained by the following limitations: (i) do not focus on hospital high critical environments; (ii) do not combine multiple strategies in different levels; (iii) do not put effort specifically in real-time data transmissions; and (iv) do not consider all relevant information for workflow analysis. Given the background, this study proposes HealthStack, a sensor middleware model for hospital settings. HealthStack collects data from sensors, stores them in a database, and delivers them to user applications meeting QoS requirements. HealthStack aims at reducing the delay and jitter in sensor data transmissions for user applications and, at the same time, reduce resource consumption. The model prototype was developed and tested in an operating room with depth cameras and an ultra-wideband (UWB) real-time location system (RTLS) for surgery workflow monitoring. This study presents scientific and technical contributions, and also contributions to society on behalf of hospital services. Its scientific contributions are twofold: a middleware model for healthcare environments with automatic QoS support for real-time data transmission; and a QoS strategy based on artificial neurons to select middleware components with poor performance. The experiments demonstrate that the strategy can improve the applications experienced jitter mean by 92.3% and delay mean by 28% for position data samples. Also, it resulted in a reduction of network, memory, and CPU consumption by up to 66.4%, 5.06%, and 48.3%, respectively. Besides the technical contributions, the solution offers a new level of reliability to time-critical applications directly impacting the patients’ health. This research provides the improvement of medical services for patients, contributing to the hospital administration processes because they can access real-time data with QoS guarantees.