In the area of healthcare, we present today a novel approach for sensor data processing in order to derive abstract level information (such as health condition) based on patient ‘s physiological data (e.g. pulse, temperature). This work follows our previous WASP monitors elderly people on elderly people monitoring with Wireless Sensor Networks.
We recently released a paper [1] on this work, which has been accepted to the 3rd international conference on sensor technologies and applications (SensorCOMM). In this paper, we demonstrate the processing of sensor data based on a contextual ontology. In the scope of European funded research project WASP [www.wasp-project.org], we aim at integrating Wireless Sensor Networks (WSN) [2] with Healthcare solutions. The ability of WSN to monitor and control physical environment made them very attractive for many application domains. And the healthcare domain is not an exception. Nevertheless, Wireless Sensor Networks encounter several technical obstacles, which may hinder their integration within business applications. Considering the fact that Wireless Sensor Networks produce a large and diverse amount of data, business applications may have difficulties to select and process relevant information. We distinguish two approaches which address this issue: in-network processing and data processing within dedicated middleware. Whereas in-network processing mainly aims at resource saving in Wireless Sensor Networks, middlewares ease the definition and execution of Wireless Sensor Network data processing for business applications. In this paper [1], we propose to enhance the common middleware processing approach with the introduction of a contextual ontology. The purpose of this approach is two-fold:(i) to provide only information that match the business application interests, and (ii) to dynamically process Wireless Sensor Network information in order to provide higher semantic level of information to business application. Following this approach, we ease the definition of data processing which are business domain-dependant. For example, Public Security and Healthcare have different types of requirements on sensor data processing; where Public Security applications would need to know if there is a fire in a building, Healthcare application would require a patient’s health status.
We consider, for example, a remote healthcare monitoring application where a patient is monitored remotely at home after surgery. His pulse, body temperature, ambient temperature and his activities are monitored 24 hours per day using a WSN. The latter is connected to a Medical Emergency Response Center (MERC) through a middleware, partially hosted for example by the patient's PDA. This middleware is in charge of detecting any irregularities in patient health condition. The MERC then registers to the middleware for a set of high level information related to patient health condition. The middleware can for example trigger an alert in case of irregularities to the MERC, which contacts a physician for a home visit.
As depicted in Figure 1, we used the Protégé tool for the contextual ontology related to the healthcare domain.
Figure 1 - Contextual Ontology
We developed a prototype based on the depicted approach in order to validate our ideas. We used a google map application as UI for the display of patients and available physicians in a city. As depicted in Figure 2, several patients are monitored around the city. Based on healthcare related information classified and characterized in an ontology, we establish relationship and rule over basic physiological data in order to infer on the patient health condition. Whenever we acquire from the WSN abnormal physiological information, the health condition is evaluated to critical, and an alert is triggered and sent to the hospital. In the google map, the hospital command center can graphically assign a physician to the patient in critical health condition.
As future work, one can envision an integration with the Collaborative Healthcare Network in order to maintain patient’s health condition.
Figure 2. Remote Patient Monitoring
References
[1], Gomez L., Laube A., "Ontological Middleware for Dynamic Wireless Sensor Data Processing", International Conference on Sensor Technologies and Applications, 2009
[2], C.-Y. Chong and S. P. Kumar. Sensor networks: evolution, opportunities, and challenges. Proceedings of the IEEE, 91(8):1247–1256, 2003.
1 Comment
