Adaptive decision support functionality for health monitoring
The growing number of cardiovascular episodes in the western world and particularly in Europe has brought a significant increase in the number of deaths and early disabilities. An important factor is the early diagnosis, but due to the nature of the disease, incidents most usually occur outside the hospital environment. Therefore, reduction of the time before treatment is highly required. Urged by this, the HEARTS project focused on disease prediction rather than on diagnosis, since not only ill and high risk, but also healthy people may also be involved. Towards, this aim the project consortium developed a new tele-health monitoring system that is non-intrusive, dynamic, intelligent, interoperable and based on an open architecture. The system features an advanced and adaptive decision control capability, where classical analytical techniques were integrated with novel approaches. The latter include Neural Network and Non-Linear analyses to reflect the specific health condition of a person. Adopting both disease-centric and the patient-centric approaches for the analysis of the heart disease diagnosis this feature has been realised in the ECG automatic analysis toolkit. More specifically, health status information of a specific subject within a specific context is collected anytime and through innovative neuro/fuzzy processing technologies, it is further analysed. By employing easy interpretability and reasoning capabilities in case of any signs of myocardial ischaemia users are given immediate warning. The modular decision support subsystem is user friendly for medical experts. The toolkit uses an ECG beat recognition method and it is based on a supervised neural network-based algorithm for early detection of myocardial ischaemia from ECG signals. It has been extensively tested and validated employing the European ECG database showing high sensitivity and specificity in the identification of ischemic beats. Moreover, more than 94% accuracy was displayed in beat classification. The reliable decision support modules can be used either as part of the telemedicine system or as standalone units in a medical environment or embedded in medical devices. For information click at: http://heartsproject.datamat.it/hearts(opens in new window)
