Real-time early detection of allergic reactions based on heart rate variability

Abstract

The popularisation of the concept of “Internet of Things” has promoted the fast increase of applications focused on obtaining information regarding people. For this reason, and thanks to the availability of the computing capacity of smartphones, over the last years a large number of low cost devices and applications have been marketed for analysing the health of users. In this thesis it is proposed to use ECG signals for early detection of allergic reactions. With this aim, a new QRS complex detection algorithm able to work in real time has been designed. This algorithm achieves an accuracy similar to those proposed by other authors, by reducing their computational complexity and the needed resources, which make it able to be implemented in portable platforms. In a previous study the effect that the occurrence of an allergic reaction causes in the heart rate variability was analysed, showing that it is noticeable even before the appearance of physical symptoms in most of the cases in which patients suffered an allergic reaction. However, the method proposed in this previous study is not suitable for detecting allergic reactions during real tests, since the computational complexity of the model designed requires hours of analysis to perform that detection. Moreover, the previous study only focused on food provocation tests in children under 12 years old. The study of the heart rate variability of allergic and non-allergic patients during provocation tests is continued in this work, with two main objectives: the designing of an algorithm capable of detecting allergic reactions in real time, and the extension of the study to include adults and drug provocation tests. The resulting algorithm has an accuracy similar to that proposed in the previous work and the achieved dose and length reduction of the provocation tests is similar as well. However, this algorithm is able to be implemented in a standalone portable device with limited resources and, what is more important, to perform the allergy reactions detection in realtime. Although the results are promising, this study should be interpreted as the beginning of further research, since it is necessary to spend more time and effort in acquiring new data to get a representative sample of the entire population of allergic patients in the case of both food and drug allergies.