MigraineCloud
Shrey Mohan, Arindam Mukherjee · 2018
Many advancements have been made in the area of health sciences and analytics since the advent of Internet of Things (IoT) and highly accurate predictive models using Machine Learning (ML). The problem of migraine, a debilitating headache which usually causes chronic neurological disorders for for patients, is one such area where both of these technologies can prove to be highly useful and help us gain insights into the causes and prevention of the problem. Prevention of migraine can only come from ubiquitous and autonomous observation and correlation of triggers and environmental parameters peculiar to a patient. However, such observations need to be aggregated without actively engaging the patient thorough questionnaires, because patients are historically reluctant to answer questions in real time, often develop false recollections, and give inaccurate answers. The advent of personal IoT smart devices makes it possible to autonomously collect trigger and environmental data, and find correlation between different factors and migraine attacks in real time so that an accurate prediction of an impending migraine attack can be made, and steps can be taken to prevent such attacks. We propose a deep learning framework - MigraineCloud - which uses a front-end migraine and trigger detection app for personal IoT mobile devices, and a back-end deep learning neural network to learn and subsequently predict the onset of migraine for a particular patient.