Artificial Neural Network and Mobile Applications in Medical Diagnosis

The aim of this paper is to present a pilot study regarding the application of an ANN to stroke recognition and diagnosis. Our study makes use of a (i) a neural network that can be trained to recognize normal limb movements (for individual patients), which may then be coupled to (ii) physical grid mattress that can he used in the patient's home. Any changes in the patient's movement that could potentially indicate that stroke has occurred are transmitted to a mobile phone app. The latter, in turn alerts a relative or ambulance to render rapid assistance to the patient. When stroke has occurred it is essential to transfer the patient to hospital very quickly in order that treatment can be given promptly. In the case of strokes that have arisen due to a blood clot in the cerebral circulation of the brain, a drug called Altcplase (an anti-thrombolytic) must he given within 4.5 hours of the stroke occurring to he maximally effective. Therefore it is important to know the exact time on stroke onset. Our system would record the time of onset of the stroke, by recognizing and recording abnormal changes in the patient's limb movements. A Feed forward neural network was used in our modelling.