Pre- Stroke Detection using K- Nearest Neighbour and Random Forest Algorithm

Abstract

Stroke is one of the deadliest diseases in the world because it causes the brain's blood vessels to burst, injuring the brain. Symptoms may appear when the brain's blood and other nutrient flow is interrupted. There are various imaging techniques to detect stroke like CT, MRI etc., but these techniques are expensive, time consuming and in these techniques, people need to depend on radiologists for disease diagnosis. The existing model incorporates only software prediction so real time prediction is not possible and also early detection of stroke cannot be predicted so that the treatment given for stroke gets delayed and the severity of the disease is increased To overcome this the proposed system uses a microcontroller and various types of sensors to detect the vital parameters like heart rate, SpO2, temperature, lump, and it also uses machine learning algorithm to detect the stroke in advance. For the accurate detection of the stroke, an efficient Machine Learning technique should be used, and it was created through a unique examination of many ML algorithms. KNN and the random forest algorithm were two machine learning algorithms employed in this case to recognize strokes. The accuracy level of KNN is less than random forest algorithm that is 52% and 94% respectively.