Falls mainly occurs among elderly and physically challenged people which results in severe wounds and even cause deaths. The main aim of this research is to create and apply a novel approach to aid in predicting the risk of falls. In order to protect a person from injuries without relying on others, this study suggests a machine learning-based fall prevention and detection system, which will improve their quality of life. Our system prototype contains a smart phone and a smart shoe with four pressure sensors and Wi-Fi communication module and detects a fall using decision tree making algorithm because decision tree making algorithm is best among all machine learning algorithms. It detects normal and cautious values and segregates these values at different points. This alerts user to be more careful whenever cautious gait occurs by sending message, email or call. With the help of all these the chances of falling of elderly people will get reduce.