Fall detection systems are crucial for the safety and well-being of elderly individuals, especially those living independently or in assisted living environments. These systems, powered by Internet of Things (IoT) sensors, enable real-time monitoring and alerting, facilitating timely intervention in case of a fall. This helps to minimize injury risks and enhances the health outcomes of elderly populations. Existing fall detection systems predominantly rely on rule-based or threshold-based methods, which may fail to differentiate between falls and other activities, leading to false positives or missed detections. Additionally, limitations in sensor types and placements may hinder the system's ability to capture all fall incidents, particularly in nonstandard positions or locations. This research proposes a machine learning-based fall detection system that leverages IoT sensor data to enhance fall detection accuracy. By utilizing supervised learning algorithms on sensor data from wearable and ambient IoT devices, the system can learn complex patterns, such as acceleration changes, orientation shifts, and spatial context. The proposed system aims to minimize false alarms and improve fall detection sensitivity, offering a more reliable and scalable solution for elderly care.