Falls represent a serious health concern, especially for the elderly, leading to serious injuries and loss of independence. This causes the second leading accidental injury deaths worldwide. Early detection of falls can help prevent the risks. Hence, a Fall Detection System can play a vital role in senior citizens’ health care. This study offers a novel technique for detecting falls utilizing the Machine Learning model integrated with a Flask app, deployed on Google Cloud Platform (GCP) for real-time data analysis combined with explainable AI techniques. The system processes this data locally on the device before being transmitted to a cloud-based framework for analysis and prediction. The prediction service uses a machine learning model specifically a GBDT model with 99% accuracy in fall prediction and 72% accuracy in label prediction, to determine the likelihood of a fall based on the incoming data. We have implemented SHAP (SHapley Additive exPlanations) for the explainability of our model, which allows us to understand how each feature contributes to the prediction process and provides a clear picture of the factors influencing the system’s judgments. Our system has achieved 97.3% accuracy, 93.75% sensitivity, and 100% precision in fall detection real-time testing and scenarios, consistently observing label performance across multiple trials. This method integrates smartphone sensors with a cloud-based framework to improve fall detection’s accuracy and transparency, showing the activity of the users, which sets it apart from traditional methods. While prior systems have explored smartphone-based sensing and cloud-based ML for fall detection or activity recognition separately, few, if any, offer an integrated, scalable framework that combines multi-output prediction like simultaneous fall and activity detection with real-time explainability (XAI) on a production-ready cloud platform like GCP.