In industrial workplaces, ensuring the health and safety of workers is essential. Health problems such as stress, fatigue, and physical strain can lead to serious health risks and reduce productivity. Traditional safety monitoring methods are mostly not so effective. They address problems only after they happen. This article presents a predictive digital twin framework to overcome this limitation and make the health monitoring system more effective, that aims to protect workers proactively. The system uses wearable sensors, that connected through the Internet of Things (IoT) to collect health data, which is then analyzed with advanced machine learning models to predict a worker's health condition in real time. By creating a “digital twin” for each worker, supervisors can continuously monitor health status, observe patterns, and receive early alerts before critical situations occur. The framework was tested using a publicly available dataset named WESAD. The dataset includes multimodal physiological signals such as electrocardiogram (ECG) and electrodermal activity (EDA). The results show that the predictive digital twin approach provides high accuracy in detecting potential health risks, which makes it an effective solution for improving occupational health management and workplace safety.