Predictive Modeling of Occupational Exposure Using Machine Learning and Environmental Sensor Data

Abstract

Such working environment exposures to harmful elements carry a great risk to workers of different works in different industries especially where they work with chemicals, airborne dusts, physical stresses. Classic exposure assessment approaches —those which tend to inform by manual sampling and past history, are relatively blunt in terms of temporal resolution and lag of response. The use of low-cost environmental sensors and machine learning (ML) techniques provide a paradigm-changing means of predictive modelling of occupational exposure, and will be able to allow real-time risk assessment and pre-emptive hazard mitigation Environmental sensors are able to monitor such things as temperature, humidity, particulate matter (PM), volatile organic compounds (VOCs), and noise levels at all times. These data are able to serve as the basis for the creation of predictive models using ML algorithms that determine the exposure trends, prediction of the high-risk scenarios, and dynamic decision-making Supervised learning models, such as random forests and gradient boosting machines, have demonstrated valid usage in predicting exposure result based on multi-variate sensor input. These days, deep learning methods, including recurrent neural networks (RNNs), have shown better results in dealing with temporal data, and in identifying complex patterns of exposure over time