Occupational diseases represent a significant economic and personal burden. This article proposes an intelligent system capable of predicting abnormal spirometry, a potential indicator of occupational disease, based on the results of a medical examination and environmental factors in the workplace. Three unsupervised techniques—Local Outlier Factor, Isolation Forest, and One-Class Support Vector Machine—in a semi-supervised approach to novelty detection using data from a specific workplace. The results show good system performance, with an accuracy of 97.81% and F1 scores of 98.81% and 86.32% for pathological and non-pathological spirometry, respectively.