The study examines how machine learning can be incorporated into a traditional Statistical Process Control (SPC) scheme to improve the early detection of deviations in industrial sewing processes. Using real production data, the researchers trained a supervised model designed to identify combinations of variables that often appear before a quality issue. The model’s performance was compared with standard control charts that apply Nelson rules. Findings indicate that the machine learning model detected deviations 3-to-12-time units earlier than SPC, providing a margin that can be used for preventive adjustments. The team also developed a prototype interface that displays SPC signals alongside model predictions in real time. According to the results, a hybrid strategy that keeps SPC as the central tool and adds machine learning as a predictive layer can support more timely and effective decision-making in manual or semi-automated textile operations.