From predictive maintenance 4.0 to 5.0: bringing humans back into the loop with a self-learning platform and implementation roadmap on automated production lines

Purpose – Industry 5.0 (I5.0) emphasises human-centric collaboration between operators and intelligent systems. This paper presents a Predictive Maintenance 5.0 (PdM 5.0) framework that combines Industry 4.0 (I4.0) technologies with I5.0 human-centric principles. Design/methodology/approach – Real-time data feed artificial intelligence (AI) models that generate probabilistic failure forecasts over short time windows and are explained through explainable artificial intelligence dashboards. A decision-support interface collects feedback from different operator roles to refine the models, and an implementation roadmap supports replication in industrial settings. Findings – Applied to three production lines in an automotive plant, the platform enabled the transition from preventive maintenance to an integrated PdM 5.0 approach, bringing the humans back into the loop, and contributed to an average 20% improvement in overall equipment effectiveness (OEE), together with positive usability scores that capture the operators’ perspective in a PdM 5.0 setting. Social implications – By embedding operators in interaction with the self-learning platform, it supports skill development, transparency and shared control over maintenance decision and a reduced routine workload, contributing to human-centric workplaces and supporting more resource-efficient operations consistently with I5.0. Originality/value – The originality of this work lies in offering the first socio-technical architecture that explicitly frames the transition from PdM4.0 to PdM5.0, responding to the need for new maintenance frameworks co-designed with organisations highlighted in recent literature. Practically, the implementation roadmap makes this transition operationally actionable, showing how to keep humans in the loop in day-to-day maintenance decisions.