Toward smarter lean practices: a systematic review of machine learning applications in lean manufacturing
Bahar Serevan, Ali Osman KuşakçıPurpose
This study aims to examine how machine learning (ML) is integrated with lean manufacturing (LM) practices in manufacturing and use “Smarter Lean” as a practice-oriented descriptor for data-driven lean improvement rather than as a distinct theoretical framework.
Design/methodology/approach
Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines, a systematic literature review was conducted on 54 peer-reviewed journal articles published between 2010 and May 2025. The review synthesizes ML techniques, targeted lean tools, operational outcomes, enablers, barriers and emerging research gaps.
Findings
Supervised learning and deep learning dominate current ML–LM applications, particularly for predictive maintenance, quality control, defect detection and process monitoring. The most frequently targeted lean tools are total productive maintenance, quality control, just-in-time, and value stream mapping. Reported outcomes include improvements in defect rates, downtime, cycle time, cost efficiency and safety; however, the magnitude of these improvements varies across contexts and should not be interpreted as universally generalizable. Key enablers include high-quality data, IT and sensor infrastructure, management support and workforce readiness, while major barriers encompass system integration complexity, skill gaps, weak standardization and limited empirical validation.
Originality/value
Unlike prior reviews that examine Lean–Industry 4.0 or Lean Six Sigma integration at a broad technological or strategic level (e.g. Vinodh et al., 2020; Sony et al., 2020), this study offers the first focused tool-level synthesis systematically mapping specific ML techniques to individual lean practices and their implementation conditions. The review contributes to the Lean Six Sigma literature by clarifying practical integration patterns and highlighting critical future research directions concerning real-world validation, scalability, sustainability and ethical considerations.