In the era of Industry 4.0, predictive maintenance (PdM) has become crucial for minimizing downtime ensuring equipment reliability and reducing costs. However, the effectiveness of PdM systems depends heavily on data quality which remains a persistent challenge in industrial AI. This data quality requires a right assessment framework to evaluate quality dimensions. Hence, this paper is to explore the gaps of approaches in Data Quality Assessment Frameworks (DQAF) for PdM by systematically review existing literature to identify their strengths, weaknesses and research gaps. A structured search was conducted in IEEE Xplore, ScienceDirect, Springer and Google Scholar covering the period 2015–2025. Inclusion criteria targeted peer-reviewed works directly addressing data quality assessment in PdM, while irrelevant, duplicate, or non-framework-oriented studies were excluded. Our findings reveal widespread issues, including data heterogeneity, semantic ambiguities, imbalanced datasets and a lack of automation in existing approaches. Machine Learning (ML) driven and ontology-based frameworks show potential in addressing these challenges by enabling semantic interpretation, enhancing anomaly detection, improving data quality and adaptability. The implications of this review provide a clear roadmap for academic researchers seeking to develop advanced Data quality assessment frameworks. Furthermore, it offers industrial practitioners’ insights into building robust and trustworthy PdM systems. This study highlights the significant need for unified and semantically rich frameworks that can adapt to the complexities of modern industrial environments.