A Scoping Review of Applications of Machine Learning for the Diagnosis and Classification of Multiple Myeloma

This review systematically maps how machine learning (ML) has been applied to improve the diagnosis of multiple myeloma through serum protein electrophoresis (SPE) and immunofixation electrophoresis (IFE), two tests that remain central but heavily dependent on expert interpretation. By reviewing ML applications ranging from early neural-network models to modern deep learning systems, the study captures the full landscape of automated M-protein detection, band classification, and electrophoretic pattern recognition. Through rigorous scoping review methodology, it synthesizes evidence from diverse datasets, model architectures, diagnostic objectives, and performance outcomes, offering the first comprehensive overview of this rapidly evolving field. Early and accurate detection of multiple myeloma is often limited by human interpretation, variable expertise, and resource constraints, particularly in settings with uneven access to trained laboratory personnel. ML models consistently demonstrate high diagnostic accuracy—sometimes surpassing expert readers—yet their clinical readiness, generalizability, and integration remain unclear without a consolidated evidence base. Understanding where these models excel, where they fail, and what methodological gaps persist is essential for guiding responsible adoption of AI in diagnostic laboratories. This review is therefore crucial for identifying strengths, limitations, and unrealized opportunities such as explainable AI, multi-centre validation, workflow integration, and multimodal diagnostic approaches. The review establishes that ML has strong potential to enhance or even automate components of electrophoresis-based myeloma diagnostics, but real-world impact depends on addressing key gaps in generalizability, transparency, and clinical deployment. It delivers an evidence map, highlights methodological weaknesses, and outlines a clear research roadmap to support future development of trustworthy, scalable AI tools. Ultimately, this project provides essential groundwork for improving diagnostic equity and accelerating early detection of multiple myeloma, especially in resource-limited healthcare systems, while informing laboratories, clinicians, and policymakers considering AI adoption