# Shaping the future of multiple myeloma with artificial intelligence and digital twins: from concept to clinic

## TL;DR

This paper explores how artificial intelligence and digital twins can improve risk assessment and treatment decisions for multiple myeloma patients.

## Contribution

The paper introduces AI and digital twin technologies as novel tools for personalized medicine in multiple myeloma.

## Key findings

- AI can create advanced predictive models by analyzing large clinical datasets.
- Digital twins offer dynamic, patient-specific simulations to refine risk assessment.
- Integrating AI and digital twins may enable tailored therapies for high-risk multiple myeloma.

## Abstract

Multiple myeloma (MM) is an incurable hematological malignancy with significant clinical and biological heterogeneity. Despite development and refinement of numerous prognostic models for MM, challenges with accurate and reliable risk stratification remain, highlighted by unexpected, early relapse or progression of disease in patients termed functional high-risk (FHR). To improve decision-making and optimise outcome, there is an unmet need for precise identification of high-risk (HR) patients, to enable tailored therapeutic strategies. With a complex and rapidly evolving treatment landscape, artificial intelligence (AI) and digital twin (DT) technology have emerged as potential tools for personalized medicine in MM. Through the integration and analysis of large data generated in clinical trials, registries and real-world cohorts, AI can inform therapy selection by creating advanced predictive models. DT, virtual patient-specific disease replicas, act as a dynamic, bidirectional bridge between real-world clinical data and computational simulations. Continuous acquisition of patient data, synchronized with DTs through AI-driven architectures, facilitates iterative risk recalibration. This ensures the virtual models accurately reflect evolving disease biology and treatment response. This review provides an overview of current and emerging risk stratification in MM, including genomic-based definitions of HR disease and the concept of FHR MM. We described the role, limitations and controversies of AI and DT in refining risk assessment, their predictive capacity for outcomes and therapy selection. Finally, we provide perspectives on the future of AI application in MM.