A Machine Learning Approach to Connect Multiple Myeloma Complexity to Early Disease Recurrence

Overview

This is a non-interventional, national, multicenter prospective non-profit observational study aiming at improving the accuracy of risk prediction in multiple myeloma (MM) by applying machine-learning tools for data processing to develop model(s) predicting response to therapy and the probability of early relapse for MM patients.

Improvements in the therapy of MM have prompted the achievement of very deep clinical responses, with significantly improved outcomes and survival \[2\]. However, despite significant therapeutic progress, MM remains a challenge, due to the composite pathogenesis and intricate networks of different interacting factors. As a result, a relatively high proportion of NDMM patients across the different therapeutic strategies have a higher risk of disease progression and worse outcomes, independently of the anti-MM regimen received. These patients currently represent un unmet clinical need, with consequent challenges in the management of the disease and identify these patients upfront is an important goal in MM. At present, risk stratification scores rely on a limited set of clinical and biological variables, not always sufficient to identify patients at a high risk of early disease progression or relapse (i.e., within 12 months from start of first-line therapy). Recently, AI tools have been explored to improve the accuracy of risk prediction, showing that high-risk diseases might be upfront recognized, based on tumor and immune biomarkers \[3-4\]. By applying Machine Learning (ML) tools for data processing, clinical, genomic, and imaging data from MM patients will be integrated and employed in models aimed at improving the accuracy of MM risk prediction. In this way, ML models will aggregate all tumor- and microenvironment-related information obtained by high-throughput technologies and omics approaches to identify and describe clusters of MM patients that best correlate with the achievement of early progression. Overall, this study will identify new knowledge to support clinical research and decision-making in MM: precise up-front stratification of patients, based on the whole landscape of MM-related features, could improve understanding of MM individual risk. Results from this study will have an impact on the possibility to access personalized treatment, with predictable overall repercussion on the effective management of MM patients and savings for the National Health System.

Conditions

Eligibility

Locations (4)

Outcomes

Central Contacts