Model uses AI to predict prognosis for patients newly diagnosed with multiple myeloma
Click Here to Manage Email Alerts
Researchers at several institutions collaborated to develop the first computational model that predicts prognosis for patients newly diagnosed with multiple myeloma based on tumor genomics and treatments.
The prediction model for individualized risk in newly diagnosed multiple myeloma (IRMMa) uses artificial intelligence (AI) strategies and datasets of newly diagnosed multiple myeloma cases to generate personalized prognoses for individual patients.
The model is the result of collaborative efforts between investigators at Sylvester Comprehensive Cancer Center at University of Miami Miller School of Medicine, Memorial Sloan Kettering Cancer Center, NYU Langone Health, Moffit Cancer Center and Heidelberg University Hospital.
“IRMMa moves the myeloma field forward in the direction of precision medicine,” C. Ola Landgren, MD, PhD, chief of the division of myeloma and director of Sylvester Myeloma institute at Sylvester Comprehensive Cancer Center, told Healio.
“In my mind, precision medicine is the only way forward,” Landgren added. “We need to be more specific and develop treatment strategies for individual patients — in particular, patients with more aggressive disease need better tools to select optimal therapy.”
Healio spoke with Landgren about the creation of the model, the types of prognostic information it can provide and how it has performed so far.
Healio: What inspired your group to create this prediction model?
Landgren: Until recently, all available models developed to predict clinical outcomes focused on the average risk. That is similar to weather reports — for example, there may be a 30% chance for rain. We wanted to test the hypothesis that, using clinical, genomic and treatment data from a large number of patients with multiple myeloma, we could predict an individual patient’s survival outcomes.
Healio: What information is IRMMa able to provide?
Landgren: IRMMa enables us to use AI to predict an individual patient’s survival outcomes based on tumor biology, clinical data and planned treatment. It allows us to answer such questions as “Which therapy will provide the best possible clinical outcomes?” or “If we give the following combination therapy, will autologous stem cell transplant delay progression or extend survival?”
Healio: How did you create the model?
Landgren: We collaborated with several groups worldwide to collect clinical, genomic and treatment data for a large number of patients newly diagnosed with multiple myeloma. In the first version of IRMMa, we collected about 2,000 cases. Using AI strategies, we developed a sophisticated prediction model that is the engine behind IRMMa. It has taken our team more than 3 years to develop this model. We have built the model so that we can continue to feed IRMMa with new datasets and teach it about new treatment strategies. The beauty of AI models is that very feature — you can continue to train them as new information becomes available. I believe there will be more immunotherapy-based therapies in the future, and we will continue to train IRMMa using such datasets. We want to collaborate with other groups around the world so we can feed IRMMa with more data in the future.
Healio: How has this model performed so far?
Landgren: Based on our estimates, IRMMa is about 80% accurate. Think about it — it is better to be 80% correct than randomly guessing and be 50% wrong. As we gather more detailed data from larger numbers of patients, the model will continue to improve and get closer to 90% to 95%, or better.
Healio: Can clinicians use this predictive model?
Landgren: Early on, we made the decision to make IRMMa available to anyone who is interested. IRMMa is a research tool for now. It cannot be used to make decisions for individual patients. We cannot take responsibility for individual patients’ treatment or care.
References:
- IRMMa Risk Calculator – User Guide. Available at: https://irmma-risk-calculator.miami.edu/. Accessed Feb. 1, 2024.
- Maura F, et al. J Clin Oncol. 2024;doi:10.1200/JCO.23.01277.
For more information:
C. Ola Landgren, MD, PhD, can be reached at Sylvester Comprehensive Cancer Center, University of Miami, Don Doffer Clinical Research Center, 1120 NW 14th St., Room #650D, Miami, FL 33136; email: col15@miami.edu.
Collapse
Click Here to Manage Email Alerts