Site Editor

Leo I. Gordon, MD, FACP


AACR 2023: Can Machine Learning Improve Predictive Prognostics in Mantle Cell Lymphoma?

By: Joshua Swore, PhD
Posted: Monday, May 8, 2023

A call to build more advanced machine learning models that use numerous features from molecular and genomic data was presented at the American Association for Cancer Research Annual Meeting 2023 (Abstract 5377/26). “Current prognostic indicators, such as the [Mantle Cell Lymphoma] International Prognostic Index (MIPI), were primarily designed with patients treated with chemoimmunotherapies,” said Ken Chen, PhD, of The University of Texas MD Anderson Cancer Center, Houston. “Using machine learning and molecular data, we provide a novel predictive method to improve upon conventional clinical markers.”

The study employed data from 785 patients diagnosed with mantle cell lymphoma. Data were labeled as aggressive, indicating relapsed or refractory status, or mild, indicating no relapse after first treatment. A total of 195 features were extracted and engineered from clinical, genomic, pathologic, and cytogenetic data and used within the XGBoost library for gradient-boosting machine learning with 10-fold cross-validation. The group compared models using an integrated feature set, many features from diverse data, and a clinical features set alone.

The authors reported that using a model with integrated features (AUC = 0.82, accuracy = 0.76) outperformed the model using clinical features alone to predict prognosis (AUC = 0.78, accuracy = 0.68). The model was further demonstrated to improve metrics from a similar multivariate logistic model that included all patients. The group went further and used the integrated model with MIPI and other prognostic indices, finding the model outperformed MIPI. Finally, the investigators announced the model has been used in an application programming interface that allows users to enter clinical, pathologic, cytogenetic, and genomic features and to receive a prediction in return.

“Future work will include expanding the features included in the model and using the rest application programming interface to construct a graphical user interface accessible to clinicians and other researchers to make treatment decisions in precision oncology,” the investigator concluded.

Disclosure: For full disclosures of the study authors, visit

By continuing to browse this site you permit us and our partners to place identification cookies on your browser and agree to our use of cookies to identify you for marketing. Read our Privacy Policy to learn more.