ADVERTISEMENT
Machine Learning Model Beneficial in Predicting Overall Survival in Patients With Myelofibrosis
According to a presentation at the 64th ASH Annual Meeting, researchers indicated that the use of a simple machine learning model (ML) known as the AIPSS-MF (Artificial Intelligence Prognostic Scoring System for Myelofibrosis) model was associated with an elevated rate of accuracy with regard to predicting overall survival (OS) in patients with primary and secondary myelofibrosis (MF), outpacing other well-established risk scoring systems such as the IPSS and MYSEC-PM.
In this study, researchers collected registry data from patients with MF between the time period of January 2000 and October 2021 in 59 Spanish institutions.
The study involved a total of 1386 patients who were arbitrarily split into a training set which comprised 80% of the cohort and a test set that included 20% of the study cohort. To model overall survival (OS) in the training cohort and to confirm the results in those individuals in the test set cohort, researchers utilized a machine learning (ML) technique (random forests).
The primary outcomes included OS as time for diagnosis of MF to death due to any cause.
Results revealed that based on assessment of eight variables, the ML model was remarkably effective in predicting OS. The eight variables indicated in the publication included: patient age at MF diagnosis, gender, percentage of blasts found in peripheral blood, hemoglobin, platelet count, leukoerytroblastosis in peripheral blood and the occurrence of constitutional symptoms. The ML was reported as exceptional to the IPSS model despite age.
The authors noted that the ML model performed better than those used currently in practice such as the IPSS and MYSEC-PM models. Advantages observed by the authors with regard to the ML model included the following :performs equally in both types of MF, offers a tailored risk estimate for each patient and it is not based on genomic data which allows its use across all healthcare settings.
The presenters concluded that this ML model is a simple and extremely accurate tool that provides clinicians with the ability to predict OS in patients with both primary MF and secondary MF and has advantages over existing risk scoring systems.
Reference: Orgueira AM, et al. Machine Learning Improves Risk Stratification in Myelofibrosis: An Analysis of the Spanish Registry of Myelofibrosis. Presented at 64th ASH Annual Meeting and Exposition. December 10-13, 2022. New Orleans, Louisiana. Abstract: 339.
