Site Editors

Shaji K. Kumar, MD

Prashant Kapoor, MD, FACP


Role of AI-Enabled Predictive Analytics for Patients With Multiple Myeloma

By: Joshua Swore, PhD
Posted: Wednesday, August 30, 2023

Risk stratification in patients with multiple myeloma could be conducted by artificial intelligence (AI) algorithms using fluorodeoxyglucose PET/CT images in the future, according to an article published in the journal Scientific Reports. “Noninvasive methods to stratify patients with high risk could help to optimize multiple myeloma management,” said author Kyung-Han Lee, MD, of Sungkyunkwan University School of Medicine, Seoul, Korea, and colleagues. “In this work, we investigated the prognostic value of image features extracted from [fluorodeoxyglucose] PET/CT scans of patients [with multiple myeloma] by a convolutional autoencoder.”

A total of 191 patients with multiple myeloma underwent fluorodeoxyglucose PET/CT scans and were included in the study. After scanning, the images were manually corrected to include the entire skeleton and exclude nonskeletal structures. Images then underwent convolutional autoencoding with three layers and three max pooling layers. Following autoencoding, the 3,072 extracted features were clustered using unsupervised K-means clustering and a supervised outcome-weighted clustering algorithm, resulting in patients being placed into one of three clusters: A, B, or C.

The authors reported that patients within cluster C had similarly high maximum and mean high standardized uptake value, total lesion glycolysis, and metabolic tumor volume. Cox regression analysis revealed that unsupervised cluster C (hazard ratio [HR] = 3.36), supervised cluster C (HR = 4.07), and high metabolic tumor volume (HR = 2.83) were significant predictors of poor progression-free survival. However, there was not a significant difference between clusters A and B. The authors further found that patients who do not receive autologous stem cell transplantation, have high tumor burdens, and possess the features found in cluster C are at high risk for disease progression and relapse, indicating the need for more frequent follow-up.

Disclosure: The study authors reported no conflicts of interest.

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.