Use of Non-invasive Machine Learning to Help Predict the Chronic Degree of Lupus Nephritis

Using a non-invasive machine learning model based on ultrasound radiomic imaging to analyze features of the kidneys, such as shape and texture, researchers were able to predict the degree of kidney injury in people with lupus nephritis, (LN, lupus-related kidney disease). Currently, a renal biopsy, an invasive test which can cause bleeding, pain and other outcomes, is the most common form of assessing a person’s chronic degree of LN. 

Using radiomics, the ultrasound images of 136 people with LN who had renal biopsies were examined. The images were divided into two groups, a training set and a validation set, and seven machine learning models were constructed based on five ultrasound-based radiomics to establish prediction models. The Xgboost model performed the best in the training and test sets. 

Knowing the degree of kidney injury in people with LN can be useful to clinicians as they develop an individual’s treatment plan. Learn more about lupus and the kidneys.

Read the study