AI could automate lupus nephritis diagnosis: Report
Current diagnosis methods like biopsies face limitations, including discrepancies in interpretation by pathologists
image for illustrative purpose
With research being undertaken in the US to leverage Artificial Intelligence (AI) to diagnose lupus nephritis (LN), there is a possibility of automating the detection of the disease, according to a report on Thursday.
Systemic lupus erythematosus (SLE) is a chronic autoimmune that can affect multiple major organ systems in the body. One of its most severe manifestations is renal (kidney) involvement, known as LN.
Although LN can be detected via blood or urine tests, a kidney biopsy is considered the most precise diagnostic approach. However, there are challenges in interpreting biopsy reports due to discrepancies in pathologists' interpretations. The report by GlobalData, a data and analytics company, detailed research efforts to automate diagnosis of LN using AI.
In September 2023, Dr. Mohan and Dr. Van Nguyen, two faculty members from the University of Houston (UH) Cullen College of Engineering, were awarded a $3 million grant from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) to develop an AI programme to aid in the diagnosis of LN.
This funding enables the team to train a “neural network” to read and classify LN biopsy slides. Specifically, the research team intends to build a dedicated computer vision pipeline for classifying LN through the analysis of histopathology imaging data using machine learning (ML).
Collaborating closely with renal pathologists, including experts from various institutions worldwide, the UH team aims to establish a computer-aided diagnosis system for LN, providing clinical decision support akin to renal pathologists.
The application of AI in the LN space was also researched in the past to assess the treatment responses to the disease. In 2021, a team of researchers at the Medical University of South Carolina (MUSC) introduced a pioneering ML algorithm tailored to forecast treatment responses in individuals with LN.
This innovative model incorporated seven key disease indicators to predict the probability of a patient's response to therapy within a year and showed promising outcomes.
