AI Can Predict Treatment Outcomes for Diabetic Macular Edema, Study Finds
A new study presents a deep learning model capable of predicting treatment response in diabetic macular edema (DME) using baseline ocular imaging.
WALTHAM, Massachusetts – Research by Voiant and its collaborators has resulted in the publication of a peer-reviewed study detailing a machine learning-based method for predicting treatment outcomes in patients with diabetic macular edema (DME). The findings were published in Nature's Eye.
The study utilized baseline optical coherence tomography (OCT) imaging data from treatment-naïve patients. A deep learning network was trained to forecast both structural and functional treatment responses at the 12-month mark. The model demonstrated high accuracy in predicting changes in central subfield thickness (CST) (R2=0.66) and functional visual acuity (BCVA) (R2=0.63).
The research aims to improve patient selection and enhance the efficiency of clinical trials. Dr. Lorena Ulla from the University of Turin served as the first author, collaborating with Voiant researchers. The results highlight the value of baseline imaging for treatment planning and the success of clinical studies.
Diabetic macular edema is a significant complication of diabetes that can impair vision. Voiant suggests that AI-driven imaging methods hold considerable potential for improving patient outcomes and advancing new therapies.