2 results about "Rheumatologic Disorder" patented technology

This invention discloses a method for predicting the efficacy of rheumatoid arthritis treatment based on multimodal feature fusion, relating to the fields of artificial intelligence and medical information processing. This method, targeting rheumatoid arthritis patients, integrates ultrasound images, radiomics features, and clinical characteristics to predict the efficacy response of traditional slow-acting antirheumatic drugs (csDMARDs). Specifically, it includes: standardizing and enhancing ultrasound images; extracting image depth features using a pre-trained ResNet convolutional neural network as the backbone model; extracting and standardizing radiomics and clinical features, and fusing them with image depth features through a Cross-Attention mechanism; optimizing feature distribution using supervised contrastive learning (SupConLoss) to improve model discriminative ability; and finally outputting the treatment response category through a fully connected layer. This invention establishes a high-performance predictive model with promising clinical applications by integrating ultrasound images, radiomics, and clinical data.