Multimodal Clinical System Patent Application US20260112493A1

← USPTO Patent Applications

METHODS, DEVICES, AND SYSTEMS USING MULTIMODAL INPUT FOR CLINICAL APPLICATIONS

Application US20260112493A1 Kind: A1 Apr 23, 2026

Inventors

Yuanxu GAO, Kang ZHANG

Abstract

Disclosed herein is a method for training a model comprising embedding data in a first modality in a first embedding layer to generate a sequence of first modality tokens, and embedding data in a second modality in a second embedding layer to generate a sequence of second modality tokens; passing the sequence of first modality tokens and the sequence of second modality tokens to a plurality of bidirectional multimodal attention blocks to generate a bag of unified tokens, wherein each bidirectional multimodal attention block applies attention to capture intramodal connections and intermodal connections, and wherein both the intramodal connections and the intermodal connections are encoded into latent representations, and representations within the same modality and across different modalities are learned and fused simultaneously.

CPC Classifications

G16H 50/20 G06N 3/0464 G16H 10/60 G16H 30/40

Filing Date

2025-12-10

Application No.

19414783

View original document →