Work

MedSecure AI: Intelligent Multimodal Fusion for Smarter Healthcare 01

Year: 2026
Type: article
Abstract: This paper presents MedSecure AI, which is an end-to-end intelligent multimodal medical diagnosis system that canbridge the gap between a unimodal AI system and real-world clinical decision-making. The system simultaneouslyprocesses three different data modalities: a medical scan image using a Vision Transformer (ViT) for deep visual featureextraction, an unstructured clinical text report for contextual NLP embeddings using BioBERT, and structured patientdata through a dedicated neural network module. These representations were dynamically integrated using an attentionbasedfusion mechanism that assigns modality-specific importance weights to each prediction, enabling more reliable andcontext-aware diagnostic output. The system generates a disease classification with confidence scores and produces adownloadable quantum-signed PDF medical report through an interactive Streamlit web interface. Experimental resultsshowed that the trimodal attention fusion model performed better than both single-modality and pairwise-fusion models.
Source: PRATIBODH
Language: English
Open Access status: diamond
OpenAlex ID: W7215262006
FWCI:
Cites: 8
Cited by:
Related to: 10