A unified computational framework for biomedical sensing, AI and assistive technologies

Abstract

Biomedical engineering is undergoing a paradigm shift as sensing hardware, signal processing, artificial intelligence, and human-centered interfaces converge into integrated systems. The complexity of these domains often results in fragmented solutions, each addressing a narrow problem without interoperability. This paper presents a unified computational framework that organizes over thirty topical areas into an architecture that bridges sensing, signal processing, machine learning, IoT systems, clinical translation, and assistive technologies. The framework is illustrated through offline case studies in electrocardiogram (ECG) analysis, neuroimaging segmentation, and biomedical natural language processing (NLP), each demonstrating reproducible methods with synthetic data. We highlight representative datasets, benchmarking strategies, and practical templates for results reporting. The discussion outlines key limitations, regulatory challenges, and the future direction of neuromorphic computing, federated learning, and accessibility-focused solutions. This framework is the first to demonstrate a cross-modality pipeline integrating signals, images, and biomedical text under a shared computational workflow. While the case studies use synthetic data, they validate architectural interoperability and provide templates for real-world biomedical deployment.