Work

An AI-Driven Multimodal Data Integration Framework for Holistic Patient Profiling in Healthcare Robotics

Year: 2026
Type: preprint
Abstract: With the fast development of healthcare data, there is a lot of opportunities are there for artificial intelligence (AI) to make improvement in disease diagnosis, patient monitoring, and clinical decision-making. While heterogeneous distribution of healthcare information makes comprehensive patient analysis challenging. This paper proposes a multimodal framework AIMDI-HR (AI-Driven Multimodal Data Integration for Holistic Patient Profiling in Healthcare Robotics), that integrates Electronic Health Records (EHR) and physiological vital-sign data which supports intelligent patient profiling and hearth disease prediction. This proposed framework concludes data pre-processing, feature normalization, multimodal data integration, prediction based on machine-learning, and feature importance analysis. In this paper three supervised machine learning models: Logistic Regression, Random Forest, and XGBoost, were estimated using a multimodal dataset which is built from healthcare datasets which are publicly available. After the experiment, the results signify that the Random Forest model got the highest performance with 90.16% accuracy, 89.29% F1-score, and 0.9448 ROC-AUC. Also feature importance analysis indicates contribution of both clinical and physiological features toward predication of disease. The proposed framework shows ability of multimodal AI techniques to increase predictive healthcare system and gives support for decisions for future healthcare robotic applications.
Institution: Parul University
Language: English
Open Access status: green
OpenAlex ID: W7235235069
Cites: 15
Cited by:
Related to: 10