This is a page that curates AI-related papers published worldwide. All content here is summarized using Google Gemini and operated on a non-profit basis. Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.
Design of an Edge-based Portable EHR System for Anemia Screening in Remote Health Applications
Created by
Haebom
Author
Sebastian A. Cruz Romero, Misael J. Mercado Hernández, Samir Y. Ali Rivera, Jorge A. Santiago Fernandez, Wilfredo E. Lugo Beauchamp
Outline
In this paper, we present a portable, edge-based electronic health record (EHR) platform for resource-poor remote environments. The platform is optimized for offline-first operation, secure patient data management, and modular diagnostic integration, and runs on small form factor embedded devices. It provides AES-256 encrypted local storage and supports optional cloud synchronization for interoperability. Through a use case integrating anemia screening module, we propose a noninvasive anemia diagnosis model using nail pallor analysis and present the performance evaluation results of a Random Forest model (RMSE 1.969 g/dL, MAE 1.490 g/dL, sensitivity 79.2%) and the performance optimization results of a YOLOv8n-based nail bed detector (inference latency 21.50ms after INT8 quantization, mAP@0.5 0.995 maintained). With a focus on low-cost deployment, modularity, and data privacy compliance (HIPAA/GDPR), it seeks to address key barriers to digital health system adoption in unconnected environments.
Takeaways, Limitations
•
Takeaways:
◦
Presenting the possibility of building an effective offline-first electronic health record platform in a resource-constrained environment
◦
Integration and expandability of various diagnostic tools through modular design
◦
Increased reliability through secure data management and privacy compliance
◦
Suggests potential for improving access to health care in remote areas at low cost
◦
Validation of the practicality of a noninvasive anemia diagnosis model based on nail pallor analysis
•
Limitations:
◦
The training dataset size of the anemia diagnosis model (250 people) is relatively small.
◦
Currently, the use case is limited to anemia diagnosis, and verification of expandability to other disease diagnosis is required.
◦
Additional evaluation of platform performance and stability in various environments is needed.
◦
Lack of specific strategies for data management and system maintenance for long-term use
