Enhancing Patient Privacy and Reducing Latency with Edge AI

Introduction

A healthcare company specializing in pain assessment required a secure, privacy-compliant mobile application for their pain detection technology. To ensure patient privacy, all processing needed to occur directly on the mobile device, avoiding cloud storage and connectivity. The company collaborated with Darwin Edge and other partners to build a robust mobile application, optimizing AI models to perform efficiently within the power, memory, and processing constraints of mobile devices.

Solution: Transition to Edge AI for Enhanced Privacy and Efficiency

Darwin Edge participated in the development of a fully offline, AI-powered mobile application for automated pain assessment. The application processes data locally on both iOS and Android devices, preserving patient privacy and ensuring reliable functionality in settings with limited or no internet connectivity.

Key features included:

• On-Device AI Processing: all inference is performed on the device, maintaining patient privacy, with no images or videos sent to the cloud.
• Optimized Machine Learning Models: models were fine-tuned for mobile hardware to achieve real-time processing without excessive resource usage.
• Multi-Stage Inference Pipeline: the app was equipped with a sophisticated pipeline, including face detection, landmark recognition and pain scoring, all optimized for edge performance.
• Cross-Platform Development: app was developed for both iOS and Android mobiles and tables, ensuring broader accessibility.

Technology Stack

To develop and deploy this edge AI solution, a comprehensive technology stack was used:

• Mobile App Development: Java and Kotlin for Android and Swift for iOS.
• Machine Learning: PyTorch for model development and fine-tuning.
• Programming: C/C++ for the integration of the models and implementation of the application layer

Results

The newly developed edge AI mobile application delivered a secure, real-time, and privacy-compliant pain assessment tool that operates seamlessly even in environments with no internet connectivity. Thank to this solution the healthcare company was able to achieve:

• Medical Device Certification: The pain assessment solution was approved as a medical device and obtained CE marking, ISO 13485, and pending FDA approval.
• Global Reach: iOS and Android applications are now used in Australia, the UK, and North America.
• Successful Implementation in Care Centers: Widely adopted in elderly care centers across three continents, enabling effective and private pain monitoring for vulnerable populations.

Interested in Edge AI?

By transitioning from a cloud-based model to an edge AI solution, healthcare companies can significantly enhance their data privacy, reduce costs, improve sustainability and eliminate the latency and connectivity issues often associated with cloud processing. This case demonstrates the transformative potential of edge AI for privacy-sensitive applications, particularly in healthcare.

Is your organization looking to transition to Edge AI? Darwin Edge specializes in on-device AI optimization and can help you transition from cloud-based solutions to efficient, secure edge processing. Request a free consultation today to explore how we can create a custom edge AI solution that meets your needs.