Stanford and VinBrain are developing an AI-based platform for improving diagnosis and treatment based on chest X-ray findings. (Photo courtesy of VinBrain)
As part of an ongoing collaboration, Stanford University and Vietnamese AI company VinBrain are designing an AI-enabled platform to advance medical imaging diagnosis and treatment.
Specifically, they're seeking to improve the accuracy of the RadGraph method, designed to extract clinical entities and relation annotations from large data sets of full-text radiology reports.
Stanford and VinBrain marked the occasion with a Data Use Agreement Signing Ceremony on January 12, which allows them to share relevant deidentified data to train and test the model, starting with 240,000 anonymized medical images and reports.
The RadGraph method was designed by Harvard professor Pranav Rajpurkar and professor Curtis Langlotz, director of the Center for Artificial Intelligence in Medicine and Imaging at Stanford. To improve its accuracy, the platform-in-the-making will be trained on geographically distributed data, medical images and patient medical records.
For VinBrain, the project is also an opportunity to improve the quality of its DrAid platform, an AI doctor assistant software that assists in diagnosing heart, lung, liver and bone diseases from digital medical images.
Its components include DrAid Radiology V1, an FDA-cleared, AI-enabled triage and notification software for clinically assessing adult chest X-rays indicating pneumothorax; and DrAid Appliance, an AI-enabled system for identifying nearly 60 abnormal findings in adult Chest X-rays with higher-than-average quality standards of primary care physicians.
Working with Stanford will allow VinBrain to incorporate data from outside Vietnam, while continuing to develop its solution as the most massive data source there. To date, the platform has analyzed 2.3 million medical images.
The DUA also includes opportunities to expand cooperation as development continues into the next stages; strengthens Stanford’s relationship with VinBrain; and provides VinBrain with greater footing in research dedicated to AI-developed solutions for patient care worldwide.
