This blog has been authored by Kimia Mavon.

Machine learning and analytics are increasingly used to improve health outcomes, enhance patient and clinician experiences, and optimize organizational performance within healthcare systems. The foundation of these solutions is data, which continues to grow at an unprecedented rate, particularly in unstructured documentation. Healthcare organizations seeking to leverage this data for machine learning, analytics, or other uses outside of clinical care may be required to de-identify health information of patients. However, manual de-identification of unstructured patient health records is both time-consuming and expensive. Moreover, many automated methods fall short of meeting the stringent requirements of healthcare data privacy, rendering them inadequate to support medical advancements. 

Today, Microsoft is excited to offer a new de-identification service in Azure Health Data Services, empowering organizations to securely de-identify clinical data while preserving its clinical relevance and adhering to the strict standards of the HIPAA privacy rule. 

The de-identification service consists of three operations, “TAG,” “REDACT,” and “SURROGATE.” The surrogation feature maximizes the balance between privacy and utility by replacing PHI elements with realistic surrogates. This process generates synthetic, de-identified data that closely resembles the original data and enables analytics and machine learning models to interact with de-identified, realistic data, found in production environments or at inference. 

The de-identification service enables healthcare organizations to leverage their data in a de-identified format to:  

Train private machine learning models, including generative models, with de-identified data.  
Develop analytics dashboards to drive data-driven decision-making. 
Generate synthetic test data to troubleshoot difficult-to-reproduce issues in test environments.   
Facilitate data sharing across collaborating institutions, fostering the creation of extensive datasets and unlocking opportunities for clinical research and discoveries. 
Conduct longitudinal studies to assess the predictive value of risk factors on diseases without revealing patient data. 

Organizations across the healthcare spectrum can benefit from the de-identification service, with early adopters already planning to leverage the service to help advance some of their most prominent use cases. 

A collaborative research group between Professors David Eyre (Professor of Infectious Diseases), Big Data Institute and Dominic Furniss (Professor of Plastic and Reconstructive Surgery), Nuffield Department of Orthopedics, Rheumatology and Musculoskeletal Sciences, at the University of Oxford, has been investigating de-identification services to support clinical research for the United Kingdom’s National Health Services (NHS) and has recommended the AHDS de-identification service for its performance protecting NHS patient data. 

Doctors Rachel Kuo and Andrew Soltan are developing multimodal foundation models aimed at advancing medical diagnostics and treatment in Plastic and Reconstructive Surgery and Oncology. Doctors Kuo and Soltan work closely with patient partners and require robust de-identification for both patients and researchers. Large volumes of clinical data are required to train models, and automated, efficient de-identification is essential for scaling data availability. By first de-identifying the vast amounts of clinical data needed to train these models, Doctors Kuo and Soltan ensure patient privacy and protect the models against memorization attacks by obfuscating the training data so the model cannot reveal patient identifiers. 

 
At Microsoft, we strive to empower healthcare providers, payors, scientists, and life sciences companies by accelerating their data and AI journey while maintaining a strong commitment to patient privacy.    

Learn more about the service: Quickstart – Deploy the de-identification service (preview) in Azure Health Data Services | Microsoft Learn

Try it out: Health De-Identification Services – Microsoft Azure 

Microsoft Cloud for Healthcare is helping your organization shape a healthier future with data and AI 

We are excited to strengthen our data and AI investments through the Microsoft Cloud for Healthcare. Our healthcare solutions are built on a foundation of trust and Microsoft’s Responsible AI principles. Through these innovations, we are making it easier for our partners and customers to create connected experiences at every point of care, empower their healthcare workforce, and unlock the value from their data using data standards that are important to the healthcare industry. 

Learn more: 

Read more about Azure Health Data Services 
Explore Microsoft Cloud for Healthcare. 

​Microsoft Tech Community – Latest Blogs –Read More