AI-Driven Technology to Protect Privacy of Health Data


On May 24th, researchers from the Technical University of Munich, College London, and OpenMined, a non-profit organization published a paper titled end-to-end privacy deep learning on multi-institutional medical imaging.

The research unveiled a privacy-preserving analysis that employs aggregated federated learning and an encrypted approach towards the information obtained from medical imaging. They experimented on pediatric chest X-Rays and used an advanced level deep volutional neural network to classify them.

Although there exist conventional methods, for instance, centralized information-sharing strategies have proved inadequate to protect sensitive information from assaults. This nascent era protects information via using federated learning, wherein only the deep learning algorithm is passed on while sharing the medical data and not the actual content material. Additionally, they implemented secured aggregation, which prevents from outside entities finding the supply where the set of rules was educated. this will not allow anyone to pick out the institution in which it originated, maintaining the privacy intact. The researchers also used any other technique to make certain that statistical correlations are derived from the information records and now not the people contributing the statistics.

According to the paper, this framework is well-matched with a wide kind of medical imaging data codecs, easily user-configurable, and introduces functional enhancements to FL schooling. It will increase flexibility, usability, safety, and performance. “PriMIA’s SMPC protocol guarantees the cryptographic safety of each model and the facts inside the inference section,” states the report.

A report by the College London quotes a professor, who co-authored the paper and says, Our methods had been applied in other research, but we are but to see large-scale studies using actual medical records. Through the targeted development of technologies and the cooperation between specialists in informatics and radiology, we have successfully trained models that deliver specific results at the same time as meeting high standards of data safety and privacy.

With the advancement of technology and the adoption of AI, the healthcare sector has been witnessing a digital boom. With electronic health records and the proliferation of telemedicine, medical data and pictures are abundantly generated every day. To enable better patient monitoring, diagnostics, and availability of statistics, these clinical records are often shared across different points and institutions. This AI-pushed privacy-preserving technology has a potential function to play right here as it does now not compromise information privateness at the same time as sharing takes place.Statistics can not be traced back to individuals, therefore protecting their privacy.

Follow and connect with us on Facebook, LinkedIn & Twitter


Please enter your comment!
Please enter your name here