Recently, a report had emerged that AI-driven technology working in conjunction with something called “swarm intelligence” has the capability of detecting some chronic diseases and also give out better treatment solutions to those diseases. However, there was a drawback, the “swarm intelligence” cannot run if different institutes and countries do not share health research data, which of course is against the rules and regulations of certain geographical locations and fraternities. Now, a recent report has emerged, which alludes to the fact that artificial intelligence can work in the reverse scenario as well, i.e., if the parties want to protect their health data instead of sharing it.
New research data has been put forward, which indicates the fact that an unhackneyed technology powered by a robust artificial intelligence algorithm does have the capability to protect the privacy (or “secrecy”) of medical images.
It was Monday, May the 24th, 2021, when a group of collaborative researchers working for institutes and non-profit organizations such as Imperial College of London, OpenMined, and TUM (an acronym for Technical University of Munich) released a paper with a very long title in which they theorized about AI-based technology to protect the privacy of health-based data, and it was titled.
The paper expounded in detail about a tech titled PriMIA (an acronym for Privacy-Preserving Medical Image Analysis) that uses the method aggregation and secure and federate learning along with encrypted storage of data acquired from medical imaging. The paper also revealed that the technology takes no charge to use and can be currently described as an open-source software framework. The researchers have also experimentally verified the tech after borrowing data from pediatric chest X-rays and also used quite a sophisticated convolutional neural network for classification.
It is imperative to understand that there are already quite handy medical data protection toolkits, but they do sometimes fall short and are found to be vulnerable. For instance, something called centralized data sharing has been proven to be inefficient in saving confidential and sensitive medical data from falling into the hands of attackers. Now, this hackneyed technology employs a sophisticated method of federated learning which only passes on deep learning algorithms filtering out the actual content and just sharing the required medical data. The approach of aggregation helps out the tech from exfiltration of any kind of information regarding the source of the algorithm. It will not only protect the sensitiveness of medical data, but it will also keep the source hidden, thereby increasing the privacy of the content. The researchers also embedded a feature in which the AI tech will only let statistical correlation take the data from the records and not from the individuals.
As per the claims made in the paper, the team is quite confident that new tech will be formidable in ensuring privacy for all medical information related to medical imaging. Apart from that PriMIA also boasts features such as user convenience and “configurability” apart from upgrades to FL training, which can prove to be effective in furthering its popularity.
Comments
Post a Comment