Photo: Al David Sacks/Getty Images
Technologists writing in the journal Nature have developed a digital “mask” dubbed DM, which they say offers a pragmatic approach to protecting patient privacy in electronic health records and during virtual health visits.
There also appear to be clinical benefits to DM.
Based on real-time 3D reconstruction and deep learning, the technology is designed to preserve the clinical attributes contained in patient videos while minimizing access to non-essential biometric information. This is for added privacy in clinical practice.
First experimental results show that with the DM, examination videos of patients with eye diseases can be precisely reconstructed from 2D videos with original faces. A clinical diagnostic comparison showed that ophthalmologists achieved high consistency in reaching the same diagnosis when using the original videos and the corresponding DM-reconstructed videos.
This technology was able to effectively remove identity attributes and received positive feedback from patients with eye diseases. They expressed an increasing willingness to share their personal data and store it digitally with that extra layer of biometric protection.
WHAT IS THE IMPACT
Patient privacy and data usage are often cited as concerns by patients concerned about data breaches. Compared to crude but still widely used options, such as covering identifiable areas with very large bars or cutting out those areas entirely, the DM represents a potentially more sophisticated tool for anonymizing facial images.
The DM selects relevant features for reconstruction, but it is impossible to reconstruct original data relevant for patient identification. And compared to other face-swapping technologies, the DM can obtain quantitative parameters — such as the degree of eyeball rotation, eyelid shape parameters, blink frequency, and rotation frequency — which the authors say could prove essential in the future for intelligently diagnosing diseases or studying the relationships between diseases and certain facial features.
The DM can also be applied to telemedicine, including automated online diagnosis and patient triage. The mask can encrypt data before it is sent to the cloud, allowing doctors or AI algorithms to review the reconstructed data and allay the concerns of patients whose medical records contain sensitive biometric data.
However, protecting privacy is not synonymous with the absolute removal of identity tokens. Under the Health Insurance Portability and Accountability Act privacy rule, protecting patient privacy refers to reducing the risk of identification of health information. One of the most important principles is balancing the risk of disclosure with the benefit of the data.
To that end, the purpose of the DM is to protect health information as much as possible without compromising the need for the clinician to make a diagnosis.
THE BIGGER TREND
Confidence in the security and confidentiality of personal health information is beginning to erode if the results of a recent survey are any indication. The study, published by the American Medical Association, showed that more than 92% of patients believe privacy is a right and their health information should not be available for purchase.
Nearly 75% of the 1,000 patients surveyed by Savvy Cooperative expressed concerns about the privacy of personal health information, and only 20% of patients said they understood the extent of companies and individuals with access to their data.
The survey found that an overwhelming percentage of patients demand accountability, transparency and control over how their health information is protected. About 94% of patients want companies to be legally accountable for the use of their health data, while 93% want health app developers to be transparent about how their products use and share personal health data.
Industries are increasingly being sued by consumers for data breaches, but the sector with the biggest rise in litigation is healthcare, according to results released in April by law firm BakerHostetler.
In fact, according to BakerHostetler, healthcare accounts for 23% of data breach lawsuits. Business and professional services follow at 17%, followed by finance and insurance (15%), education (12%) and manufacturing (10%).
Email the author: [email protected]