Patient data is the fuel for healthcare research
The UK Government has a grand challenge mission to ‘Use data, Artificial Intelligence and innovation to transform the prevention, early diagnosis and treatment of chronic diseases by 2030′
To achieve this mission the NHS needs to make more patient and image data available to researchers. And it needs to do this more quickly and efficiently than it can today.
NHS England’s recent plans to share GP patient records with third parties have highlighted the concerns of some citizens and doctors to wider data sharing. These concerns need to be addressed before this mission can be met, starting with more education and better communication.
New challenge, new tools?
To gain the confidence of the public and professionals in relation to the wider use of patient data for research, we need to provide hospitals and clinicians with better tools to de-identify patient data before sharing.
Although de-identification tools exist, they require manual intervention and template building or involve exposing data to third-party providers over the Internet. And these tools struggle to cope with the full range of image types, formats and data sources, routinely, efficiently and reliably. As data volumes increase these challenges will become more of an obstacle.
Why patient data de-identification?
In practice, de-identification means removing Personal Health Information (PHI) from patient records and images before sharing outside of an NHS Trust. Penalties for PHI leaks may be an issue but the biggest problem is reputation damage to the Hospital Trust and the NHS. Any wide-scale leaks are sure to set back progress and lead to calls for tighter data sharing restrictions, so getting this right is key.
The medical image de-identification challenge
Medical image volumes are increasing as imaging technology becomes smaller, cheaper and more common.
Each image contains patient data, either as metadata or as pixels burnt into the image itself. Unfortunately, there is no universally applied standard where PHI can be stored in the image or in metadata. There is also a big range of image types and formats from different vendor machines and the ways technicians use them. Images clearly present a particularly tough de-identification challenge.
Existing image de-identification tools are narrow in focus, do not scale to meet rising image volumes and are manual or semi-automatic in nature. Some tools require transmitting raw data over the Internet (including Google and Amazon), an unacceptable risk for most Trusts.
Meeting the Government’s grand challenges means we need local, automatic, and autonomous tools for de-identification. And until we do there will always be a risk that personal data will be exposed.
Glendor PHI Santizer: the first fully automated in-situ PHI tool
PHI Sanitizer is a fully automatic AI tool for the de-identification of Medical Images. Its unique features are:
- Works with a wide range of image types (X-rays, CT Scans, MRIs, …) and formats (DICOM, png, jpeg, …),
- Works fully automatically to identify PHI in both burned-in and meta data,
- Does not require templates or manual interventions,
- It is vendor neutral, easy to instal and run in-situ on inexpensive local hardware.
Glendor (www.glendor.com) is a Utah-based startup founded by two Silicon Valley entrepreneurs. The company’s focus is to provide Protected Health Information (PHI) de-identification to accelerate Medical Data sharing and aggregation for research. PHI Sanitizer is its first product.
Want to know more?
Contact us at firstname.lastname@example.org to find out more or organise a live demo.
Glendor Company Contact – Julia Komissarchik, CEO, Glendor, Inc. email@example.com.