Medical imaging is a difficult issue for the blockchain approach. The files are typically huge, and they need to be kept strictly confidential. Blockchains are publicly visible, and they accumulate data without limit. All participants’ records are stored on the blockchain, and they’re never deleted. A blockchain of medical images could grow to gigantic size in a short time, making it difficult to store and update.
The software is called MedNetwork. It handles the privacy issue by encrypting the image data. The patient holds the “private keys or password” for decryption and can grant access rights to others on a selective basis. MedAI argues that this is more secure than existing systems, where each healthcare provider has its own repository. The lack of a consistent method of granting access can lead to security issues when trying to share information.
According to the whitepaper, telemedicine has led to growing use of imaging, and imaging is one of the biggest factors in increasing medical costs. The company expects that its approach will provide a better infrastructure for images and decrease the storage and access costs.
Artificial intelligence is a major component of the design. AI diagnostics, built into the blockchain as smart contract code, will assist physicians in diagnosing conditions from images.
Current status and plans
An alpha release of the software is currently available, and people can sign up to try it out. A beta release, version 0.1, is promised for March 2018. The alpha version includes a prototype for the storage engine and AI code based on algorithms licensed from SemanticMD.
The storage engine focuses on security that will satisfy HIPAA requirements. Imaging conforms to DICOM standards.
The company is based in Singapore and plans to open an office in the US in 2018. Its ICO of 10 million MedAI tokens began with a private pre-sale. The first public phase is scheduled for April 1 to April 14, 2018, and the second for May 1 to May 14. By running an ICO, the company hopes to better engage the blockchain community. MedAI also has backing from Mercatus Capital.
The ultimate goal is “to become the storage layer of the decentralized Internet.”
Questions and concerns
MedNetwork is taking on two challenges at once: the storage of images on the blockchain and the incorporation of AI software to interpret them. Either one is a major task by itself.
To succeed in its aim, MedNetwork needs to address two difficult issues. The first is the scaling problem. Medical images are typically megabytes in size, even with compression. If it’s going to be the destination for images for millions of patients, the blockchain could grow into the terabytes or petabytes. This presents scaling issues, as each copy of the blockchain has to be kept in sync with the others. MedAI needs to explain how it will manage scaling, and what economic incentives storage providers will have for handling so much data.
Another issue is granularity of access. The patient is supposed to be able to grant specified parties access to specific records. According to the whitepaper, records are secured by the patient’s private key, and “without the key, the data is secure and inaccessible.” That seems to imply all-or-nothing permission for all a patient’s records.
A whitepaper is not a technical document, so it’s quite possible answers to these questions exist. Companies developing new technologies are naturally reluctant to reveal too many technical details. Still, MedAI will need to convince investors that it has answers. If it does, it has a solution to one of the most difficult issues in healthcare blockchains and could become a very important player.
Subscribe for the latest reviews on artificial intelligence in blockchain healthcare innovation.
Brennan is a blockchain technical adviser in the healthcare sector and blockchain entrepreneur who has worked on developing proprietary concepts for both artificial intelligence and enterprise blockchain. He is a graduate of Rutgers University School of Health Professions where he earned a M.S. in biomedical informatics.