 |
Technical Specification: AI for Health (MPAI-AIH) – Health Secure Platform (AIH-HSP) V1.0 published with a request for Community Comments. |
AI for Health (MPAI-AIH) is an MPAI project developing Technical Reports, Specifications, and Reference Software in the Health domain leveraging Artificial intelligence Technologies. Comments should be sent to the MPAI Secretariat by 2026/03/16. Register to attend the online presentation om 2026/02/09T15 UTC.
| Watch video recording (YouTube, WimTV) | Read PowerPoint presentation. | Read An introduction to AIH-HSP V1.0 |
What MPAI-AIH is about
MPAI has developed and published with a request for Community Comments Technical Specification: AI for Health V1.0. This specifies a System where AIH-HSP subscribers use personal devices running AI Framework implementations (front ends) locally process and submit health data to the AIH-HSP back end. A licence attached to health data specifies the types of processes the back end or specific organisations called Third-Party Users may perform and the type of use they can make of process data. The back end is connected to a blockchain storing the smart contracts governing the processes that the back end and the Third-Party Users can perform on subscriber data. From time to time, the backend collects neural networks trained by front ends and distributes an updated neural network used by AIH-HSP subscribers with the knowledge acquired by the community using Federated Learning.
Read What is the AI for Health Call for Technologies about?
MPAI was seeking proposals of technologies to enable the implementation of standard components (AI Modules) that make real the vision described above. The deadline for submitting a response was October 19 at 23:59 UTC. Here were the relevant documents:Keep up to date with the project
See the About MPAI-AIH page.Contact secretariat on how you can participate in this work.
| Call for Technologies (closed) | html, pdf |
| Use Cases and Functional Requirements | html, pdf |
| Framework Licence | html, pdf |
| Template for responses | html, docx |
