(Informative)

1     Introduction 2     Actors 3     Services

1        Introduction

The overall architecture of the AI Health Secure Platform, in the following also called AIH Platform, comprises a set of different systems, specific distributed services, and APIs as depicted in Figure 1.

Figure 1 – Reference Model of AIH Platform

A concise description of the AIH Platform is the following:

  1. End Users acquire Health Data using AIH Apps running on their handsets – in the following also called AIH Frontends.
  2. Health Data is AI-processed using the local AI Framework – in the following, processed and unprocessed Health Data is called AIH Data
  3. End Users may upload their AIH Data to the AIH Platform’s AIH Backend according to the terms of the Smart Contract between the End User and the AIH Backend.
  4. The Smart Contract resides on a Blockchain and regulates the use that the AIH Backend and its Third-Party Users can make of the AIH Data.
  5. The Backend processes AIH Data in its local AI Framework.
  6. Registered Third-Party health-related entities – in the following called Third-Party Users – may access the AIH Backend and request AI-processing of AIH Data present on the AIH Back End.
  7. A Third-Party User may access and process AIH Data present in the AIH Backend based on the terms of a Smart Contract between the AIH Backend and the relevant End User.

2        Actors

The AIH Platform identifies and recognizes the following:.

  1. Users
    1. AIH Frontend User (End User): a user who collects and processes AIH Data their with their AIH Front Ends, and licenses their AIH Data to the AIH Back End.. The End User controls and audits the access of any Third-Party Entity to their AIH Data based on the terms of a Smart Contract issued at the time and to the Third-Party User who requires access to and processing of that AIH Data.
    2. AIH Backend User (Third-Party User): any third-party entity requiring access to the data on the system or to process that data and extract knowledge through the usage of some AI-based mechanism (or through the orchestration of multiple AI-based mechanisms). Third-Party Users include hospitals, clinics, research centers, caretakers, and others identified according to the Taxonomy specified by MPAI-AIH. Access is granted according to the Smart Contract between that Third-Party User and the End User. The Smart Contracts are based on approved templates that are verified for legal compliance and technical security before release.
  2. Data
    1. AIH Data: collected, locally processed, and uploaded  by End Users to the AIH Back End, and stored, processed, and sublicensed  by the the AIH Back End to Third-Party-Users respecting the Terms specified by the relevant End-User.
    2. External Data Sources: represent data from platforms other than the AIH Platform from which the AIH Back End may collect subsidiary data for the integration of relevant information for health-related predictions. Access and Provenance of External Data Sources are regulated via Smart Contracts between the Sources and the AIH Backend.

3        Services

Figure 1 depicts the AI Health data system composed of a set of distributed components and services:

  1. The AIH Frontend, the application (AI-Health App running on a smart device (e.g., a smartphone)  that is capable of:
    1. Capturing End User’s Health Data, e.g., from Google Fit and Apple Health, and from external biometric sensors that capture Health Data.
    2. Locally storing Health Data in a “Secure Data Vault” controlled by the End User (see Figure 2)
    3. AI processing Health Data using standard AIMs and AIWs downloaded from the MPAI-Store which perform the computational operations of the End User’s Health Data, including transformations, training, and inferences.
    4. Alerting the End-User about any deviation of the value of the captured Health Data that may be caused by disease.
    5. Uploading the processed Health Data (AIH Data) to the AIH Back End.
    6. Receiving Health Data further processed using the End User’s and other End Users’ AIH Data.

The AIH Front End represents the personal gateway to the user-data and any external biometric sensors that capture health-related data, and the connection with the AIH-Backend.

 

 

Figure 3 – AI Front End architecture of the AIH Platform

  1. The AIH Back-end, composed of a set of tools that implement the necessary services to securely store, de-identify and anonymize data, control entity authentication and access to data, and license and audit the access to AIH Data on the AIH Back End. The AIH Back End gathers anonymised data from all the various sources (the End Users providing it and other External Sources) and acts as a broker gateway between the Third-Part Entities requesting access to the AIH Data and those who provide it. The backend grants the rights without referring to the identity of the End Users who have provided the data. The Back End may not grant the Third-Party User the right to process The AIH Data that the relevant End User did not grant to the AIH Back End.
  2. Blockchain and Distributed Ledgers (B&DLT) enable the transparency and auditability of the system. Each provision of and access to AIH Data requires the emission of a license in the form of a Smart Contract that is stored on the B&DLT. The Smart Contract contains information about:
    1. The parties, e.g., the End User storing AIH Data and the AIH Back End, and any future Third-Party User requesting access to and processing of AIH Data.
    2. The Type of Third-Party User (per the AIH Taxonomy).
    3. The AIH Data and AIH Models to be used.
    4. The Rights granted to use the AIH Data:
      1. Type of use of the AIH Data (per the AIH Taxonomy).
      2.  Type of use of the processed AIH Data (per the AIH Taxonomy).
    5. The duration of the Licence.
  3. The AI Services offered by the AIH Platform Backend can be used directly to treat and process the AIH Data on the AIH Front End and extract the specific knowledge sought by the End-User or Third-Party Users based on the licence. These services are selected from those available from the MPAI Store and may be orchestrated to produce specific analyses for the Third-Party Users that request access to AIH Data. AI services through data processing enable specific and customised training of machine learning models to identify and assist in the identification of medical diagnosis and prognosis.
  4. The AI Federated Learning System (FLS) orchestrates the learning of a central model for medical diagnosis and prognosis, namely by working as a medical anomaly detection tool, receiving model weights data from the client models at each AIH Frontend and using it under the terms of the Smart contract that has been established between the End User and the AIH Backend. When an improved model is obtained by the FLS, this is distributed to the client models via a model update.