The FAITH project transitioned from concept to a functional federated AI solution for monitoring the mental health of cancer survivors, focusing on six primary objectives:
1. Defining the FAITH framework: The team identified use cases, user stories, and a reference architecture, ensuring alignment with user and technical requirements.
2. Developing the Federated Learning framework: A proof-of-concept system using the Flower framework was deployed on OpenStack, with automated setup via Ansible playbooks.
3. Ensuring data privacy and security: A robust framework was developed to securely handle patient data, supported by visualization tools for hospital data analysis.
4. Voice/NLP integration: A voice component within the app captured audio data for analysis as potential depression markers.
5. Real-world applicability: Clinical trials and usability studies validated the system's practical use, integrating modules for activity tracking, sleep monitoring, appetite tracking, and voice analysis.
6. Sustainability planning: A business and exploitation strategy was developed, including the CS-AIW cluster, which has grown to over 13 projects and 250 members.
The FAITH project successfully delivered a functional and validated federated AI system for monitoring the mental health of cancer survivors. Our key achievements include:
1. A privacy-preserving data management framework.
2. A mobile app with activity, sleep, nutrition, and voice data modules
3. A federated learning framework for decentralised data analysis.
4. A standalone component to process audio vocal data.
5. A clinical interface to visualise and analyse patient data.
6. A sustainability plan and open-source platform for long-term impact
7. Plans for exploitation of key assets including:
- the open source platform,
- mobile app and an additional 8 x components of the solution
(activity monitor, nutrition and appetite tracker, sleep tracking module, voice module, federated learning framework, advanced analytics, adaptive AI module, activity structure and performance, XAI)
8) Continued dissemination of the project and its results
- Promotion on social media, including regular blog posts
- Final conference organised by clinical partner Champalimaud - ‘AI and Smart Technology in Mental Healthcare for Cancer Survivors’
- Participation in AI brokerage event organised by SETU in Ireland
- Presentation of paper ‘Digital Oncology: supporting clinical trials with biomedical sensors and wearables in cancer treatment and post-cancer follow up’
The project showcased AI's potential to address mental health challenges in cancer survivorship, emphasizing ethical practices and sustainability. Its robust foundation supports future development and commercialization efforts.
