Client needs

KIC (Knowledge & Intelligent Computing Lab, University of Ioannina) seeks a privacy-preserving solution for ADHD risk assessment in children. Existing tools lack robust security and compliance, risking data exposure. Requirements include:

• Secure, decentralized data collection via gamified tablet applications.
• Role-based access controls for parents, clinicians, and educators.
• Integration of federated learning and differential privacy to avoid centralizing raw data.
• Compliance with GDPR, eIDAS 2.0, and IDSA standards.
• Scalable consent management inspired by ConInSe architecture.

Forkbomb process

Requirements analysis  

• Collaborated with KIC to define user roles, data flows, and compliance needs.
• Identified integration points for DIDroom, Attribute-Based Credentials, and federated learning (TensorFlow Federated).
• Mapped GDPR and eIDAS 2.0 requirements to technical specifications.

Implementation and solution development

• Configured DIDroom Dashboard for issuing and verifying Verifiable Credentials tied to child status and parental consent.
• Integrated DIDroom DIDroom microservices with the Behavioral Machine Learning Backend to enforce role-based access controls.
• Implemented differential privacy techniques to anonymize aggregated data.
• assistance with deploying DIDroom microservices in a lab setup and in production

Quality assurance and project management

• Conducted weekly alignment sessions with KIC to ensure adherence to timelines.
• Tested credential issuance/verification flows using DIDroom Wallet and Verifier Apps.
• Collected and analyzed data output and logs.

Ongoing support and continuous improvement

• Refined DIDroom’s credential templates based on clinician feedback.
• Continuous follow up for support, improvements and bug fixing

Forkbomb solution

Forkbomb implemented the use case using DIDroom (https://didroom.com/):

• Credential Issuance: Parents receive Verifiable Credentials to approve their child’s participation; clinicians obtain credentials for accessing anonymized risk indicators.
• Federated Learning Integration: Data analysis occurs locally on tablets, with only encrypted model updates shared to the backend.
• GDPR Compliance: DIDroom’s W3C DID Federated Service ensures decentralized identity management, while differential privacy adds noise to datasets.