Project Description

CARE-AI (Cognitive Adaptive Rights-based Engagement for AI) is a pioneering healthcare technology platform that digitalizes the i-SUPPORT international standards for pediatric medical procedures, transforming evidence-based children’s rights frameworks into an intelligent, adaptive system accessible to patients, families, and healthcare providers. The platform addresses a critical gap in pediatric healthcare: the absence of scalable, technology-enabled solutions for implementing trauma-informed, rights-based care during clinical procedures.

The project responds to research showing that pediatric medical trauma has lasting psychological impacts, with children who experience negative healthcare encounters developing long-term anxiety, medical avoidance behaviors, and post-traumatic stress symptoms. Despite the development of comprehensive i-SUPPORT standards by over 50 international experts in collaboration with 203 children, 78 parents, and 418 healthcare professionals, these evidence-based guidelines remained largely theoretical without digital implementation tools. I architected CARE-AI to bridge this gap, creating a production-ready system that operationalizes children’s rights under UNCRC Article 12 through intelligent automation and adaptive AI communication.

The platform’s core innovation lies in its age-adaptive AI conversation engine, which integrates Anthropic’s Claude API with developmental psychology principles to deliver procedure preparation support across a 14-year age spectrum (4-18 years). The AI system employs contextual prompt engineering that dynamically adjusts language complexity, information disclosure depth, and interaction patterns based on the child’s age band, cognitive development stage, and specific medical procedure type. The conversation logic respects children’s autonomy by never forcing information, allowing question-led exploration, and maintaining dignity throughout the interaction — core principles embedded directly into the system prompts and response validation layers.

The front-end architecture implements a multi-stakeholder single-page application (SPA) built with React and styled using Tailwind CSS, with three distinct interfaces optimized for different user roles. The child-facing interface features vibrant, encouraging design with large interactive elements, simplified navigation, and a conversational AI chatbot that provides real-time procedure explanations, anxiety management techniques, and rights education. The parent dashboard includes progress tracking visualizations, evidence-based timeline guidance, and detailed breakdowns of how each interaction aligns with i-SUPPORT standards. The clinician portal provides analytics dashboards with patient readiness metrics, standards compliance indicators, and aggregate data for quality improvement initiatives.

The backend integration layer connects Claude’s language model API through a secure REST architecture with rate limiting, error handling, and response caching for optimal performance. Each API request includes structured metadata about the child’s age, procedure type, previous interactions, and specific i-SUPPORT standards being addressed, ensuring contextually appropriate responses. The system implements prompt chaining techniques for complex queries, where initial responses are refined through multiple API calls to achieve age-appropriate language while maintaining medical accuracy.

The platform also incorporates a rights compliance engine that maps each system interaction to specific i-SUPPORT standards and UNCRC articles, creating an auditable trail of rights-based care delivery. The system tracks 10 core standards including information provision, consent processes, support mechanisms, pain management approaches, and dignity preservation — generating compliance reports that healthcare organizations can use for quality assurance and accreditation purposes.

Technical infrastructure prioritizes accessibility and cross-platform compatibility, with responsive design supporting desktop, tablet, and mobile devices. The application employs progressive enhancement principles, ensuring core functionality remains available even in low-bandwidth environments common in resource-constrained healthcare settings. The codebase follows modern JavaScript best practices with component-based architecture, state management through React hooks, and clean separation of concerns between UI logic and API integration layers.

Project Outcome

CARE-AI establishes a functional prototype for how AI technologies can be ethically deployed in pediatric healthcare, demonstrating that advanced language models can be constrained and contextualized to serve vulnerable populations while respecting their rights and autonomy. The platform validates the feasibility of digitally implementing international healthcare standards at scale, with architecture designed for eventual integration into Electronic Health Records (EHR) systems, Hospital Information Systems (HIS), and clinical workflow management platforms. Beyond its immediate application in procedure preparation, the system serves as a blueprint for rights-based AI deployment in healthcare contexts, showcasing techniques for age-adaptive communication, multi-stakeholder interface design, and standards-compliant automated decision support. The project positions technology as an enabler of better healthcare outcomes while centering children’s voices and choices in their medical care experiences.