CarefulAI
  • Accelerating Safer AI Investments
  • Accelerating Safe AI Users
  • Accelerating Safe AI Providers
  • AI Assurance
  • Learning Academy
  • Public Interest AI
  • Contact Us
  • Feedback
  • cn
Co-designing an AI Support Agent for Families: The ND Parent Guide. 

Report SBRI_GDS_GovTech 5


On the development of

https://www.carefulai.com/nd_parent_guide.html


Project Duration: January 2020 - January 2021

Executive Summary

We have completed the "ND Parent Guide" project. Our team designed an AI support agent for families awaiting ADHD assessment. The UK faces an ADHD care crisis, with waiting times often exceeding three years. Our AI agent provides support to carers during these extended waiting periods.  It was designed with families, Torfaen County Council Social Service, and in collaboration with AB NHS University Health Board clinicians.

The agent has shown benefits for families. These include better ADHD knowledge, improved coping strategies, and lower stress levels. Notably, children showed significant reductions in anxiety and depression levels. We've created a framework for similar digital solutions in other healthcare areas.

Key Achievements1. Co-Production WorkshopsWe initially ran workshops with participants known to Carers UK and the Carers Trust. These included parents and clinicians. We achieved gender balance with equal numbers of mothers and fathers. We included parents of both boys and girls with ADHD.

Our workshops delivered:
  • Twelve priority support areas for families
  • User requirements for the AI agent interface
  • Guidance protocols for sensitive topics
  • A stakeholder advisory group

2. AI Agent Development and Testing.  

We created the "ND Parent Guide" AI agent with the carer families. The agent was trained on:
  • Materials from the Social Services and NHS AB Health Board.
  • Educational content from our research team
  • Anonymised peer group discussions
  • Approved health data and research
We tested the agent in three phases:
  1. Technical validation
  2. User acceptance testing with 24 families
  3. Preliminary effectiveness evaluation

The agent includes:
  • Personalised ADHD information
  • Crisis support protocols
  • Healthcare system navigation help
  • Support for diverse family structures
  • Safety mechanisms for human intervention

3. Research Capacity Building and Knowledge ExchangeWe held a conference on "ND Parent Guide" with 96 delegates from the Carer Community. These included healthcare professionals, researchers, service users, and policymakers.

Key Findings 

Usage Patterns
  • The agent was used most in the evenings and at weekends
  • Sessions lasted 18 minutes on average
  • Users returned about three times per week
  • Common topics were school challenges, homework, and emotional regulation

User Outcomes
  • 87% of users showed increased ADHD knowledge
  • Parental stress levels fell by 47%
  • 73% reported improved family dynamics
  • 91% felt more confident navigating healthcare systems

Mental Health Improvements
  • Anxiety reduction: Children in the families that used the AI Agent showed a significant decrease in anxiety as measured by the GAD-7, with mean scores dropping from 12.8 to 7.3
  • Depression reduction: PHQ-9 scores decreased from an average of 11.2 to 6.5
  • Clinical significance: 64% of children showed clinically significant improvements, moving from the "moderate-to-severe" category to "mild" or "minimal" symptoms
  • Sustained results: Follow-up at 3 months showed maintenance of these improvements in 82% of cases
  • Improvements in the status of both parents and children

System Performance
  • The agent maintained appropriate boundaries
  • Response accuracy was 94% compared with expert input
  • User satisfaction was high (4.7/5 average rating)

Cost-Effectiveness: measured by the Nudge Unit of GDS
  • Cost per family: £32 (versus £187 for face-to-face via the Carers Hub)
  • Potential NHS savings of £560 per family
  • Return on investment ratio of about 1:8 at national scale

Challenges and Adaptations

We faced several challenges:
  1. We improved the agent's specific advice through better training data
  2. We refined the agent to support diverse family contexts
  3. We created a simplified interface for those with limited digital access
  4. We enhanced privacy safeguards following user feedback

Future Directions in 2021

We've identified several steps:
  1. Testing the AI Agent with NHS Trusts interested in understanding the value of the agent in waiting time management
  2. The framework could support other neurodevelopmental conditions
  3. Partnership with Universities to study personalised AI interventions in the region's targeted family support
Our project has shown that AI can support families awaiting ADHD care. Our co-production approach ensured we met real user needs while maintaining clinical boundaries.

Post Project Deployment Update

Following the GDS project conducted in 2021, the ND Parent Guide has continued to be used by parents to discuss issues they face in managing their lives and those of their children.  Currently, 59% of parents using the application have children that either have or are expected to have a diagnosis of ADHD.  The application has an active backlog of and product development pipeline and is supported by CarefulAI as part of its commitment to social responsibility.

Within the ADHD parent communities, it is valued.  NHS Trusts and Social services recognise its value.  As and when the demand for ADHD mental health services results in the NHS needing to embed it into clinical pathways, it is recognised that the GDS project evidence and future research may accelerate its adoption.
Our Privacy Policy

Customers Quotes

PRIDAR was the first and remains easiest to use AI Assessment Frameworks"

''AutoDeclare speeds up the process of winning and growing business in our regulated markets."

'"Insurance costs for our GenAI application decreased because of CALMS"


  • Accelerating Safer AI Investments
  • Accelerating Safe AI Users
  • Accelerating Safe AI Providers
  • AI Assurance
  • Learning Academy
  • Public Interest AI
  • Contact Us
  • Feedback
  • cn