NHS Diagnostics: Eliminating Unnecessary Appointments (DNAs) and Improving Patient Outcomes with AI
Exploring how conversational AI could make complex healthcare journeys easier to understand, navigate and prepare for.
Making healthcare simpler with human-centred AI
Across NHS diagnostic services, increasing demand and workforce pressures have created a growing need for more efficient ways to support patients.
As part of a wider AI transformation programme led by Hudson & Hayes, an opportunity was identified to address an estimated 90,000 unnecessary appointments through improvements to patient engagement, operational processes and access to information.
My involvement focused on the patient-facing experience, exploring how AI could make complex diagnostic journeys easier to understand and navigate without compromising accessibility, patient trust or clinical safety.
The challenge extended beyond designing a conversational interface. It required understanding how patients interact with healthcare information, identifying where automation could genuinely help, and creating an intuitive experience that worked alongside existing clinical services.
The result was a series of UX/UI concepts and interactive prototypes exploring conversational support, pre-assessment questionnaires, preparation timelines, appointment information and post-procedure guidance.
The challenge: Complex patient processes
Preparation instructions, medication questions, travel arrangements, appointment logistics and uncertainty around what happens next can create additional cognitive load during an already stressful experience. The original product hypothesis identified recurring questions around preparation, parking, transport, recovery, appointment changes and procedure requirements. When patients cannot quickly find or understand this information, they may contact clinical or administrative teams for clarification. It was to explore whether AI could become an accessible layer between the patient and a complex healthcare journey, helping people find approved information while recognising when a question requires human judgement.
Limited AI literacy among clinicians and leaders, making it difficult to identify, assess, and implement opportunities.
My Role
Working within the wider programme, I focused on UX/UI design and the evolution of the patient-facing assistant.
My contribution included reviewing existing interface concepts, simplifying navigation, exploring alternative interaction patterns, designing patient preparation journeys, improving complex questionnaires and embedding accessibility considerations throughout the experience. A central part of the work involved balancing three competing needs: making the service easy to use, ensuring patients could complete important tasks accurately, and recognising the boundaries of AI within healthcare.
The design approach
Rather than designing the entire experience around a chatbot, I explored how conversational AI could complement more familiar digital interactions.
Patients could ask questions when they needed support, but important information such as preparation stages, appointment details and questionnaire progress should not require a conversation to discover.
This led to an experience built around clarity, visibility and predictable behaviour.
Structured forms supported important health information. A visual timeline helped communicate preparation requirements. Accessible navigation and resources reduced reliance on the chat interface.
Where questions required clinical judgement or fell outside approved information, the proposed experience prioritised human escalation rather than automated certainty.
The outcome
The design work helped translate a broad AI opportunity into tangible, testable patient experiences.
Through iterative prototypes, accessibility reviews and stakeholder collaboration, the concept evolved from a primarily conversational assistant into a more connected patient-support experience.
The resulting direction brought together AI-assisted information retrieval, structured health questionnaires, preparation guidance and accessible service navigation.
While the wider programme identified substantial opportunities for operational improvement, the patient-facing design work remained focused on creating and refining concepts for further evaluation.
The project demonstrated how thoughtful UX/UI design can help define where AI belongs in healthcare, where conventional interfaces remain essential, and where human support must remain central.
