Inspiration
Many people who live in the United Kingdom (U.K.) find it difficult to get a face to face appointment with a real medical doctor π (Yes itβs a struggle). Research has shown that around 44% of people in the UK are struggling to get an appointment (Global Med, 2022). The findings also show that 1 in 4 General Practitioner (GP) Appointments are avoidable π€ Alternatives are available with paid Β£ online sessions to see a doctor π©ββοΈ. Considering the current cost of living crisis in the U.K. (ONS, 2024), charging for this type of service would prevent access to the people who need help the most. This project sets out to provide medical assistance for people in the U.K. with AI (Amazon Bedrock) π¦Ύ and reduce the need for a doctor appointment. Benefits also include reducing the strain on the NHS for medical advice GP appointments.
What it does
Gives U.K. citizens free NHS medical advice from an accredited and published source. The Generative A.I. accumulates and synthesises relevant NHS health solutions / treatments based on your own description of the issues you are facing (symptoms) via user input text. Just like how you would speak and talk to a real medical doctor and get advice or next steps.
How it was built
The cycle of how it was built was the following flow:
Idea -> Try it -> Feedback -> Adjust Idea -> Try it Again -> More Feedback π
Following the above development flow, I set out to DIVE π¦ straight in... I looked through the introduction videos and guidance available via the PartyRock Generative AI Hackathon by AWS. A good starting point is the AWS Bedrock Introduction. Followed by Build your first generative AI application with Amazon Bedrock (AIM218). Once I got some grounding on the fundamentals, I experimented with a few different ideas. I also made a Batman π¦ butler called Alfred who makes me laugh with his analysis on bat cowls. I know irrelevant, though it's good to share that my ideas ranged far and wide (maybe that happens to you too). A bigger problem / issue I've seen and experienced is seeing a doctor. I think my time would be better spent trying to find a solution that could bring greater help to a larger amount of people. The UK NHS Online Medical Doctor was built during my lunch breaks / spare time. The speed to which it was setup was incredibly fast π¨ (No Code!). To make it less generic and specific to NHS users inputs, I spent some time learning Advanced Prompt Techniques on Amazon Bedrock - really good session! Also, other Youtube videos on projects built with AWS Bedrock enabled me to try different ways to craft good replies via A.I. Additionally, during development it was shown to a real NHS Medical doctor for feedback and development adjustments.
Challenges encountered
I'm building this before / after work and during lunch breaks. Like most we have a lot going on outside of work. Time is a big constraint and challenge.
I've never used AWS Bedrock before... so it was learning on the go π . This meant learning how to use and build with AWS Bedrock took time β³. I looked through previous projects and couldn't find something relatable. Idea generation of identifying something that can genuinely help to solve a problem and provide the end user with value took time too.
I wanted the generative A.I. to provide the user with infographics from the advice. Though the image results were not accurate (maybe someone out there can help me create something that can generate better medical infographics based on the generative A.I. responses).
I also tried to format the text so it's easily readable and understood by a wider age range. I formatted the advanced prompts in specific ways to aid better understanding. I looked at reducing the amount of text for the solution. Though narrowing the scope proved difficult as I didn't want to limit the user in pursuing different options (one of them might work).
Accomplishments to be proud of
- Trying and learning something new :)
- Helped me to save time β³ in searching the vast NHS website for possible solutions
- It works π₯³ I and others use it π
What was learned
- You can build out an idea really quickly π¨
- Take the time to work on refinement as it can increase the end user experience dramatically
- Learn by doing (Do Tutorials and Test learnings)
- Disagree and Commit (Perfection paralysis is real)
What's next for UK NHS Online Doctor
- Explore visualisation techniques to turn the information into infographics.
- Test the bedrock A.I. with more users including a wider demographic of intended users
- Seek help from the community to improve and provide better value
- Make the text diagnosis more readable
- Improve the A.I. π€ chat to be more engaging
- Text to speech (audio) version with doctors voice
References
Edited by Dr Shivani Kochhar
Built With
- amazon-web-services
- partyrock
- partyrockaws


Log in or sign up for Devpost to join the conversation.