Paul Bradley – Health Interoperability Expert

This blob is my personal space to share thoughts, insights, and experiences accumulated over the years. It’s a site unafraid to eschew so-called “modern web” best practices. Web development broke when it started prioritising trends over substance. Forget hydration; just frigging use HTML!.

Honestly, if you’re here you probably know me; hopefully there is something useful or vaguely entertaining here.

Table of Contents

• Career
  • Healthcare
  • Public/Private Sector
• Projects
  • Healthcare
  • Side Projects
• Notes
  • AWS HealthLake
  • AWS Medical Comprehend
  • Git Commands
• Contact
• Colophon
  • Access Keys; Keyboard shortcuts used on this page

Notes

AWS HealthLake

HealthLake is a FHIR compliant data store for healthcare information. I’m investigating its capabilities and how it can be leveraged for managing healthcare data effectively. Primarily, I am interested in understanding how it can be used to provide an EPR Downtime Solution.

I’ve investigated using the DocumentReference and DocumentManifest resources to manage and organise healthcare documents within HealthLake. We can convert XDS document entries to these FHIR resources as long as we use contained resources for elements such as authors, custodians, and related entities.

XDS document entries contain home community and repository information, which doesn’t map directly to FHIR fields. As such, we need to add extensions to capture this information. These extensions can be mapped to the corresponding FHIR resources to maintain the provenance and context of the healthcare documents.

As we intend to embed the Patient resource within the DocumentReference resource when searching for documents for a given NHS number we need to use the resource modifier format patient:identifier, instead of the chained search parameter patient.identifier.

After consideration, it’s too expensive for my use case. AWS Aurora DSQL is a much cheaper option; although we do have to design the database schema ourselves.

AWS Medical Comprehend

AWS Medical Comprehend is a natural language processing service that uses machine learning to extract medical information from unstructured text. I’d like to explore whether I can use Medical Comprehend to analyse a discharge summary PDF document and modify it to highlight key medical information for easier review and processing.

For example, it would be nice to colourise different types of medical entities such as medications, diagnoses, and procedures for quick identification.

A tool like this which accepts a source PDF and outputs a highlighted version feels like a project that could be vibe coding in one session, if a sufficient enough prompt was supplied to the LLMs.

Git Commands

Here are some useful Git commands that I frequently use:

• Create a feature branch: git checkout -b {jira-ticket}-feature-description
• Discard changes in the working directory: git restore .
• List files that would be removed by git clean: git clean -n -d
• Remove untracked files & directories: git clean -f
• Move or rename a file: git mv {old} {new}

Rebase with Main Branch

git checkout main
git pull origin main
git checkout {feature-branch}
git rebase main
git push --force-with-lease
            

Colophon

This domain has existed in one form or another since 2009. I actually bought the domain name while on honeymoon, so that the yearly renewal reminder would prompt me to get my wife something special for our wedding anniversary.

Its current inception as a blob was created in September 2026. It's a single HTML file that I edit and maintain. The single page is then processed through a custom static site generator that I developed in Go. The site generator minifies the HTML before creating compressed versions of the page in both GZIP & Brotli formats. The precompressed files are then served directly by the web server (Caddy) to improve performance.

The site is designed to be simple, fast, and easy to maintain, reflecting my philosophy of prioritizing substance over trends in web development. I tried to go with a full No CSS site, but in end I wanted to make use of a sticky header for navigation and use a smooth scroll behavior so the page has a minimal CSS declaration.

Access Keys; Keyboard shortcuts used on this page

Access keys are a cool feature of web browsers introduced in 1999 that allows a certain keyboard combination to jump to a certain part of the webpage. This is without any CSS or javascript, just purely implemented in the browser and HTML. It's a neat way to improve accessibility and navigation for users who prefer keyboard shortcuts. These are the access keys available on this page:

• 0 - Table of Contents
• 1 - Contact