a very early, unelaborated idea about a Markdown-like specification for quickly writing free text medical notes that can be programmatically parsed into structured data, with terminologies where appropriate
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
medical-markdown-python
.gitignore
README.md

README.md

Medical Markdown

A very early, partially elaborated idea about a Markdown-like specification for quickly writing free text medical notes that can be programmatically parsed into structured data, using standard Academy of Medical Royal Colleges/Professional Records Standards Body document headings, with insertion of clinical coding systems/clinical terminologies where appropriate.

If you haven't previously encountered or used the Markdown system for rapidly formatting HTML documents using plain text delimited by punctuation characters, then I would recommend reading about it here. An understanding of what Markdown is and how it works is advantageous in order to understand both the similarities and differences of Medical Markdown to 'vanilla' Markdown. In common with Markdown, Medical Markdown is both a specification and a series of reference parser implementations.

Additionally, the Extended Specification allows for some real-time features, such as selection of codes from a pick list populated in real-time from a REST terminology server. This pick list can be constrained or pre-processed such as to offer the clinician a subset of the available terminology codes, based on that clinicians' commonly used codes, locally recommended codes, or reference codesets for a particular use-case.

As with Markdown, this specification is not intended to be all-encompassing and we will not implement every possible use-case. That way lies madness. A good example is in prescribing of drugs, where (at present at least) the EPR manufacturer's dialog, with a mixture of numeric and text fields, makes more sense and is probably safer to use. Medical Markdown is intended simply to streamline entry of a clinical consultation into an EPR without text-box overload, get clinicians' hands off their mice, and get them onto their keyboard where they can get stuff done.

This specification is still evolving, and contributions are welcome. Please see the section below on Contributions.

Basic Specification

This is the basic specification for Medical Markdown, which specifies only features that could be included in an asynchronous parser (ie no real-time features) and is here:

'Asynchronous parsing' means the text is first written and saved as a text document, and then a parser processes it at some later stage. The key difference is there is no real-time interaction between the parser and the user, making it simpler to implement.

An example of Basic Medical Markdown

PMH/ => is parsed to the AoMRC heading 'Past Medical History'. The subsequent lines of text are included as structured data under this heading, teminating at the first newline. Where an entry matches exactly the rubric (text portion) of an existing terminology, the plain text can be replaced with structured text including the code.

Much of the standard Markdown specification is included in the Medical Markdown specification, where this made sense and didn't interfere with medical/clinical utility.

Basic Medical Markdown = Daring Fireball Markdown + Medical Markdown bits

Further details are in the medical-markdown Wiki Medical Markdown Specification

Extended Specification

This is the extended specification for Medical Markdown, which includes all of the Basic Specification, and adds some features which can only be offered if the parser is interacting with the user in real-time.

An example of a feature that requires real-time interaction between the user and the parser is interactive selection of a clinical code/terminology from a pick list. This may require any or all of the following data to be processed synchronously, in order for the correct set of codes to be shown to the user:

  • Which section of the document is currently being filled in - for example, in a diagnosis section, one may wish to constrain the available codes to 'diagnosis' codes, in a medication list section, similarly one would constrain the list of options to a local medicines data dictionary.
  • The current user's recent usage of codes - offering the clinician's most-recently-used or most-commonly-used codes
  • Local, National or International guidance on appropriate codes for recording a particular piece of clinical information - for example this allows a medical institution to guide their clinicians to recording using an approved code-set.
  • Natural language processing of the free text of the rest of the clinical document - this may enable an advanced system to intelligently suggest the codes that the clinician might want, based on analysis of the document. But, unlike other suggestions I've heard for automatic-text-to-terminology conversion, the clinician remains in charge of what is actually entered.

Extended Medical Markdown = Basic Medical Markdown + Synchronous Features

Further details are in the medical-markdown Wiki Medical Markdown Specification

Credits and Acknowledgements

  • Major credit is of course due to John Gruber of daringfireball.net, for the original 2004 Markdown specification which inspired this work, and has saved me countless hours of point-and-click formatting. Medical Markdown is, like Markdown itself, released under an open source license.

  • I'd like to acknowledge the influence of those projects that introduced me to the wonders of Markdown and that have gradually spread the utility of markdown all over the web: GitHub, Discourse, Jekyll, Trello, Prose.io, and many others.

  • I'd also like to credit Gina Trapani, creator of Todo.txt, whose inspiration I have also used in the way I've handled the `TODO/`` 'jobs list' tag.

  • The Professional Records Standards Body deserve due acknowledgement for the work done on developing clear standards for medical records in the UK, and the Academy of Medical Royal Colleges and it's constituent colleges are recognised for adding their endorsement to the medical record content standards.

Contributing

Here are some ways to contribute to the Medical Markdown project:

  • Talk about the project with your colleagues and see what they think.
  • Submit improvement suggestions, problems, bugs, or just plain verbal abuse as GitHub Issues (see here for a guide on how to do this, and here for our Issues section)
  • Talk about Medical Markdown with your CCIO, CNIO, or even the vendor of your EPR!
  • Donate to the project, to allow me to spend more time developing it.
  • Spread the word via social media
  • If the AoMRC headings make sense in your country's flavour of medicine, you could to translate Medical Markdown into your language.