I have been archetype wrangling now for 10 years and I must admit that I had initially thought that we would identify patterns pretty early on. This has not been the case. Maybe I'm just a slow learner... but the end result is very pleasing and I'd like to start to share the thinking... Through trial and error and implementation and iteration, we are starting to identify pragmatic and sensible patterns that will fast track future archetype development, but it has been a long and slow process to get to this point.
I know that others in the modelling world claim to have identified patterns, but the logic has eluded me. Some describe patterns for 'scores' - a series of data elements with a total score. Well, that is just representing the data as the original authors have defined it and it is dead simple to mimic that and create an archetype/DCM etc, whatever your favorite flavour of clinical model you prefer. This is NOT a pattern in my mind.
Just recently we have identified patterns for:
- 'Normal' statements
- Exclusion of critical data, which others might refer to as negation - we are much more explicit by creating specific archetypes to minimise any ambiguity around presence or absence assertions
- Simple frameworks that can be extended further with more specific data when clinically appropriate
- Procedure - caters for anything from insertion of a catheter through to triple bypass surgery
- Test results
- Laboratory Test Result
- Panels
- Histopathology
- Microbiology
- Serology etc
- Imaging Test Result
- Ultrasound
- MRI etc
- Laboratory Test Result
- History taking, including symptoms
- Physical examination findings, especially difficult when you consider the fractal nature of medicine and the need for a flexible approach to allow different clinicians to record exam findings in different ways, but based on fixed archetype structures.
Over the next series of blog posts, I plan to gradually explain the basis for these patterns and provide examples that you can find on CKM.
Stay tuned for pattern details to come...