Computation in Medicine
Some scenes from the "trenches"...
rudolph.pienaar@childrens.harvard.edu
New approaches to old ideas..
Drivers of medical innovation
From "pharma" to "computation"
From "pharma" to "computation"
Technical design debt
Technical design debt
- The "kernel" is almost vanishingly small
- Supporting structure and software systems
- Always underestimated
- Critical importance
- Deployment environment
Promise of computation
Promise of computation
- Explore complex solution spaces
- Quickly
- Efficiently
- Imaging informatics/analytics
- Data integration
- Genomics
- Molecular folding
The reality of computation
The reality of computation
- There is no such thing as a free lunch
- Commodity solutions do not replace domain knowledge
- Quick and "sexy" is not the same as deep solutions
- Core understanding of the problem and solution are the real keys
Why do they look so bad?
- Bad design paradigms
- No competition
- Closed source
- Locked to a device
- Core conservative philosophy
Multiple, deep data sets exist in the healthcare dataverse
Largely disconnected and independent of each other
clinical
billing
health records
Current
- Unidirectional
- Isolated
- Non-integrative
- Static
Current
- Unidirectional
- Isolated
- Non-integrative
- Static
Future
- Multi-directional
- Collaborative
- Integrative
- Dynamic
How to apply these ideas?
How to apply these ideas?
Research apps
- Sampling of some current
- Sampling of what we do...
What about research apps?
- What do they do?
- What is the development cycle?
- What is the deployment cycle?
Sampling of what we do...
Skeletal Survey Report Tool
"Old Way"
"Old Way"
New way
Normative image reference
http://fnndsc.childrens.harvard.edu/library-anon
ChRIS
Thank you!
Computational_Medicine
By Rudolph Pienaar
Computational_Medicine
An overview of some computational aspects relating to value-enhanced Radiology
- 761