At 09:43 PM 6/10/2007, Kashyap, Vipul wrote:
Matthias,
Thanks for the well organized e-mail, This is clearly one area where
the HCLSIG
community can provide some feedback. Chimezie had similar constructs
related to
"Patient Records" in his POMR. Let's try to distill out some issues so that
HCLSIG/BIONT can give some feedback on this:
1. Some notion of "records" is important at least in healthcare due to
temporal issues and provenance, for e.g., what if another physician
assigns a different diagnosis to the same patient. The patient's disease
is still the same, but there are two versions of the diagnosis in his
Record.
2. Similarly, the same biological fact could be viewed differently by two
different versions of the biological record"
3. That said, we have to figure out a way to partition the world into two
parts: biological or clinical facts/hypotheses; and records that record
those facts and hypotheses.
For e.g. as you suggest:
> 'Protein_2 encoded_by Gene_1'
> 'Gene_1 described_by gene_record_1'
A similar example for the healthcare scenario would be:
Patient_1 suffers_from Disease_1
Disease_1 described_by Snomed_code_1
4. The representation of evidence:
Is evidence a collection of facts that supports the hypothesis or
Inference processes that derive the hypotheses from the facts:
- An example of the former is existence of a cross_reaction, similarity
of substrate specificity
- An example of the latter is "inferred from genomic analysis.."
In the healthcare context, evidence for a diagnosis is typically
phenotypic clinical observations that drive diagnoses or could also
be Pubmed publications as in Evidence based medicine.
It will be great to reconcile the two. And of course there are
Uncertainty related issues here as well!
"Evidence" is important, but is complex. If evidence is from
collection of facts, then exactly how do those facts provide the
evidence (there are many types of analyses you can do on raw data to
produce evidence). So evidence is a function of the facts, the
analysis method, the method of inference, and perhaps even the
observer (e.g., if the evidence is a radiology image or physical
exam, there is inter-observer variation).
And it's definitely necessary to relate the hypotheses to the
evidence with probabilities.
5. Of course this leads to the problems in using process related artifacts
from BFO ontologies. I think HCLSIG has a role to play for further
validation and use of BFO constructs, which are currently not well baked
IMO. Guidelines on how to use BFO constructs is not clear, for instance.
> These issues will be discussed in the BioRDF (BioOnt?) teleconference
> tomorrow.
[VK] I guess there's an overlap with BIONT here... Was wondering if we could
move the discussion out to the next BioRDF call (6/18) as it will give time to
notify the BIONT and Clinical types who might be interested...
This is certainly relevant to the clinical domain (e.g., I'm
interested in it from the radiology perspective)
Regards,
Daniel
Also, Mattias (and Chimezie), was wondering if it would make sense to create
a new wiki on this topic and link it to the BFO Process discussion Wiki and of
course point to it from the BIONT wiki as well?
Cheers,
---Vipul
The information transmitted in this electronic communication is
intended only for the person or entity to whom it is addressed and
may contain confidential and/or privileged material. Any review,
retransmission, dissemination or other use of or taking of any
action in reliance upon this information by persons or entities
other than the intended recipient is prohibited. If you received
this information in error, please contact the Compliance HelpLine at
800-856-1983 and properly dispose of this information.