A discussion about application of Bayes' theorem to diagnosis

Tom Benzoni benzonit at GMAIL.COM
Wed Jan 3 16:50:50 UTC 2018


Excellent summary from Dr. Wells, copied from MDCalc site:
*Dr. Phil Wells*
from  the Creator

*Dr. Wells on use of his scores for MDCalc:*

The model should be applied only after a history and physical suggests that
venous thromboembolism is a diagnostic possibility. it should not be
applied to all patients with chest pain or dyspnea or to all patients with
leg pain or swelling. This is the most common mistake made. Also, never
never do the D-dimer first [before history and physical exam]. The monster
in the box is that the D-dimer is done first and is positive (as it is for
many patients with non-VTE conditions) and then the physician assumes that
VTE is now possible and then the model is done. Do the history and physical
exam first and decide if VTE is a diagnostic possibility!

*Dr. Wells on testing in medicine for MDCalc:*

The importance of Clinical pretest probability is underutilized in
medicine. Recognizing the power of a simple concept, derived essentially
from Bayes theorem, that discordance between the clinical PTP and the test
result should raise suspicion of a false negative test (if high PTP) or
false positive test (if low PTP), we sought to derive prediction rules for
suspected DVT and for suspected PE. Used appropriately these rules will
improve patient care.

On Wed, Jan 3, 2018 at 6:21 AM, Bob Latino <blatino at reliability.com> wrote:

> Thank you Dr. Bimal
>
>
>
> This is helpful for me as a layman to understand traditional Bayes Theorem.
>
>
>
> I think your argument applies to what I see in many HC related
> investigations of past undesirable outcomes.  When hypotheses are developed
> about how certain facts came to be, many in HC seem inclined to use a
> 'literature search' as a valid form of evidence to support any given
> hypotheses.
>
>
>
> To me, a literature search merely provides an average of the experience of
> a population of others, and cannot confirm or refute the hypotheses in a
> single undesirable outcome (the focus of an investigation).  I can't see
> going to court and having my evidence be literature search results showing
> that in 7 out of 10 people, this XX was found to be a contributing factor.
> Whatever the literature search revealed as a higher probability would still
> have to be validated that in this specific case, that it did indeed
> contribute to the bad outcome.
>
>
>
> Not sure it's a like-for-like analogy, but in my world it appears to be:-)
>
>
>
> Thanks for helping me clarify how Bayes Theorem works in theory and
> application.
>
> Bob
>
>
>
> *Robert J. Latino, CEO*
>
> Reliability Center, Inc.
>
> 1.800.457.0645 <(800)%20457-0645>
>
> blatino at reliability.com
>
> www.reliability.com
>
> [image: linkedin logo signature file]
> <https://www.linkedin.com/company/958495?trk=tyah&trkInfo=clickedVertical%3Acompany%2CclickedEntityId%3A958495%2Cidx%3A1-1-1%2CtarId%3A1464096807851%2Ctas%3Areliability%20center%2C%20inc.>
>
>
>
> *From:* Jain, Bimal P.,M.D. [mailto:BJAIN at PARTNERS.ORG]
> *Sent:* Tuesday, January 02, 2018 1:10 PM
> *To:* IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
> *Subject:* [IMPROVEDX] A discussion about application of Bayes' theorem
> to diagnosis
>
>
>
> In this attached paper, I discuss application of Bayes ’theorem, as a
> mathematical relationship, to diagnosis, which is a process in the real
> world. This application is not correct, I argue, because the mathematical
> term, probability, does not correspond to evidence during diagnosis.
>
> Please review and comment on this paper.
>
> Thanks.
>
>
>
> Bimal
>
>
>
> Bimal P Jain MD
>
> Northshore Medical Center
>
> Lynn MA 01904.
>
> The information in this e-mail is intended only for the person to whom it
> is
> addressed. If you believe this e-mail was sent to you in error and the
> e-mail
> contains patient information, please contact the Partners Compliance
> HelpLine at
> http://www.partners.org/complianceline . If the e-mail was sent to you in
> error
> but does not contain patient information, please contact the sender and
> properly
> dispose of the e-mail.
>
>
> ------------------------------
>
>
>
> To unsubscribe from IMPROVEDX: click the following link:
> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?
> SUBED1=IMPROVEDX&A=1
>
> or send email to: IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG
>
>
>
> Moderator:David Meyers, Board Member, Society for Improving Diagnosis in
> Medicine
>
> To learn more about SIDM visit:
> http://www.improvediagnosis.org/
>
> ------------------------------
>
>
> To unsubscribe from IMPROVEDX: click the following link:
> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?
> SUBED1=IMPROVEDX&A=1 or send email to: IMPROVEDX-SIGNOFF-REQUEST@
> LIST.IMPROVEDIAGNOSIS.ORG
>
> Moderator:David Meyers, Board Member, Society for Improving Diagnosis in
> Medicine
>
> To learn more about SIDM visit:
> http://www.improvediagnosis.org/






Moderator: David Meyers, Board Member, Society to Improve Diagnosis in Medicine


HTML Version:
URL: <../attachments/20180103/fab99443/attachment.html> ATTACHMENT:
Name: image001.jpg Type: image/jpeg Size: 17486 bytes Desc: not available URL: <../attachments/20180103/fab99443/attachment.jpg>


More information about the Test mailing list