Less life threatening differential diagnoses

Vipindas Chengat syncopesystem at GMAIL.COM
Mon Aug 3 01:57:37 UTC 2015


Excellent questions. Here are my thoughts.

1. With fever and cough, how much likely does the patient have pneumonia.

 What if the answer is 76%? For epidemiological data for an ER or an
insurance company, it is a valid information. But, for a clinician, he/she
has only one patient at a time and does not know if that patient belongs to
76% or 24 % or 5%. So, how can this be a relevant number unless he/ she has
gathered all relevant clinical information for that combination of
symptoms?

2. What if the cough is chronic and fever is intermittent for months? What
if the patient has fever, cough, night sweats, weight loss and
lymphadenopathy? Or fever for two days and cough for months? To consider
every single possibility, how many patients should be studied and for how
long?

3. Say, if the patient has cervical lymphadenopathy and analysis has shown
that there is 5% chance of having lymphoma. What does that 5% mean? Whether
the physician should do biopsy or wait? Whether the patient be happy or
unhappy as the number can be perceived as high or low depends on
individuals. Rather, shouldn't the physician try to gather data like the
size and consistency of the LN, presence of systemic signs etc?

4. Our algorithm create dynamic checklists. Example: What if fever and
cough has started after chest trauma or an episode of seizure? These are
specific contexts that are relevant for that combination. It can also look
for other symptoms like purulent sputum or hemoptysis or unilateral leg
swelling.

5. The study that you have mentioned can be used for one diagnosis at a
time. However, what if there are hundreds of diagnoses for a combination of
symptoms? Example, if you suspect PE, you can calculate pre test
probability and then order d dimer or CT angiogram or VQ scan and evaluate
the post test probability based on the performance characteristic of the
test. However, is PE missed due to lack of awareness of the literature or
inability to consider it for uncommon presentations like unexplained
tachycardia? Is diagnosis missed due to lack of availability of data based
on multivariate analysis or inability to see patterns and inability to
gather all the relevant clinical data before making diagnostic assumptions?

6. If I input fever and cough for a 40 year old man, my algorithm considers
all the possible causes- several hundreds and list them in the order of
relevance. How does it test for relevance? How do we get the pre- test
probability ? Well, even practicing physicians don't know what the exact
pre test probability is. We don't try to solve it purely mathematically. We
aren't rounding with tables with LRs and sensitivity/ specificity. But, we
do have an understanding that some are common and some are rare. Some
symptoms and signs and very specific and some are not. To consider certain
diagnoses, we absolutely need certain symptoms. We know what questions and
to ask and try to find patterns that are relevant based on
patho-physiology. Then we look for certain signs that are highly suggestive
of certain diagnoses. Or we do a test that can be used to significantly
reduce the possibility of certain conditions. In brief, this is what my
algorithm is doing. We have identified about 50-60 relations between a
symptom or sign or lab test and a diagnosis and try to apply it
simultaneously. We believe that if the system can create dynamic check
lists for any random combinations, it will improve the performance and
consistency in decision making. If we can get structured studies like you
mentioned that gives us precise numbers; we can plug that information into
my algorithm to improve its performance. However, can we use the same
number for different populations in a different demographic regions?

My humble attempt with the algorithm is to solve a complex problem with a
simple method. Like, catching a fly ball by keeping the angle to the ball
constant rather than using calculus. But the option to use the numbers are
kept open. However, we had only limited success when we used predictive
analytics from a big database compared to current patho-physiologic
algorithm.

Vipin



*Vipindas Chengat, MD FACP**  |  *Chairman, Physician Cognition, Inc.
  —————————————————————————————————
  Mobile: +1 (773) 575-3550
  Email: Vipin at PhysicianCognition.Com <Vipin at PhysicianCognition.com>
  Website: PhysicianCognition.Com <http://physiciancognition.com/>



On Sun, Aug 2, 2015 at 4:59 PM, Mark H Ebell <ebell at uga.edu> wrote:

> The question I”m asking is: “In patients presenting with <symptoms x>, how
> accurate is <sign/symptom/combination> for the diagnosis of <disease y>”.
> For example, “In patients with acute RTI, how accurate are fever, cough,
> and combinations of symptoms for diagnosis of influenza? (or pnuemonia?)”
>
> Not familiar with your database, but to answer these questions you need:
>
>    1. A moderately large group of patients with a clinically relevant
>    spectrum of disease (I.e. Patients presenting to a primary care doc or ER
>    with undifferentiated acute RTI)
>    2. A prospective assessment of a range of signs and symptoms
>    3. An accurate reference standard test (in this case PCR) given to all
>    patients
>
> Clinical decision rules are typically developed using this kind of data
> and using multivariate analysis to identify the optimal set of predictors.
>
> Best,
>
> Mark
>
> From: Vipindas Chengat
> Date: Sunday, August 2, 2015 at 5:43 PM
> To: Society to Improve Diagnosis in Medicine, Mark Ebell
>
> Subject: Re: [IMPROVEDX] Less life threatening differential diagnoses
>
> Very interesting post. Addressing sensitivity/specificity/ LR or
> predictive values would pose another challenge. How would we calculate the
> base rate? Is it just the prevalence of the population or the clinic's
> population? How can we generalize those numbers? Mathematical strategy
> using Bayesian might not be the perfect method; but looking for patterns
> based on patho-physiological relations might be. That is how I built
> physician cognition's decision algorithm and it is doing very well( I
> think). It can even incorporate combined wisdom of professionals to
> optimize the decision rules ( you can find it here
> beta.physiciancognition.com).
> Dr. Ebell, if you would like to conduct any research to optimize decision
> pathways, I can provide it for free of cost. You can try any number of
> symptoms, signs and labs in any combination and system will provide
> differential diagnosis in the order of relevance, suggest further testing -
> can be a clinical sign, symptom or lab and decision tree will move forward
> based on answering those questions. You can also teach the system in real
> time or it can learn from a big database. Please let me know what you
> think. I think I have solved the very specific problem of combination of
> symptoms. I believe that if there are any sub-optimal results, it is just a
> matter of data and not the algorithm and can be solved within minutes.
>
>
>
> *Vipindas Chengat, MD FACP**  |  *Chairman, Physician Cognition, Inc.
>   —————————————————————————————————
>   Mobile: +1 (773) 575-3550
>   Email: Vipin at PhysicianCognition.Com <Vipin at PhysicianCognition.com>
>   Website: PhysicianCognition.Com <http://physiciancognition.com/>
>
>
>
> On Sun, Aug 2, 2015 at 4:08 PM, Mark H Ebell <ebell at uga.edu> wrote:
>
>> While very useful in many ways, that approach would not address
>> sens/spec/LR, would not help with acute problems, and would not address
>> combinations of symptoms, at least not easily
>>
>> Sent from Outlook
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Ftaps.io%2Foutlookmobile&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>>
>>
>>
>>
>> On Sun, Aug 2, 2015 at 9:47 AM -0700, "Jochanan Benbassat" <
>> benbasat at jdc.org> wrote:
>>
>> Unlike hospital based specialists, primary care physicians have the
>> advantage of prolonged follow up of patients. Outpatient records offer the
>> opportunity to determine the predictive value of abnormal syptoms and
>> signs. In 1982 Dr Froom and I published in the BMJ the prevalence of
>> microscopic hematuria in otherwise healthy male airpersonel aged 18-35 and
>> the (negative) results of an average 7 years fillow up. I believe that a
>> fsimilar ocus on the prevalence and outcomes of selected disease
>> manifestations in the community after follow up offers an opportunity for
>> such research
>>
>>
>>
>> Jochanan Benbassat MD
>>
>> Jerusaltm Israel
>>
>> ------------------------------
>>
>> *From:* Mark H Ebell [ebell at UGA.EDU]
>> *Sent:* Sunday, August 02, 2015 16:21
>> *To:* IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
>> *Subject:* Re: [IMPROVEDX] Less life threatening differential diagnoses
>>
>> Blow up NIH. Well ok, not literally. The average age at first RO1 is now
>> 40, and less than 1% of funding goes to what one could generously call
>> primary care research. And none to study of clinical diagnosis. PCORI even
>> expressly forbids research to develop and validate clinical decision rules.
>> This research is not expensive, but proper reference standards (PCR,
>> imaging) cost money.
>>
>> Sent from Outlook
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Ftaps.io%2Foutlookmobile&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>>
>>
>>
>>
>> On Sun, Aug 2, 2015 at 5:19 AM -0700, "Robert Bell" <rmsbell200 at yahoo.com
>> > wrote:
>>
>> Excellent point. How to change things? What to do to get the funding?
>>
>> Rob Bell
>>
>> Sent from my iPad
>>
>> On Aug 1, 2015, at 8:58 PM, Mark H Ebell <ebell at UGA.EDU> wrote:
>>
>> As a primary care physician, we often see diseases early in their course,
>> when signs and symptoms overlap with other conditions and biomarkers may be
>> negative (think lupus). Unfortunately, there is no funding in the US to
>> study clinical diagnosis in the primary care setting, at least not as long
>> as NIH is dominated by basic scientists and sub specialists. That would be
>> the only way to identify the truly useful signs and symptoms (or more
>> likely combinations).
>>
>> Mark
>>
>>>> Mark H. Ebell MD, MS
>> Professor of Epidemiology University of Georgia
>> Editor, *Essential Evidence* Deputy Editor, *American Family Physician*
>> ebell at uga.edu
>>
>>
>> From: Carmel Crock
>> Reply-To: Society to Improve Diagnosis in Medicine, Carmel Crock
>> Date: Saturday, August 1, 2015 at 9:55 PM
>> To: " <IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG"
>> Subject: Re: [IMPROVEDX] Less life threatening differential diagnoses
>>
>> Dear Rob
>> I thought I would mention that at our Eye and ENT hospital emergency
>> department, shingles is one of our commonest missed or delayed diagnosis,
>> as patients present early with severe eye or ear pain but nothing (or
>> little) to show on clinical examination.
>> Regards
>> Carmel Crock
>> ------------------------------
>> *From:* robert bell [
>> <0000000296e45ec4-dmarc-request at LIST.IMPROVEDIAGNOSIS.ORG>
>> 0000000296e45ec4-dmarc-request at LIST.IMPROVEDIAGNOSIS.ORG]
>> *Sent:* Sunday, 2 August 2015 9:08 AM
>> *To:* <IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
>> *Subject:* [IMPROVEDX] Less life threatening differential diagnoses
>>
>> To me it seems that despite frequency patterns medical students are
>> rightly taught at great length to focus on not missing the life threatening
>> diagnosis. Is this done at the expense of missing the diagnosis with less
>> dangerous competing conditions that are on the differential diagnostic
>> list.
>>
>> For example, how much do most medical students/residents know about
>> shingles pain and symptoms prior to lesion development, with any lack of
>> knowledge leading to the “diagnosis” of possible acute abdomen?
>>
>> I would argue that in training knowing FAR more about the less dangerous
>> differential diagnoses would get us more quickly to an accurate diagnosis,
>> at possibly lesser cost.
>>
>> So should there be in training more focus on the less serious competing
>> differential diagnoses?
>>
>> Rob Bell
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> ------------------------------
>>
>> Address messages to: <IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> To unsubscribe from IMPROVEDX: click the following link:
>>
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FSUBED1%3DIMPROVEDX%26A%3D1&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1
>>
>> or send email to: <IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> Visit the searchable archives or adjust your subscription at:
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FINDEX&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX
>>
>>
>> Moderator:David Meyers, Board Member, Society for Improving Diagnosis in
>> Medicine
>>
>> To learn more about SIDM visit:
>>
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Fwww.improvediagnosis.org%2F&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://www.improvediagnosis.org/
>>
>>
>> ______________________________________________________________________
>> Attention:
>> The information in this e-mail message may be confidential, and may also
>> be subject to legal privilege, public interest or legal professional
>> privilege.
>>
>> If you are not the intended recipient, any use, disclosure or copying of
>> this e-mail is unauthorised.
>>
>> If you have received this message in error, please contact the sender.
>>
>> This footnote also confirms that this email message has been checked for
>> the presence of computer viruses.
>>
>> The Royal Victorian Eye and Ear Hospital however does not warrant the
>> message is free of viruses.
>>
>> It is recommended as a prudent business practice the recipient perform a
>> virus scan of any message received.
>> ______________________________________________________________________
>>
>> ------------------------------
>>
>> Address messages to: <IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> To unsubscribe from IMPROVEDX: click the following link:
>>
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FSUBED1%3DIMPROVEDX%26A%3D1&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1
>>
>> or send email to: <IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> Visit the searchable archives or adjust your subscription at:
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FINDEX&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX
>>
>>
>> Moderator:David Meyers, Board Member, Society for Improving Diagnosis in
>> Medicine
>>
>> To learn more about SIDM visit:
>>
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Fwww.improvediagnosis.org%2F&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://www.improvediagnosis.org/
>>
>>
>> ------------------------------
>>
>> Address messages to: <IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> To unsubscribe from IMPROVEDX: click the following link:
>>
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FSUBED1%3DIMPROVEDX%26A%3D1&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?SUBED1=IMPROVEDX&A=1
>>
>> or send email to: <IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG>
>> IMPROVEDX-SIGNOFF-REQUEST at LIST.IMPROVEDIAGNOSIS.ORG
>>
>> Visit the searchable archives or adjust your subscription at:
>> <http://t.sidekickopen13.com/e1t/c/5/f18dQhb0S7lC8dDMPbW2n0x6l2B9nMJW7t5XZs5v_-llVcV_H264zw8vW653gwx56dBNpf9ks1lC02?t=http%3A%2F%2Flist.improvediagnosis.org%2Fscripts%2Fwa-IMPDIAG.exe%3FINDEX&si=5919500165185536&pi=0c808af1-4c80-483f-b574-ba96e200f54f>
>> http://list.improvediagnosis.org/scripts/wa-IMPDIAG.exe?INDEX
>>
>>
>> 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 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 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/
>>
>
>






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


HTML Version:
URL: <../attachments/20150802/f59ab27e/attachment.html> ATTACHMENT:
Name: PhysicianCognitionLogo_email.png Type: image/png Size: 3718 bytes Desc: not available URL: <../attachments/20150802/f59ab27e/attachment.png>


More information about the Test mailing list