Discussion Board

 View Only
Expand all | Collapse all

BMJ review on diagnostic reasoning in cardiology

  • 1.  BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 10:35
      |   view attached

     

    Thanks to John Brush, Geoff Norman and Jon Sherbino for this New Year's present – a "State of the Art" review of diagnostic reasoning in Cardiology, just out this week in the BMJ, and open access to boot:

     

     

    Mark L Graber, MD FACP

    Founder and President Emeritus, SIDM

    Professor Emeritus, Stony Brook University, NY

    Work   607 305-0050

    Cell      919 667-8585

    A close up of a signDescription automatically generated

     

     

     

     

     





  • 2.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 13:30
    Thanks, Mark!  This is great!!!  And the price is right....


    Michael A. Bruno, M.D., M.S., F.A.C.R.  
    Professor of Radiology & Medicine

    Vice Chair for Quality & Patient Safety

    Chief, Division of Emergency Radiology

    Penn State Milton S. Hershey Medical Center
    ( (717) 531-8703  |  6 (717) 531-5737

    * mbruno@pennstatehealth.psu.edu  

    1571679014277





  • 3.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 14:19
    Thanks for sharing this great article. I have a question for the members of this group. The Abstract starts, "Research in cognitive psychology shows that expert clinicians make a medical diagnosis through a two step process of hypothesis generation and hypothesis testing. Experts generate a list of possible diagnoses quickly and intuitively, drawing on previous experience. " 

    In my experience working with many patients (and in my family's personal experience), diagnostic errors often happen because the provider does not have the patient's condition in their experience and thus does not consider it as a hypothesis. This may be because it is a rare disease or because the provider simply hasn't seen it recently or ever. 
    Does the field accept that hypotheses should be based on the provider's experience (including education)?
    Is there no step where the provider does something to consider conditions that do not immediately come to mind?
    Thanks,
    Megan Golden


    Megan Golden, CEO


    Sent via mgolden@mission-cure.org" target="_blank">Superhuman






  • 4.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 15:16
    This is precisely the place for differential diagnosis decision support systems: DXplain, with which I work, Isabel and others.
    No one can (or should be expected to) know the clinical features of every disease in the world.  Yes, most doctors should get a hypothesis when the disease is common and the findings typical, but rare disease do exist, and common diseases may have atypical findings.

    ------------------------------
    Edward Hoffer
    Massachusetts General Hospital
    ------------------------------



  • 5.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 16:53

    The most interesting aspect of this, to me, is exactly this question of how much we can\should learn about 'expert' diagnosis when in reality most diagnosis is tackled by non-experts.  I accept the argument that true experts would receive very little benefit from software that helps with differential diagnosis; they already know the variants and the rare\exotic diseases that might be responsible for a patient's condition.  But for everyone else, decision support seems like an excellent way to mention diagnoses that didn't initially spring to mind.  There are other solutions as well – consults, and second opinions are at the top of the list along with decision support.






  • 6.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-05-2022 17:19
    Mark,
    Happy New Year!  Our experience, when we look at our user data @VisualDx we see that experts also use diagnostic CDS.  I've been a dermatologist for 27 years and I know my brain has limited capacity.  I often blank on rare genodermatoses and the 1 in a million diagnoses I learned about during residency but also tis unusual variants of the common are a van large part of dx error.    
    As an example, Dr Mathis in Lisa Sanders NYT column this past weekend (link below) used point of care information as part of his diagnostic process and he is a specialist. 
    Best
    Art





  • 7.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 12:13
    Is there anyway that the advances taking place in diagnostic Artificial IntelligenceI can be extrapolated into practice to improve diagnosis today?

    Would regional differential diagnoses improve things?

    Robert Bell, M.D.







  • 8.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 12:51
    Hello
    AI may help improve diagnostic accuracy, however I fear it may actually worsen it for some populations. The AI algorithms are only as good as the underlying data they are based on. In many cases this data does not include under represented populations. Therefore they algorithms perform well for the populations of patients they were based on but perform very poorly in the under represented populations.

    Ted E. Palen, PhD, MD
    Colorado Permanente Medical Group
    Sent from my iPhone
    NOTICE TO RECIPIENT: If you are not the intended recipient of this e-mail, you are prohibited from sharing, copying, or otherwise using or disclosing its contents. If you have received this e-mail in error, please notify the sender immediately by reply e-mail and permanently delete this e-mail and any attachments without reading, forwarding or saving them. v.173.295 Thank you.




  • 9.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 14:12
    In radiology, the major equipment vendors (Siemens, GE) are now investing heavily in developing workable AI products to add on to their equipment, as well as commiting to support research to demonstrate the value of these AI tools.  So there is growing momentum in the "AI" space for improving diagnosis at the testing level, at least as regards to image interpretation.  

    The leading national radiologic society, the Radiological Society of North America (RSNA) has launched a journal entirely devoted to Radiology artificial intelligence research.  There are a LOT of research papers!


    All the best,

    Mike

    820 Jorie Blvd., Suite 200 Oak Brook, IL 60523-2251 U.S. & Canada: 1-877-776-2636 Outside U.S. & Canada: 1-630-571-7873
    pubs.rsna.org


    Michael A. Bruno, M.D., M.S., F.A.C.R.  
    Professor of Radiology & Medicine

    Vice Chair for Quality & Patient Safety

    Chief, Division of Emergency Radiology

    Penn State Milton S. Hershey Medical Center
    ( (717) 531-8703  |  6 (717) 531-5737

    * mbruno@pennstatehealth.psu.edu  

    1571679014277





  • 10.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 15:43
    H Michael,

    Thanks, important information.

    Do you think it could be valuable to discuss AI issues here on the Discussion Board. Could it help the developers?

    Rob Bell







  • 11.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 19:05
    Could regional diagnostic differential diagnosis be used to compensate for any inequalities?

    Rob Bell, M.D.




  • 12.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-09-2022 17:55
    It will require that regional data are both recognized and available. The importance of epidemiology in the development of diagnostic databases is key.

    Eric S. Marks, MD
    Professor Emeritus Medicine
    Uniformed Services University

    Sent from my iPhone




  • 13.  RE: BMJ review on diagnostic reasoning in cardiology

    Posted 01-06-2022 12:32
    Congratulations to the three authors for their fine review of diagnostic reasoning in BMJ.
    I shall make a few comments about Bayesian reasoning in diagnosis which is mentioned in their review,
    1. If we look at actual diagnostic reasoning in practice, it does not appear to be Bayesian. In practice, a disease is suspected from a presentation and formulated as a hypothesis without any prior probability attached to it, so that it does not have any prior degree of belief for or against it. The hypothesis is then tested and inferred to be correct with a high degree of confidence if a highly informative test result with likelihood ratio (LR) greater than 10 is observed, regardless of prior probability of disease anywhere in the world where the disease is suspected. For example, pulmonary embolism is diagnosed from positive chest CT angiogram, LR 20; deep vein thrombosis from positive venous ultrasound study, LR 16; covid-19 disease from positive covid-19 antigen test, LR 14 and acute myocardial infarction (MI) from acute ST elevation EKG changes, LR 13 in this manner in practice.
    2. This method of diagnostic reasoning in practice appears to be frequentist, which is the other major method of statistical inference (other than the Bayesian method). In frequentist reasoning, a disease hypothesis is inferred to be correct based on performance of a highly informative test result in inferring a disease accurately with a high frequency in a random series of patients with varying prior probabilities in whom the disease is suspected. For example, it is the performance of the test result, acute ST elevation EKG changes, in inferring acute MI accurately with the high frequency of 86 percent in a random series of patients in whom it is suspected, which leads to acute MI being diagnosed with a high degree of confidence in the high accuracy of this diagnosis in any patient in whom it is suspected.
    3. Frequentist reasoning is employed in practice, I believe, because it leads to highly accurate, experience- based diagnosis of a disease in patients with varying prior probabilities. I have not read about or seen any patient in whom a disease was suspected and a highly informative test result with LR greater than 10 observed, and yet the disease was not diagnosed because its posterior probability was low! Further details about Bayesian and frequentist reasoning are discussed in my attached paper which I posted earlier on Discussion Board.
    4. I have looked at the history of prescription of the Bayesian method for diagnosis (see attached paper) and found it to be prescribed in early 1960s due to its coherence defined in terms of not losing a bet placed on a Bayesian inference (diagnosis). I did not find any mention of diagnostic accuracy, which is the main concern of practicing physicians, in this prescription. There is no reason for us to employ the Bayesian method for diagnosis in practice, I believe, as our main concern is achieving high diagnostic accuracy in patients with varying prior probabilities of a disease and not achieving coherence.
    5. The recognition that the frequentist method, which is employed for diagnosis in practice, is the correct method of diagnosis as far as diagnostic accuracy is concerned, has important implications, I believe, in teaching diagnosis to medical students. For example, we note that prior probability of a disease does not play any significant role in frequentist diagnostic reasoning apart from its role in prioritizing testing of various suspected diseases in a non-urgent diagnostic situation. I t may not therefore be important, I believe, to teach them to estimate prior probability of a disease with a high degree of accuracy during diagnosis.
    Bimal
    Bimal Jain MD
    Salem Hospital/Mass General Brigham
    Salem MA 01970

    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 Mass General Brigham Compliance HelpLine at http://www.massgeneralbrigham.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.

    Please note that this e-mail is not secure (encrypted).  If you do not wish to continue communication over unencrypted e-mail, please notify the sender of this message immediately.  Continuing to send or respond to e-mail after receiving this message means you understand and accept this risk and wish to continue to communicate over unencrypted e-mail.