3

# Webinar Series — Reasoning Under Uncertainty (Part 1): Differential Diagnosis of Diseases

### Webinar Overview

With the outbreak of the COVID-19 pandemic, reasoning about diseases has gone mainstream. No longer is it just healthcare professionals that perform differential diagnoses. Newspapers and social media have been publicizing charts that compare symptoms of COVID-19, the "regular" flu, and the common cold so individuals can potentially self-diagnose and reduce the burden on healthcare providers.

While a chart can list symptoms, it is not an "inference engine." Deliberate reasoning still has to happen in the mind of the self-diagnosing individual to reach a conclusion. That turns out to be the difficult part, as humans are ill-equipped to handle probabilistic inference from effect back to the cause, i.e., from symptom to disease.

In this webinar, we present Bayesian networks as a framework for encoding knowledge about diseases and symptoms. Given this knowledge base, we then use BayesiaLab's inference algorithms to update the probabilities of the potential conditions given the observed symptoms. A very similar model, the so-called "Visit Asia" network, was one of the earliest examples that illustrated the reasoning capabilities of Bayesian networks.

Please note that this webinar does not constitute medical advice. Although the example is based on current events, we focus solely on the reasoning process. Thus, all numerical values and probabilities shown in the presentation should be considered fictional.

### Event Update

Please also see the next installment of this webinar series on reasoning under uncertainty on March 26: Differential Diagnosis of COVID-19 and Influenza-Like Conditions.

5replies Oldest first
• Oldest first
• Popular
• Great presentation! Slides 17 and 18 show that p(Infected | Test = positive) = 50%, where sensitivity and specificity = 99.9%. Assuming same parameters for COVID19 tests, doesn't this mean that each "confirmed" COVID19 "case" is equally likely to be "COVID19" and "not COVID19"?

Along similar lines, a Medium.com article* cites a WHO study on China** to the effect that "if you come in contact with someone who tests positive for COVID-19 you have a 1–5% chance of catching it as well." Might this low perceived transmission rate be partly a function of false positives?

Like
• Kurt Schulzke In this hypothetical scenario, I described universal testing in a population. For instance, one time when I was in elementary school in the 1970s, a bus with a mobile x-ray station pulled up, and all the students were tested for tuberculosis.

The current testing practice for COVID-19 is obviously different. So, if you have symptoms like coughing, fever, etc., your baseline probability will be much higher, and the test will be more informative. So, try changing the marginal probability (prevalence) of the diseases from 0.1% to 1%. Then, a positive test result will move the probability of the disease to 91%.

Like

Like
• Stefan Conrady Speaking of prevalence, check out this Foxnews article:

The Times quoted a front-line healthcare provider as saying that county doctors were interpreting Thursday’s advisory to mean they should only test patients who are going to be hospitalized or have something unique about the way they contracted the virus.

They are not planning to test patients who have the symptoms but are otherwise healthy enough to be sent home to self-quarantine — meaning they may never show up in official tallies of people who tested positive.

Maybe they don't have COVID (not that prevalent) or maybe COVID just ain't that dangerous (except for a small demographic), after all? See my attached mortality model.

Like 1
• Due to current events, we updated the program of our next webinar on March 26: Differential Diagnosis of COVID-19 and Influenza-Like Conditions.

Like
Like3 Follow
• 3 Likes
• 8 mths agoLast active
• 5Replies
• 422Views
• 4 Following