DXplain logo with stethoscope and keyboard in the background

DXplain for Differential Diagnoses 

What do you get when you combine a medical textbook with a computer science lab with Bayesian logic? You get DXplain, a clinical decision support system that is as impressive as it is incredibly fun and useful for education and generating a differential in those mystery cases.

On the DXplain platform, you input clinical findings, such as wheezing or dry cough, add in some details, and a differential list is generated in real time and changes as you add in more information. Since discovering the tool, I’ve been playing with it and twisting around clinical findings to get the rarest diagnoses to appear at the top. Lifelong learning, right?

DXplain search interface with listed clinical findings
Example of adding clinical findings in DXplain.
view of adding a clinical finding to DXplain
Results change as you add more information.
Differential diagnosis list in DXplain
A differential list is generated with additional information available for each diagnosis.


A bonus is the clinical information and alternative diagnosis list you get if you click on any disease.

You may be thinking, “Cool, but we already have several versions of this diagnostic tool.”

Right, but DXplain stands out with its knowledge base of over 2600 diseases and 5700 clinical findings, including symptoms, signs, lab criteria, radiologic findings, and more. As you input information, the system analyzes the disease/finding relationship based on over 300,000 data points to create a disease probability. What’s impressive is this computer-based tool was created over 30 years ago (80’s ingenuity!) and has evolved just like a fine wine. Recently, I asked a sub-I to generate a differential diagnosis for a toddler with a fever of unknown origin and then had her generate a list with DXplain, and we compared the two – this is what I call lively midnight teaching!

Check out this demo video to see DXplain in action:

Ready to give it a try? Head over to Lane Library’s clinical tools, toy around with DXplain, have your trainees generate differentials with it, and/or use it on your next mystery case! To use DXplain, create an account with your Stanford email address. If you have any access issues, please contact Lane Library.

With that, I’m off to input a different set of symptoms to see if I get a rare tropical disease to show up at the top.

References

  1. The laboratory of computer science | dxplain. The Laboratory of Computer Science. Accessed June 26, 2024. http://www.mghlcs.org/projects/dxplain
  2. Barnett GO. Dxplain: an evolving diagnostic decision-support system. JAMA. 1987;258(1):67. doi:10.1001/jama.1987.03400010071030

Leave a comment