Computer-Assisted Diagnosis


I think that better systems could be developed to make and review a diagnosis.  It is human nature to hang onto an original diagnosis and not to continually review it in the light of a gathering list of symptoms and effects.  It is more likely that the symptoms will be made to fit the original diagnosis.  A doctor who is handling 30 patients a day can hardly be expected to remember the previous symptoms and results of someone who they review 3 weeks later.  They could look at the patient’s notes but in my case they are about 4 inches thick, a task that would take a few hours each time.  A simple program could be written that had the patient's history, results of tests, and as the symptoms were continually updated it could re-evaluate the possibilities. The aim is to assist the doctors by providing the relevant information and the diagnostic possibilities. Possibilities would not drop off the radar as they did in my case.

 

For example:

 

Symptoms Blood occurring in bowel movements at random intervals.
Possibilities Bowel cancer 60%; hemorrhoids 38%; something else 2%.
   
Test Colonoscopy – No abnormal areas noticed in bowel.
Possibilities Hemorrhoids 90%; Bowel cancer 8%; something else 2%.
Symptoms Chest pains, bowel movements - occasional blood, previous DVT.
   
Test D-dimers - high.  ECG - normal.
Possibilities PE 90%; Bowel Cancer 7%; something else 2%; ECG error 1%.
   
Test VQ scan – shows blood not reaching all parts of lungs.
Possibilities PE 90%; Bowel cancer 8%; something else 2%; ECG error 0%.
   
Symptoms Increasing pain 1 week after PE treatment
Possibilities PE 85%; bowel cancer 13%; something else 2%; ECG error 0%.
   
Symptoms Continuing pain, increasing rectal bleeding
Possibilities PE 75%; bowel cancer 23%; something else 2%; ECG error 0%.

 

The point of this is that other possible diagnoses are not allowed to drop off the radar, they just vary in strength of probability.  Consultants or doctors who at various junctures take up a patient's case don’t just have the carried forward diagnosis to consider.  They could find that as various symptoms had been added through time that the probability of the other diagnosis had increased.

A way this idea could be implemented is if patients had a memory device (USB stick, SD card etc) and when a doctor came to review a patient they took the device and plugged it into their handheld PC. Previous history would all be available in a collated, easy-to-access form, together with their current symptoms, tests carried out and all the possible diagnoses together with a rating on their probability.

 

Cloned by dolly@sundown.me.uk