Medicine 30 years from now - how will Medical diagnosis evolve?


I'm a portuguese med student just starting in my last year in med school. Today i was watching a show about the evolution of technology.... space travel and all that almost-Sci-fi stuff in Discovery channel. At some point i thought: "How far did technology evolved in medicine?".

Let me rephrase this: when we are face-to-face with a patient what technological tools do we have at hand that can immediately improve our diagnostic accuracy? I'm still siting alone with the patient.... there isn't any machine doing the work for me.... Maybe that's the future? A machine that does the diagnosis for us? Well diagnosis made by us, humans, has always been associated with the risk of error. I don't have to start talking about lawsuits..... Maybe in the future a patient could enter in a room with this diagnose-making-robot that would scan him/her and give the final and correct diagnosis..... 100% error free.....

Well, i think that some day this is what is going to happen. This makes me happy and sad at the same time..... but that's a discussion for other time.....

But this technology is too far away. However, if you see the reality of today you see that physicians are forced to work at a pace that resembles that of a robot..... a Diagnose-making-robot...... And in this ever changing and evolving world of today, what tools have been given to this lhuman-diagnose-making-robiots to help them cope with all the diagnosing demand?


So it crossed my mind.... at can we, med students and physicians, do to help ourselves? To answer this question i created this topic... So we can discuss how to use the technology and the knowledge of today to help us in those moments when we are alone with the patients.

Let me start with some ideas.

- Getting a complete history:

a) It can be difficult by so many reasons.... an annoying patient.... not enough sleep.... not enough coffee.... And it can be specially difficult when we are trying to make fast and potential life-saving decisions in an ER. So why not using WORD ASSOCIATION and VOICE RECOGNITION in the computer interfaces that are used to type in the patient's file?

i.e. An 67 years old male patient comes in to the ER and complains of "shortness of breath". Immediately we think: STEMI, HF, PE, pneumonia, acute exacerbation COPD. As we type or say "shortness of breath" the system would show us questions that we should obligatory ask helping us making the right diagnosis;

- Physical examination

a) Using the example above.. We listen to the the lungs sound but we only find sparse crackles.... But when looking to the patients face we see that he has very pale skin.... We type that and our interface recommends us to look for enlarged lymph nodes and hepatomegaly or slpenomegaly. And in fact the patient has slightly enlarged cervical lymph nodes but that could be easily missed.... This patient had anemia secondary to a chronic lymphocytic leukemia.

We could easily diagnose this patient nowadays in a ER. But the same patient could be elusive in a different setting, for instance if we had mild shortness of breath and was seen by a primary care physician with lesser tools. If there's was a way of helping us making the right questions and helping us thinking outside the box we could easily be better at diagnosing without necessarily spending much more money.

Please reply




For simple cases I think machines would do great. But a lot of the time you present with very similar symptoms so the real differentiator apoears in the patient history - and that is way, way too varied for computers to process at the moment. The suggestion is always to use technology to assist your judgement (like you said, giving a list of suggested diagnosis based on your obs) instead of supplicanting it entirely. I'd like to imagine one day there'll be a machine that maybe could take over the work of a physician, but at that point it'll basically be conscious! Modern AI is no where close...

Though that reminds me, a computer did beat a human in interpreting ecgs, but I'd still like to know whether the two working together would have been more accurate.