The 2026 Drupal Business Survey is open. Your response will help shape the future of the Drupal business ecosystem.
By samcortex on
Hi:
I am a physician who has played with Coldfusion, ASP and PHP some time ago but I am not a programmer. I installed Drupal last year and I felt it's great but when I saw Lullabot and its products I felt it can be a solution for my idea.
I need your advice and opinion about this problem:
Is it possible to make a Clinical Decision Support System with drupal? I know that AI is a great part of any Clinical Decision Support System (CDSS) but making a small CDSS with drupal is my idea. Is it possible. Did anyone try it? Does it need advanced skills or just using CCKs and Views (learning with Lullabot) will give me enough skill to do it.
Thanks
SamCortex
Comments
If...
...I knew what a CDSS was maybe I could help :-)
Maybe you could give us an example of what you mean by that?
Let me guess:
Someone comes to your surgery with a high temperature, you want a list of diagnoses?
Views, CCK and node reference spring to mind.
This scares me.
Let me put it a different way:
I'm a really good programmer. Can I do surgery without going to medical school?
I mean that to be a little inflammatory!
While Drupal is very flexible, and could with only modest programming, I suspect, be made to do something like what you are thinking of, here is the concern:
People believe computers. They believe stuff that is on the screen. That means if something is not on the screen, they will tend to put it out of their minds. The risk in your case, doc, is that if you have a glitch (yes mistakes are made even without programming) that leaves out something like a differential diagnostic point, or something, the user may be accidentally guided to overlook that thing as well.
Thus, any kind of branching, etc, is a risk.
If all you are seeking to do is create a searchable reference, then maybe fine, but if it's some sort of "does the patient have fever (yes/no)" "is the patient speaking in tongues(yes/no) "has exorcism been tried?)" logic, then I suspect that, as we do with medicine, you are well advised to associate someone with some solid background.
My humble suggestion is to take it as far as you can (and that will be pretty far). see if you like it, but spend some time with a PHP-smart person to talk about how the core function will work.
But yes, to above, CCK and Views, maybe node reference too.
:)
No you can not do surgery
No you can not do surgery without going to med school.
These apps just help us, nothing else. The final decision is always the doctor's and these apps just give us more options and more insight. That's it. They are not perfect and the human factor will not be eliminated.
You are right that we use logic and it is critical in medicine and anything about it. But isn't it possible to get such a logic from Drupal. I really like you to see http://dxplain.org/demo2/frame.htm
It shows you what CDSSs are about by example.
Here is the part that makes a CDSS a real helper.
http://dxplain.org/demo2/sld004.htm
Yes. Basically it's true.
Yes. Basically it's true. It's an application that helps physicians to make an accurate diagnose. Sometimes such a thing can be literally life saving. This may help:
http://en.wikipedia.org/wiki/Clinical_decision_support_system
http://www.openclinical.org/aisinpracticeDSS.html
َAnd here is the exact definition:
http://www.openclinical.org/dss.html
Thanks for your help.
Yep
Right to be scared!
But why not?
Sounds like a good idea to me!
This is interesting
If a php based CMS could do this Drupal is the best one. You may like to get in touch with Acquia if its a money-supported project. If its a hobby based project you could give a try yourself.
You can make a knowledge-based CDSS but not AI based CDSS, that is a CDSS that "learns" with time.
You need to input a whole lot of clinical data with custom fields for which CCK module can help, and then you can use Views to query,filter and "get answer". And Taxonomy. This is feasible to a certain extent and probably can be extended by apt and judicious input of data in first place and extending that with many other modules, may be custom ones too.
You can of course give it a try, for example, left iliac fossa pain, disturbed bowel, but no malaena, age 50+ ---> possible diagnoses which can be further filtered out according to various filters.
Nothing is going to harm if you give a try, you never know what you can do unless you embark upon. I bet this will be interesting.
Ah ha
You say:
"Nothing is going to harm if you give a try, you never know what you can do unless you embark upon. I bet this will be interesting."
What about repetitive strain injury......?
None the less, this has to be embarked upon!
It's just an idea. Nobody
It's just an idea. Nobody supports it but me:D
There are some commercial versions but the problem is that they are expensive and some are abandoned. Plus it's nice to make a more personal CDSS.
I will spend some time to learn Drupal. It seems I have lots to learn.
We support you. Go ahead.
We support you. Go ahead. At least a primitive version can be made.
For a beginning, from a set of symptoms it can bring out a set of differential diagnoses which can then be pinpointed further.
Thanks! I don't know what to
Thanks!
I don't know what to say. It's great to see a community with such a support!
I hope I can do something that gives a better service to patients. As a matter of fact the important thing is that it leads to a point that I make less mistakes and maybe my colleagues too. You know what, everyday diagnosis is not very hard, many patients have classic signs and symptoms and after a while it's not very hard to instantly see the diagnosis(or at least you see two or three which will be narrowed to one with lab tests or imaging). Yes, it is a very narrow line but all cases are not what you see in House MD tv series:D
But sometimes there is a patient which comes with a symptom that doesn't fit with the signs or some lab tests and their repeats don't match the patients status. Then you feel, this puzzle needs to be solved. Then you make a list of probable diseases and try to eliminate then one by one. But rarely, everything doesn't go smoothly. Your brain tries to see through the puzzle pieces and get the picture but you need some fresh eyes and something that looks from outside of your head. A colleague is great(if has the time)... a diagnostic book is great(but it takes time to search in a book) but if a software can suggest more choices instantly without any pressure or something like that is a great relief in my opinion. Such a thing can be great even for everyday cases but the real use will remain for more complicated cases. Plus it can help me keep a neat track of every patient. It can be very educational because I may be able to write the final diagnosis for the patient (which most of the time eventually will come out) and learn a lesson.
I wrote the above to give a clue about my situation, my ideas and why I like to do it. But it always looked so far away because of my limited computer skills and my limited tome. But I hope with Drupal CMS and with the help of Drupal community I can do it.
That's a fantastic approach.
That's a fantastic approach.
The main problem or the main "trick" lies in the fact that as much raw data needs to be put as possible. The greater the knowledge database is greater can be the problem solving skill of the software.
Who puts in the data, what will be the sources of the data, what are the fields to create can be challenging but not unsolvable. Drupal has an inbuilt wiki so a collaborative approach over the net can be possible. I guess such approaches must have been made in some form or other, either web based or standalone programs. Yet there lies a huge field of work which can be started by an individual and later expanded to a group and perhaps, the population of the web.
For example, epiploic appendagitis - it can be misdiagnosed as acute appendicitis and unnecessary surgery may happen, as most often this condition does not come to mind. Software can suggest this and CT images of epiploic appendagitis stored in image database can be compared to the current CT, if a CT scan is done. Possibilities are immense!
Very well put!
Very well put!
Ok that said, had a good look at http://dxplain.org/demo2/frame.htm
Two big ifs:
1 Where does the data come from?
2 Can Drupal do a better job than http://dxplain.org/demo2/frame.htm ?
Don't have the answer to question 1.........
Some possible answers
Answer to question 2, imho, is as follows:
Whether Drupal can do a better job or not depends on how Drupal is used. There was one CMS does not mean another did not succeed, there was one social network does not mean another did not succeed. At a glance, design-wise, dxplain has a clunky feel. So yes Drupal can do perhaps better.
There must be more such there and I remember a non net based similar something years ago. Which means there is always room for one more. Which can evolve to be better, at least potentially.
Answer to 1 may be best given by OP but I think as always the best place to start is the standard textbooks like Harrison's Principles of Internal Medicine and Oxford's. Both of these have interactive CDs but not with the charming outlook of Views/CCK/Taxonomy/Book(wiki).
Then enrich it with web acquired data and collaborative data from like minded people.
Just my scanty thoughts.
Dxplain is commercial and it
Dxplain is commercial and it seems MGH sells it just to hospitals and medical organizations and not individuals (I didn't check it if they do, just read the website and it says MGH sells the license to organizations, price: 1000$ for 6 months membership). Where does data come from? I believe Dxplain and other CDSSs all use textbooks.
I will add some data and my
I will add some data and my colleagues may join me soon after they realize the process is beneficial for them. There are some dependable online resources but I am not sure how drupal can use them. An example is http://www.medscape.com which is reliable. Another example is UpToDate.
About primary epiploic appendagitis which as you said can be diagnosed using CT but there are more than one differential diagnosis and this is exactly what I mean. Giving more options to think about and a list of tests which can narrow the diagnosis. It's great that you know the disease and the approach :) thanks it gives me lots of hope! But first of all I should learn how to use drupal. Will start next week and hope to learn it quickly.
ok RE the data
Having looked at http://www.medscape.com I see at the bottom of the page, the dreaded "All material on this website is protected by copyright, Copyright © 1994-2009 by Medscape"
The data is the only thing stopping rapid progress with this project, oops did I say project!
So much for Plan A..........
Anyway, Plan B would be as follows:
1 Stop the talking and start the doing!
2 Create hosted development site (I'll do that, ready for Monday)
3 Proper analysis of DXplain web site (looks like CCK, Views, nodereference, voting API, filefield for the CT scan data and some php trickery. will do the job just fine to me!)
4 Get the basic functionality and navigation presentable using data from OP and others
5 With a good working demo and backing from the Drupal community, we should be able to convince a data supplier of worth to share their info with us. (don't worry about data import, easy!)
So guys, if you want to grab the bull by the horns and run with this, I'm up for it!
Knowing my luck "grabbing a bull by the horns" will have a completely different meaning in both Farsi and Hindi!
Assuming the grabbing of horns is not received as an insult, hit my contact link and we do this thing!
Regards,
Ludo.....
Thanks for the reply. I am up
Thanks for the reply.
I am up for it.
Does "Create hosted development site" part mean that it will be a public project? I really like it to be such a thing. I really like open source, free and public services for everybody. As I said before this idea is living in my head for quite some time now but now I have a little extra time to spend on it. I really appreciate any help and support I can get. Okay lets do it:)
...
Sam, have you come up with a plan for how you might actually build this?
I'm a pretty experienced Drupal site builder and my thoughts turn instantly to Taxonomy, Apache Solr and probably something like Open Calais (to pull in external resources).
I have been asked to do something vaguely similar but with psychometric evaluation. I'm currently evaluating either building a custom search module or using Views exposed filters for the input. Two very different approaches, and for us perhaps the Apache Solr + Custom search fields is going to win - mainly because Apache Solr is such a good search engine and can generate facets.
I would be thinking seriously about how you are going to get the data into nodes - surely importing it is the only way to go, sucking it in from somewhere (the sheer volume needed is huge) and you would want it to self-update to keep up to date.
I'd be most interested in talking to you about this project, drop me a line if you feel so inclined.
Pimp your Drupal 8 Toolbar - make it badass.
Adaptivetheme - theming system for people who don't code.
Ok Sam
Glad we got a reply from Jeff, he's heavy weight!
My thought was to set up a test server, start with some data and establish a framework/design to build upon. The early stage would include defining the Taxonomy, once again Jeff has pointed that out, as that would drive the system.
Then, address the issue of data import as Jeff has also pointed out, and take it from there.
Just drop me a line and we will get started. And I would suggest you contact Jeff too.
As for it being an open project, yeah access for all!
Regards,
Ludo...
Sounds interesting. What's
Sounds interesting. What's the latest on this?
Ask him! I set up the
Ask him!
I set up the devolopment site and did the groundwork FOC, he disappeared!
Granny was right "Never trust an....."!!
Great idea - usability?
I would be really interested in this project - was there any proceeding? Any possibility to try out what is already done? The usability will be a major aspect...