Home

Postscript 17 February 2009 - It is looking good that a system will be built and be online (and offline) before 2009/06. Funding would however help this to happen - the total amount pledged so far is a cool zero.... We have some interest from national and international organisations here in Harare Zimbabwe, but even they have not yet put any money on the table...

Postscript 30 November 2008 - although this was written some 9 years ago, it is still completely relevant, if not more so, and - amazingly or not - nothing has happened in the last 9 years to bring this concept closer to reality. What does this say about MSF, UN, WHO etc etc?? - and the only correspondence I can remember having on this subject has been with an Australian school girl some 5 or so years ago - she was doing a special project on this subject... - OK - so now is the time to get something going chaps!....

MEDICAL EXPERT SYSTEM -


1. Poverty in 3rd world is increasing and quality and availability and affordability of clinics, gp and hospital care is getting rapidly worse. and as we know 95% of cases are 5% of problems (some of these 5% are minor and easily treatable, others are life threatening or even incurable)

2. I checked out recently on the web with search thru http://www.altavista.com for medical expert systems - I find almost nothing, nothing which is open and free to users, and most of which is oriented to USA medicine. I find nothing for 3rd world. Amazingly there is nothing on the www.who.org site about expert systems... (UN WHO - World Health Organisation).

3. I am looking at something simple for malaria, diarrhoeia, cholera, cerebral malaria, etc etc etc... will have a menu structure, can or will have some photos or sketches when required, will prescribe drugs or non-drug treatment (e.g. home made saline solution) where suitable.

4. The system will be sited (free) on the internet and open to all countries. can also be sited locally on a pc hard disc to run locally to save connect time and also for cases where no internet or web connection (i.,e. most cases) but where computer and maybe electricty exist...

5. So - we need doctors and pharmacists to create the material - simple text will do quite often - and i can work that into web pages with connections / links etc...

6. I guess we start with an outline - flow chart etc etc...

7. Any volunteers? - maybe circulate this email / webpage to a bunch of people you each know who could contribute either at framework or detail level.

System will at start be in english then we will commission translations to french, portuguese , arabic , spanish, swahili, ndebele, shona, etc..

8. We will have big disclaimer notices all over the system and in many cases the system will diagnose - GO TO A DOCTOR!

9. The expert system can be used by individuals, employing companies, clinics, pharmacies, paramedics and even by trainee and experienced doctors.

10. The product will be dynamic, with expansion and correction probably on a monthly basis.

11. Maybe such a system exists already and I just could not find it - but I dont think so.

Alex Weir 27 December 1999


Postscript 26 December 2008

Some idea of how MedExpSys or MES or TWoMES will appear and how it will work....

12. The interface as per http://easydiagnosis.com/ seems to have some merits. Each subject area will have a single page with maybe 10-20 questions. The answers will be a pull-down list, of which usually only one option can be chosen. At the end of the answers there is a submit button, and then a diagnosis page will be displayed - with a number of options and their probability, based on the answers to the questions. Each of these diagnoses will have one or several local and/or internet links to text content on the diagnosis. Sometimes also one of these local links may go to another single page with 10-20 drop down lists of questions.

13. The alternative to that scenario would have been a more complex menu structure which is finely tailored to each answer. The disadvantage of (12) is that some of the questions will not be relevant, but the simplicity of menu design I think has significant advantages.

14. The main menu page will have a list of these pages as outlined in (12) above. Suitable topics for these item (12) type pages will include:

Diarrhoeia
Vomiting
Chest Pain
Bladder Pain
Headache
Fever
Fainting
Stomach Pain
Kidney or Liver Pain
Gall Bladder Stones
Colds and Influenza
Other Pains
Sore which does not heal
Swelling
Child cries continuously
Pregnancy Problems
Gynacological Problems
Post-Birth Problems
Blood in Urine
Cholera
Dysentry
Worms
Mango Fly
Dengue Fever
Sexually Transmitted Diseases
Crabs
Lice
Malaria
AIDS
General Weakness
Lack of Appetite
Excessive Appetite
Psychiatric Problems
Erectile Disfunction & other sexual problems

15. There should be some attention to country, geographical location, and time - i.e. there are epidemics of various diseases in all societies, and the system should be able to accept knowledge about such epidemics and respond accordingly.

16. XML files would seem an ideal way of accepting data. Storage can be in conventional RDBMS or in XML files.

17. Systems must be flexible to accept local input by local experts and even by the local offline system attendant or manager, but without enabling malicious input. One way to do some of that would be to have xml files which also have an encrypted checksum content - if any of the human- and machine-readable content has been changed then the checksum no longer matches, and the screen gives a warning that the xml file being used is not as per original and cannot therefore be 100% trusted. Local doctors and others producing their own xml files and variations of official xml files can submit to the central system admin people to examine, proof and approve and to have their xml file(s) issued with a checksum and resent back to them... That will also ensure that locally-produced xml files get wider circulation...

18. There should also be links on the (12) output page to drug treatments

19. It will make a lot of sense to allow people to dynamically add a page at a time - the processes will be:

a. make the multiple choice selection boxes (the usual header input (country/region, race, age, sex, bodyweight, income band, existing medication, tobacco, alcohol, drugs) - will exist already as standard)

b. create the code which makes the decisions on output - e.g. if Q1.R=2 and Q2.R>4 and Q5=R1 then display sheet 1; if Q2.R=3 and Q3.R>2 then display sheet 2; etc.. If question 1 response=2 and question 2 response > 4 and question 5 response = 1 then ......

c. create the response sheets (disease/condition 1 88% probable; disease/condition 2 10% probable; disease/condition 3 2% probable;

Part (b) will be the difficult part, and it might not be able to run the prepared minisystem immediately - it may require batch operation by the system operator. Also the files, database entries etc created might be directly portable to the offline version, or they may require some manual reworking...

This way, an empty or almost empty (and possibly primitive in content and diagnostic terms) system can be made available at go-live, with the expectation that real doctors in the third world will create their own material which is immediately or very quickly made live and available to the global public.

20. The diagnosis webpages will also have links to the description and treatment of the disease/condition - and these will normally be from online medical encyclopedias etc.. Each condition will link to a number of such alternative description and treatment sources. Offline systems will have as many of these as possible, but normally online systems will have more options.... Dont forget also that offline systems may sometimes have online capability, therefore offline systems should offer links to both offline and online material. The best route is from diagnosis sheet to disease/condition masterpage and then to the offline and online materials...

I will add more content here as ideas gestate.....

Where there is no Doctor
MSF Diagnostic Manual
www.hon.ch/MedHunt
www.selfhelpguide.nhs.uk
www.patient.co.uk
www.webmd.com

Email: alexweir1949@gmail.com