E. Coiera, When Conversation is better than computation, Journal American Medical Informatics Association, 7,277-286, 2000. PDF here
(Extremely influential recent article on the role of computers and decision support in health care.)E. Coiera, Mediated Agent Interaction, in Quaglini, Barahona and Adreassen (Eds). Proceedings of the 8th Conference on Artificial Intelligence in Medicine Europe, AIME 2001, Cascais, Portugal, 1-15. Springer Lecture Notes in Artificial Intelligence No. 2101, Berlin (2001). PDF here
(The more technical follow up. Likely tobe seminal for introducing the notion of "grounding cost")Rector A (2001). AIM: A personal view of where I have been and where we might be going. Artificial Intelligence in Medicine 23: 111-127. PDF here
(A broad overview of what I think matters, plus a large bibliography.)
Handbook of Medical Informatics www.mieur.nl/mihandbook/r_3_3/handbook/ Also about as close as this course comes to having a text book. The book, Handbook of Medical Informatics, ed. by J.H. van Bemmel and M.A. Musen, 1997 can be purchased through Amazon or is in the ISBE library. The content is variable, but the web site is growing.
E. Coiera. Medical Informatics, the Internet and Telemedicine (An excellent overview of health informatics generally if you haven't seen it. Much of it is on line. The links under 'Additional Information' are very helpful) http://www.coiera.com/Stanford Medical Informatics www.smi.stanford.edu A rich source from which not a little of the course has been borrowed. (The site for the more or less parallel course is www.smi.stanford.edu/courses/bmi210a) The Technical Reports section covers virtually all Stanford papers less a few which are copyright elsewhere.
OpenClinical www.openclinical.org - John Fox's site at ICRF with links to many demos and things to try
MIT Clinical Decision Making medg.lcs.mit.edu (Includes links to Guardian Angels www.ga.org. One of the very innovative serious computer science departments, perhaps the only one besides stanford now that Carnegie Melon is less active.
The Manchester Medical Informatics Group - our own site www.cs.man.ac.uk/mig. Jeremy Rogers maintains a particularly useful collection of links and resources. (Follow the links to people)
JAMIA - Journal of the American Medical Informatics Association. www.jamia.org. One of the main journals in the field and the only one with full content on line, although you have to be a subscriber to get to it. The University should have an institutional subscription. Otherwise, speak to me.
University of Freiburg's catalogue site: http://www.hiww.org/
Books: Degoulet P and Fieschi M (1996). Introduction to Clinical Informatics. New York, Berlin, Springer. is a surprisingly good short introduction. The really good textbook on Medical Informatics has yet to be written.
MIG - us - http://www.cs.man.ac.uk/mig , http://www.bhif.man.ac.uk, http://www.clinical-escience.org and http://www.opengalen.orgCHIME - UCL site under David Ingram. Primarily responsible for GEHR/ OpenEHR http://www.chime.ucl.ac.uk/
SCHIN - Sowerby Centre at Newcastle, primarily responsible for the Prod igy project http://www.schin.ncl.ac.uk/
(Ian Purves)ICRF - Advanced Computer Laboratory at ICRF. John Fox's unit. Responsible for ProForma, research on safety critical systems, http://www.acl.icnet.uk/lab/index.html also home for OpenClinical
To Err is Human: Report on Clinical Errors by Institute of Medicine http://www.nap.edu/books/0309068371/html/
It is also worth following up the editorials and commentary in BMJ at the time if you can find them
.
A spoonful of Sugar. Audit Commission report on medication errors in UK (Dec 2001). Abstract and full text at http://www.audit-commission.gov.uk/publications/spoonfulsugar.shtml (You have to hunt a bit to find the full text but it is there.) Strong call for electronic support for prescribing as one of the solutions.Two books linked to Marc Berg on "rationalising medicine". Berg M (1997). Rationalizing Medical Work: Decision support techniques and medical practice. Cambridge MA, MIT Press.
Berg M and Mol A, Eds. (1998). Differences in Medicine. Durham, NC, London England, Duke University Press.
Original Pediatrics Guideline in pdf format (60pp) http://www.pediatrics.org/cgi/reprint/103/4/e54.pdf
NELH http://www.nelh.nhs.uk/ (Also provides good links to Cochrane Databases with Athens password)National Institute for Clinical Excellence (NICE) http://www.nice.org.uk/
The Cochrane Collaboration www.update-software.com/cochrane/ - Abstracts on line. Full text by subscribing to CD only. I presume available through JRL.
BMJ Clinical Evidence. www.clinicalevidence.com - someone needs to check on the licensing for U Manchester. I presume there is an institutional license but can't confirm it now. And of course BMJ's own site www.bmj.com - in most ways the most innovative use of health informatics in the UK.The BNF at www.bnf.org has some material but the most valuable is probably a list of guideline providers at www.bnf.org/LinksFrame.htm - Given this comprehensive portal, I see no reason to duplicate the others here.
Prodigy - the major UK based decision support project for prescribing for chronic diseases in primary care http://www.prodigy.nhs.uk/PE Johnson's work, in paper in the library, is still the best in the field of understanding diagnostic processes and has never really been equalled.
Feltovich PJ, Johnson PE, Moller JH, Swanson DB. The role and development of
medical knowledge in diagnostic expertise. In: Clancey WJ, Shortliffe EH, eds. Readings in medical artificial intelligence: The first decade. Reading, MA: Addison-Wesley, 1984:275-319.Johnson P (1983). What kind of expert should a system be? Journal of Medicine and Philosophy 8: 77-97.
Johnson P, Duran A, Hassebrock F, Moller J, Pietula M, Feltovich P and Swanson D (1983). Expertise and error in diagnostic reasoning. Cognitive Science 3: 91-118.
Johnson P, Hassebrock F, Duran A and Moller J (1983). Multimethod study of clinical judgement. Organisational Behaviour and Human Performance 30: 201-230.
(Good summary and links from OpenClinical at http://www.openclinical.org/methodssummaries.html)
GEM http://ycmi.med.yale.edu/GEM/ - a proto standard for semi-structured guidelines expressed in XML. Not executable but very helpful in understanding the process of converting typical text guidelines into a structured text form suitable for delivery by computers for interpretation by doctors or implementation in computable form.Medical Logic Modules (MLMs, also known as Arden Syntax). This is the syntax that grew out of the original HELP system in Salt Lake City. "Simple" first generation trigger and action. Famous for the "curly bracket problem" - all terminology and links to medical record expressed as text inside {...}s http://www.cpmc.columbia.edu/arden/
GLIF - the output of the Intermed project between Stanford, Columbia and Harvard that tried to harmonise Stanford's Protege/Eon approach with MLMs. Only partially successful and now carrying on in as GLEAM - see HL7 Guidelines TC below. http://www.glif.org
Protege/Eon - the main Protege web site for papers, downloads, tutorials, etc. http://protege.stanford.edu/index.shtml and its companion for Eon/Dharma http://smi-web.stanford.edu/projects/eon/
PROforma - a 'complete' approach to clinical guideline authoring and exception, developed at the ICRF http://www.acl.icnet.uk/lab/proforma.html. A very clear graphical approach set over a rigorous logical engine and John Fox's specific approach to "argumentation" and emphasizing safety critical features. Also the basis of a book, Safe and Sound by J Fox and S Das. The tools and execution engine (Abrezzo) are now being commercialised by Infermed ( http://www.infermed.com) We hope to have a version of the tools available for the course.
ASBRU - perhaps the most ambitious programme for a 'complete' computable guideline and protocol language - output of the Asgaard programme. The main effort is now split between U Vienna and ben Gurion although others are still involved. http://www.ifs.tuwien.ac.at/asgaard/ also the older Stanford site http://smi-web.stanford.edu/projects/asgaard/
DXplain - the most widely used diagnostic suggestion from literature method. Available to medics on line at www.lcs.mgh.harvard.edu/dxplain.htm
HL7 Decision Support Technical Committee - available through www.hl7.orgbut sadly no documents are on the web as of Jan 2002
The CO-ODE web page which includes the tutorial and links to most of the tutorial material http://www.co-ode.org
The Protege-OWL site - http://protege.stanford.edu and follow plugins-->backends-->OWL. Or just download the complete version.
Both CO-ODE and Protege have mailing lists. CO-ODE also has a web based forum. We encourage you to join in.
The W3C Semantic Web Best Practice and Deployment Working Group (SWBP) http://www.w3.org/2001/sw/BestPractices/ and the official website for OWL http://www.w3.org/2001/sw/WebOnt/ where you can find all the official papers and definitions.
Unified Medical Language System (UMLS) www.nlm.nih.gov/research/umls/umlsmain.html (beware, it has moved recently. This is the outer home page, but the part you really want is in the Knowledge Sources Server which gives you access to OntoSaurus and other facilities. http://umlsks3.nlm.nih.gov/KSS/ (no 'WWW'). You need a password which will be given out on paper. We also have the resources mounted in the Department for heavy duty use, but for exploration the net interface is fine.The OpenGALEN - our large project on medical ontologies and a source for our tools. http://www.opengalen.org Where most of our work is available along with papers and documentation
SNOMED - the US site for the combined SNOMED-RT/SNOMED-CT projects http://www.snomed.org - usually, but not always a better source of information than the UK web site: http://www.nhsia.nhs.uk/terms/
The Digital Anatomist project - a comprehensive and very principled approach to structural anatomy with lots to browse and pretty demos as well as probably the definitive version of human anatomy in symbolic form. Also available in the department in PROTEGE compatible form (probably, if we get it up). http://sig.biostr.washington.edu/projects/da/. The forms based interface (http://sig.biostr.washington.edu/~brinkley/cgi-bin/fms-term-viewer.pl?term=Anatomical%20Entity) is relatively quick but has poor searching facilities. The Node based interface is prettier, but slower, but does have a good string match. (http://tela.biostr.washington.edu/~kmayes/SplitInterface/Page.html) I usually find I need both.
Rector A L (1999). Clinical Terminology: Why is it so hard? Methods of Information in Medicine 38: 239-252. PDF here
Cimino, J (1998) Desiderata for controlled medical vocabularies in the twenty-first century. Methods of Information in Medicine37(4-5): pp. 394-403. PDF here
Rector, AL. OILed tutorial for Biomedicine. To be distributed
The semantic web page www.semanticweb.org/
The MyGrid homepage: www.mygrid.co.uk
Cycorp home page www.cyc.com/ and the OpenCyc web page www.opencyc.org/
The OpenEHR site with the design principles, demo software, etc. www.gehr.org/openEHR/HL7 - www.hl7.org - the major standards group for Medical Standards which is rapidly expanding.
The HL7 RIM is one of the major teaching items for the course. HL7 RIM can be found by following the links under Resources Data Models under Current Model Archives HL7 Reference Model. The other link to HL7 Meta-Model (Methodology & Modeling) is also of interest.) These links apear now to be open to the public and not protected by membership passwords.
HL7 maintains slinks to all the other major standards organisations so I won't duplicate that here. (Unfortunately some things can only be downloaded if you are a member. It has also been in the running for the world's most user hostile and slowest website. There is a concerted effort to improve it, but if you have problems, let me know. Or try going via HL7-UK which has tried to tame some of the most awkward features, sometimes successfully.)
HL7-UK - www.hl7.org.uk - The UK subsidiary with information on the UK group and various aids t
The Hugin web site www.hugin.com the Hugin-Lite package used in the course can be downloaded from
http://www.hugin.com/Products_Services/Products/Demo/A convenient summary site is at www.cs.ualberta.ca/~greiner/bn.html. (I wouldn't try to improve on this so go there and look.
Russel and Norvig's AI textbook has a good manageable section on Belief Nets.
Recent comprehensive but expensive textbook. R.G. Cowell, A.P. Dawid, S.L. Lauritzen and D.J. Spiegelhalter, Probabilistic Networks and Expert Systems, Springer-Verlag, 1999. (This is the team that made belief nets practical. The original Lauritzen and Spiegelhalter paper at the Royal Statistical Society was a landmark and presented the first tractable algorithms - orders of magnitude improvement in efficiency. However, they are mathematicians, so it is not an easy read. )
Pathfinder papers - the best known commercial system in medicine, although now a decade old
Heckerman DE, Horvitz EJ, Nathwani BN. Toward normative expert systems. I. The Pathfinder project. Methods Inf Med 31 (2): 90-105, 1992.Heckerman DE, Nathwani BN. Toward normative expert systems. II. Probability-based representations for efficient knowledge acquisition and inference. Methods Inf Med 31 (2): 106-16, 1992.
Heckerman DE, Nathwani BN. An evaluation of the diagnostic accuracy of Pathfinder. Comput Biomed Res 25 (1): 56-74, Feb 1992
Ledley RS, Lusted LB. Reasoning foundations of medical diagnosis: symbolic logic, probability, and value theory aid our understanding of how physicians reason. Science 1959; 130: 9-21. This is the paper that is usually credited with beginning the field of medical informatics or at least medical decision making. It is still as clear an exposition of the rationalist Bayesian stance and naive Bayesian techniques as anything ever written. (paper only - in the library)de Dombal, F, Leaper, D, Horrrocks, J, Staniland, J, and McCann, A,(1979) Human and computer-aided diagnosis of abdominal pain: further report with emphasis on performance of clinicians. British Medical Journal, 1974. 1(1974)(904): pp. 376-380.
de Dombal, FT, Computers and the surgeon - a matter of decision. Surgery Annual, 1979. 11: pp. 33-57.
Two of several papers from de Dombal's work on Bayesian diagnosis of Abdominal Pain and related problems - the group I came here to work with originally. Still the most careful data collection and best attention to inter-observer variation in the literature. (paper only in the library)Warner HR, Olmstead CM, Rutherford BD. HELP - a program for medicaldecision-making. Computers and Biomedical Research 5:65-72, 1972.
One of the original descriptions of HELP, still one of the most successful deployments of decision support, although never successfully migrated outside its original milieu in Salt Lake City. There's a recent update article that I haven't seen: Gardner RM, Pryor TA, Warner HR. The HELP hospital information system: update 1998. Int J Med Inf. 1999 Jun;54(3):169-82.
Clancy WJ and EH Shortliffe, editors. Readings in Medical Artificial Intelligence: The First Decade.
Addison-Wesley, 1984.E.H. Shortliffe. MYCIN: Computer-Based Medical Consultations. New York: American Elsevier,
1976. The original book on Mycin of Ted Shortliffe's thesis. If you ignore the stuff on uncertainty and Shaeffer Dempster theory at the back, it is still an amazing description of where we thought we were going. Copies in the library.AI in Medicine Workshop book. The four classic papers. I shall try to arrange a copy of the papers not in Clancy and Shortliffe, but now precious as out of print and some of the papers available nowhere else
Blois MS. Information and Medicine: The Nature of Medical Descriptions. Berkeley CA: Univ of California Press, 1984. (Out of print. We are trying to resurrect a copy. Still worth reading)
The UML Reference centre www.rational.com/uml/. This is in many ways the most useful UML reference site. Links to tutorials, documentation, the latest variations in the standards, and much more. It is maintained by Rational whose tool RationalRose has become the industry standard despite being horrendously expensive and with a licensing policy that ties it to a single hard disk. Rumour has it that an educational price is now available. .The UML Home Page www.uml.org/ The primary source maintained by the Object Management Group (OMG), Look under 'resources' for various links.
ArgoUML - the tool we are using in this course. See under software.
Or type UML into Google and stand back for the avalanche (nearly 1M sites)
There are innumerable books on UML and 'object oriented design'. This is a small sample.
Albir S S (1998). UML in a Nutshell. Sebastapol, CA, O'Reilly. O'Reilly's ...in a nutshell books are good handy references and sensible non-doorstop sized. This one won't teach you UML, but if you need a quick clear reference to supplement the various online sources and handouts, this is probably it.Use in Medical StandardsMartin J and Odell J J (1992). Object-Oriented Analysis and Design. Englewood Cliffs, New Jersey, Prentice Hall. What used to be the best source text. Still lots of examples and clear. But big and expensive.
Jacobson I, Christerson M, Jonsson P and Overgaard G (1992 (revised 1996)). Object-Oriented Software Engineering. Reading, MA, AddisonWesley. The 'bible' for software engineering for much of this
Rumbaugh J, Jacobson I and Booch G (1999). The Unified Modeling Language Reference Manual. Reading, MS, Addison Wesley. If you want the official source on paper, here it is. Very well put together.
HL7 RIM http://www.hl7.org/Library/data-model/RIM/C30113\rim0113l.zip or navigate to the current version from www.hl7.org -> Technical Committees -> Modelling and Methodology -> Data ModelsCEN Electronic Health Care Record Architecture ENV13606 most easily available from http://www.chime.ucl.ac.uk/HealthI/EHCR-SupA/documents.htm
Beyer H and Holtzblatt K (1998). Contextual Design. San Francisco, Morgan Kaufmann. One of the most exciting things about designing real systems I have seen. It has given rise to a whole range of daughter methodologies. Worth reading for anybody designing systems, even if you can't follow the entire methodology.Checkland, Peter(1999). Soft Systems Design in Action. John Wiley and Sons. The most recent of Peter Checkland's books. See also web sites for Gilbert Cockton http://osiris.sunderland.ac.uk/~cs0gco/welcome.htm
Loosely related but well worth reading, Norman D A (1998). The Invisible Computer: Why good products can fail, the personal computer is so complex, and information appliances are the solution. Cambridge Mass, MIT Press. Donald Norman is one of the gurus of HCI going all the way back to Xerox Parc and the invention of the GUI interface.
Donald Norman, The Invisible Computer
XML and its relatives are not really the subject of this course, but lots of the material uses XML in one way or another and you need at least to know what it is. We'll cover the very very basics in the course. Below are a few directions to go if you want or need more information. But beware, this is an area in which information is inevitably obsolescent even before it is published.
W3Schools - by far the best source for XML related technologies including RDF (but not OWL). www.w3schools.com. Much more up-to-date than any text book.
The XML Cover pages. www.oasis-open.org/cover/sgml-xml.html Comprehensive but daunting. Everything you ever wanted to know. There are good tutorials etc..W3C - www.w3c.org - the committee that runs the Web and an ultimate source for everything with many useful links concerning XML, RDF, the Semantic Web, etc. A flood of material, but if you want to get involved, her is where it is.
There are even more books on XML than UML. If you buy a book, try to get something that is no more than a year or so old and that covers XSL and XSLT and XML Schema. The standard changes rapidly. If you want to use XML with IE5 you need XML for IE5 because the Microsoft version is non-standard. (For once you can't blame Microsoft. They rushed a version out to get it in IE5 before the standard was finished, and it changed under them - or at least that accounts for some of the differences.) The books tend to be doorstops. I have a variety of older ones which you are welcome to browse. There is an XML in a Nutshell published in Jan 2001 which I haven't look at closely. Reviews are mixed. From the table of contents it seems to cover everything but XML schema. Appears to be unusual in making a clear distinction between XML for data and for text. There are any number of tutorial XML books, but be careful that you know what you want to do with XML and that the tutorial covers it. In a big, fast moving field, it is easy to get tutored on everything except what you need to know.Another option - or rather a supplement - is Crane Softwrights - www.cranesoftwrights.com/training/ - electronic books perpetually updated on a variety of XML related topics. Not cheeap (US$40) and not for the fainthearted, but useful if you really have to get 'down and dirty', particularly with things like the differences between dialects used in different tools and software, etc.
Russell and Norvig, Artificial Intelligence: A modern approach. Prentice Hall. Available at Blackwells. Probably the best introduction to AI but not always an easy read. Includes some material on knowledge representation, rule based systems, belief nets, etc. plus all the AI basics including logic and search. There is an excellent web site with lots of material. at www.cs.berkeley.edu/~russell/aima.htmlWinston P H, Artificial Intelligence, 3rd Edition. Addison Wesley, 1992.
Cheerful and accessible but shallow. (Be sure it is the 3rd edition at least.) No material on belief nets.Rich E & Knight K, Artificial Intelligence, 2nd Edition. McGraw Hill, 1991. A compromise between Russell & Norvig and Winston, lacks material on Belief nets.
Jackson P Introduction To Expert Systems, 3rd Edition. Addison Wesley Longman, 1999. An excellent set of examples of developing real "expert systems" (it has kept the title even though the phrase is now deprecated) Lots of practical examples in different fields. Should be in the JR library, as it was removed from the CS library when it was consolidated. May be on short loan in the CS resource centre.
Schaum outline on Logic. There isn't a lot of logical notation in the course, but it is worth getting familiar with. If you don't find enough in one of the other sources - or you ever find yourself needing to get up to speed quickly, students consistently report that this is a good way to do it. Lots of examples and exercises on the basics.
See under main links above.
See under main links above
ArogUML is an opensource UML tutorial tool. Unlike TogetherJ (also available on the Department of Computer Science machines) it does not actually generate code and IDL interfaces. But it does step you through the main issues in developing a UML model. The main web page with downloads etc. is at http://argouml.tigris.org.
The belief net software we will use in this course. See under Belief Nets.
CLIPS is the open source descendent of OPS5, the original forward chaining rule based language used in systems such as Internist-I and Casnet. DLIPS is still heavily used for the programming under the EON project and various other things. We won't actually get to grips with it in this course - just too much, but here are some basic references.
Giarratano & Riley, Expert Systems: Principles and Programming, 3rd edition, Thompson. Comes with a CD-ROM containing CLIPS 6.05 executables, source code, and documentation. The first half of the
book is theory oriented and the second half covers rule-based programming using CLIPS.Clips home page http://www.ghg.net/clips/CLIPS.html The full monty on CLIPS including a gargantuan manual (Should you actually want to, print it out in 100 page segments. The whole things chokes most print queues - 300Mb or more of TCL)
Knowledge Couplers - Larry Weed's approach to diagnosis. Take it with large grains of salt but worth looking at. Make up your own mind. http://www.pkc.comMiscellaneous AI links including The Brain, Turing, Hofstadter, news stories, etc. all from general introductory AI courses last year. www.cs.man.ac.uk/ai/Modules/CS1412-2001/sites.html
A good reference list on computerised guidelines, at least at time of updating www.nmsr.labmed.umn.edu/~relson/guideline_refs.html
General article on Computer Based Evidence Based Medicine from Lazoff http://www.medicalcomputingtoday.com/0nvebm.html
Shortliffe: The Adolescence of AI in Medicine: Will the Field Come of Age in the 1990's? Artificial Intelligence in Medicine, 5:93-106, 1992. Abstract and PDF here
(Old but still asks pertinent questions. Sadly the answer is probably "not quite in the 1990s, but looking more hopeful for the 2000s)
I presume everybody has an Adobe Acrobat PDF reader but if not they can get one from
http://www.adobe.com/products/acrobat/readstep2.html - or therabouts if they've moved it.