Imagine with us that you are an Oriental Medical practitioner in the early twenty-first century. You are examining a patient referred by an internist, a colleague in your H.M.O. . . . . . . .

As you complete the survey portion of your examination your record keeping program prompts you for further signs suggested by the information you have already entered. As you make a selection, entering it into your patient's record, the interview form rearranges, becoming more specific as the record becomes more complete. Because your record keeping system is based on standard pattern definitions developed and researched in China, Taiwan, Japan and Korea, your system can remind you to pursue more subtle areas of investigation. As in the classical ideal, your examination is unique to each person. Thus, for example, if symptoms of heat in the jue yin began to mount, you would be reminded to explore not just direct organ associations but, for example, historical, psychological or biomedical signs that would indicate a predisposition to liver wind. When you were unsure of a diagnostic clue help from the fourth edition of Paradigm's Practical Dictionary would be available on-line. Where appropriate, visual images would be displayed. When useful, an objective scale such as a thermometer reading or a pain inventory would be immediately available. Thanks to decades of Chinese attention to the integration of T.C.M. and biomedicine, and more than a half-century of Japanese biostatistical research, the referring physician's diagnostic data will simply be another source of clinical clues. Although you will only rarely think in terms of a biomedical diagnosis, your ready access to epidemiological and clinical research, as well as basic science studies, will allow you to effectively integrate the results of biomedical tests and measures.

When you examine your patient's tongue, you may chose to use ordered, rigorously labeled visuals for comparison. So too with pulse examination, abdominal and point palpation. Graphics labeled with strictly used English terms will insure that the relationship to a pattern that evolves is precise and consistent from patient to patient. Because the language of your colleagues' patient records would be similarly consistent and precise, your case records will be mutually valuable. Indeed, because the standards that have evolved through the Council of Oriental Medical Publishers are available to everyone, records stored in a variety of systems will be reliably summarized. Statistics drawn from thousands of practitioners will be of inestimable value, not just to clinicians but also to experts in education and research.

Before you began treatment you could ask for a list of examination and treatment points for consideration. You might select Japanese palpatory assessments, a list of potentially reactive points, Korean hand acupuncture points, Manaka tests, or abdominal patterns, depending on your training, particular talents and interests. Because of inter-rater testing you would know the general reliability of any particular assessment as well as the strengths and weaknesses of your own diagnostic skills. During the treatment you could select and record acupoints and techniques drawn from your prior treatments, similar cases or academic research. When you wrote a prescription, you would use an on-line database of formulas, modifications and outcomes. Because of its size and cost the Chinese formula outcomes database will have never been translated into English. Yet, this vast resource of experience will be available to you because a one-to-one link between medical Chinese and English and international record keeping conventions allow accurate ``data mining.'' What you needed to know in English, your programs would search in Chinese, because your formal terminology developed through continuous review and update by linguists, clinicians and scholars from many countries.

As you consider each individual medicinal in your prescription you will be able to reference pharmaceutical and botanical databases through standard Latin, common and commercial names. Because these generic names are freely-available, all of your favorite commercial products will be accurately related one to the other as well as to the published clinical trials on which their advertising claims are based. A translation of the 7,200 item translation of the Materia Medica of the People's Republic cross referenced to their 3,200 entry formulary will be available on CD-ROM. Because your Oriental Medical records never use biomedical terms, except by biomedical standards, referring physicians' measurements will fit seamlessly into your records. Disease names and biomedical standards will be easily compared to similar information from Asia. When you wish, resources such as MedLine, the P.D.R. or other standard biomedical references would be available on-line.

Clearly, these potentials remain in the future and not all would apply to any one patient. It is also likely that practitioners would develop reference sets that fit the needs of their clientele. In fact, that might become one of the major advantages of using computer systems in clinical practice. And, while this story is fiction today, there is not one of these features that is not now possible. In fact, it seems to me that there is really nothing in this fiction that is not already programmable.

Roughly quantified the 3,200 formulas, 7,200 medicinals of P.R.C. formularies, the upwards of 2,000 channel and non-channel acupoints compiled among several systems, and the 30,000 concepts and symptoms of modern P.R.C. dictionaries are within the storage and speed capacity of today's PC hardware. There are several database and inquiry systems -- Oracle, Sybase, Progress and others -- that have a design capacity and an application development environment suitable for the task. While the potential scope of the data and its relations is sufficient to suggest that personal hyperstack products and older relational systems might be less than ideal choices (because they lack double-byte characters, record locking, roll forward/back recovery, multi-user and multi-processing features), there is certainly no lack of reasonably priced tools that could do the job.

Although the choice of an application development system is an interesting forum topic, the fact that there is plenty of room for discussion in that regard serves to highlight the fact that what we need is not hardware or software but conventions and agreements. This then is the intent of this forum: to discuss how computers can be well used in Oriental medicine; how developing and employing information technology could help research, train and deliver Oriental medicine in today's medical environments. Ideally, the discussion will attract scholars, linguists, clinicians and experts in the information sciences. We will see! For now, we think several topics suggest themselves.

Information Management vs. Automation
First, please distinguish this intent from an effort to establish an ``automatic'' Oriental medicine. It is a broad field. There are those who see Oriental medicine as essentially and inherently psycho-spiritual, those who see it as entirely clinical, and many of us somewhere between. There are those whose practice is solely traditional, those whose practice is biomedically founded. Because Oriental medicine developed at the bedside and is rooted in highly trained human perceptions, mental and physical skills, sophisticated Oriental medicine will probably always demand hands-on training and human discipline. Thus the effort we envision here is not the ``automation'' of diagnosis or the construction of a therapeutic automata. Instead, it is the building of information tools of benefit to clinicians, medical researchers and scholars.

Discussion of Design Issues
Topics that are usefully developed in a public forum are those that benefit from a global perspective. We think that first among these is a discussion of the most basic design issues. For example, it is clear that there are several relations (where ``names'' is the English nomenclature associated with each data set and ``indications'' is the world of definitions, information about illness, etc.):

These are simple relationships, at least in part, because formulas can be related to constituents, etc. In reality though the set of relations is very rich. There are patterns. These patterns are composed of primary and secondary indications. Primary indications are those without which a pattern cannot be said to exist and secondary indications are those that may or may not exist in association with that pattern. These may or may not indicate a variance in the therapy associated with that pattern. Furthermore, in modern Asian practice there are also one-to-many relationships to biomedical diseases, or ``western medical correspondences.'' From this view, the data actually looks something like this (simplified for display):

• i = indication/pathocondition (primary/secondary not distinguished here)
• p = pattern
• d = disease

   d1                  d2                  d3                   d4
   |                   |                   |                    |
   |                   |                   |                    |
__________________     |                _____________________   |
|   |       |   |______|_________________|__|  |   |    |   |   |
|   |       |   |       |   |   |   |    |  |      |____|___|___|__
i1  i2  i3  i4  i5  i6  i7  i8  i9  i10  i11  i12  i13  i14  i15 i16
|       |   |    |   |  |   |   |    |    |    |    |    |    |   |
|       |   |    |   |  |   |   |    |    |    |    |    |    |   |
---------        ---------------     -----------------------   -----
  p1                  p2                   p3                   p4

Notice that the relationships are not generally one-to-one, although in modern Asian practice there are some one-to-one relations between biomedical diseases and therapies, what the Chinese call ``syndromes.'' Patterns and diseases relate to indications unequally, and some indications are excluded from one or the other. From a databasing view, the file structure looks like it should have two upper level files, patterns and diseases. These relate to a lower-level file of indications, each of which is a record based on a unique, Chinese-named concept. Patterns also have a second dimension, where the indications that identify a pattern (distinguished as primary and secondary) also identify acupoints and herbs used to treat those indications, thus we get:

• i= indication/pathocondition
• p = pattern
• a = acupoint
• h = medicinal

f = formula

i1  i2  i3  i4  i5  i6  i7  i8  i9  i10  i11  i12  i13  i14  i15 i16
|       |   |    |   |  |   |   |    |    |    |    |    |    |   |
|       |   |    |   |  |   |   |    |    |    |    |    |    |   |
---------        ---------------     -----------------------   -----
| p1 |  |        |    p2 |  |   |    |     p3  |    |    |    | p4 |
-------------    ---------------     -----------------------   -----
|   |   |   |    |   |  |   |   |    |    |    |    |    |    |   |
|   |   |   |    |   |  |   |   |    |    |    |    |    |    |   |
a1  a9  |   a3   a8  a5 a6  |   |    a3   a8   a9   |    a1   a6  |
|   |   |   |    |   |  |   |   |    |    |    |    |    |     |  |
h1  h2  h13 h4   h1  h2 h7  h8  h9   h1   h9   h10  h11  h1    h5 h1
______________   ________________     _____________________     ______
        f1                f2                    f3                   f4

Patterns relate to acupoint and medicinals via formulas which are again not one-to-one in any fixed way.Thus, the pattern records cross reference formulas, which cross reference medicinals and acupoints, as well as cross referencing indications
(where CR = cross reference):

Pattern Record =
Names,
Descriptions (definitions per Chinese sources)
CR = Indications,
CR = Formula
Formulas Record =
Names,
Descriptions,
CR = Acupoints,
CR = Medicinals.
Medicinals Record =
Names,
CR = Indications.
Acupoints Record =
Names,
CR = Indications.
Indications Record =
Names,
Descriptions (definitions per Chinese sources),
[In some] Descriptions (definitions per biomedicine).

An indication field in a formula or pattern record is really just an index to indications files. This provides an additional quantification, there being some 30,000 individually-defined concepts preserved in modern Chinese medical dictionaries. However, theoretically, were the conceptual database to achieve a similar richness, a query from any direction could reach anywhere (e.g. knowing an indication can give me not only its names and definitions, but its relations to patterns, formulas, medicinals, acupoints, and even diseases).

In practice however this diagram changes according to different ``schools of thought'' where different sets of indications lead to different pattern and formula relations to medicinal and/or acupoints. This can be seen as:

                          Modern T.C.M.
________________________________________________________________
|                    |                       |                 |
 p1                  p2                      p3                 p4
-------------    ---------------     -----------------------   -----
|   |   |   |    |   |  |   |   |    |    |    |    |    |    |   |
|   |   |   |    |   |  |   |   |    |    |    |    |    |    |   |
a1  a9  |   a3   a8  a5 a6  |   |    a3   a8   a9   |    a1   a6  |
|   |   |   |    |   |  |   |   |    |    |    |    |    |     |  |
h1  h2  h13 h4   h1  h2 h7  h8  h9   h1   h9   h10  h11  h1    h5 h1
_________   __________________  _________________   _______________
   p1                p2                p3                p4

                           Classical

In other words, schools of thought re-order acupoints or medicials relations to patterns. To account for this, some pattern files may need to be subdivided into classes. As well, treatment patterns, which relate to both medicinal and acupoint use, serve to summarize and distinguish indications and their relationship to patterns.

SO, the question becomes . . . How do we get there from here. What's wrong with these admittedly simplified descriptions and what needs to be discussed and decided before the beginnings of the system described in the introduction to this piece might really come together? What do the records look like? What data do we need and what are its criteria? How are the many one-to-many relationships to be made and what is and what is not a valid relation?

A few questions come to mind:

• What are the file structures?
• What are the important relations?
• What are the most useful inputs and outputs?
• What tools can practitioners, scholars use?
• What are the appropriate uses?
• What systems and languages would be most appropriate?
• What are the educational uses, needs and concerns?
• What about cost?

Obviously, there could be many more areas of discussion. Chief among these would be functional criteria ---- what are and what are not valid criteria for relation. Can one, for example, establish a relationship between symptoms and signs used in prior eras with those used today? How do we establish that entirely quantitative concepts are used in usefully similar ways by different clinicians? When the Oriental medical vocbulary is so soundly rooted in practitioner responses and actions (e.g.: the perception of color), how can we accomplish clinically reliable standards? How do we resolve conflicting claims of clinical validity?

Intents, conventions and participation guidelines
What would be nice for this to become would be a relatively specific discussion among those who have a specific interest in the topic: users, educators, information system specialists, designers, programmers, and researchers. There need not be many participants, fewer people with specific interests would actually be more interesting. Neither should those who are commercially-interested be discouraged; indeed, one effective use of a forum might be the development of cooperative labels, such as those achieved by the Council of Oriental Medical Publishers, that would make it easier for people to effectively use more than one system. However, I don't feel this is an appropriate place for either advertising or reviewing commercial products. The OrMed mailing list is entirely suitable for such reviews and I am afraid that people whose commercial interests diverge would be less likely to participate openly if they felt a need to defend their products (or the urge to promote them). So, while I think it would be completely appropriate to say, for example, ``I wish these systems could be open so that I can add my own indication relations,'' I feel it would be inappropriate to say, ``System-such-and-such is a dud because I can't add my own indications.'' If we stay generic, there are several things we might accomplish.

The idea is not so much a moderated forum but an edited one. If in time there are contributions that become a foundation of the discussion , I would like to include these in a developing overview -- with of course footnotes to the contributors. So, my current intent is to let the posting be very open, direct and unedited. This is not to say that spam and slam posts won't be excised, only that I will be looking to filter posts with an eye toward eventual inclusion in a forum publication. I intend that the forum materials be ``copy lefted,'' such that they are available for anyone to publish in any form, provided they remain intact and are appropriately documented as to source.

Bob Felt

Forum Editor

bob@paradigm-pubs.com