HISA (Vic) home Magazine back issues Previous Article Next article Title Contents

Overcoming Medical Notation Blues - The Docle90 Method


Dr Y Kuang OonThe basis of a good medical record system is a shorthand notation that is quick and easy to write. A real bonus is that the system is eminently computer ready.

Back in 1987, while consulting with a patient with some symptoms suggestive of multiple sclerosis, I jotted down MS in my notes. She saw my scribbling and confronted me if she had MS. This was an uncomfortable episode. During this period I had been working in the Department of Community Medicine, Monash University and had looked at ways to record medical data in computers. I was unhappy about the ICD and ICHPPC numeric coding system - GPs will never accept number codes to represent disease entities. At that moment, the germinal concept of a medical notation based on abbreviations took hold. One dividend is that the patients will not be able to understand what we write again!

Practically all doctors use abbreviations in their notes. The problems are that they are not standardized. Hitherto, there is no agreed method for converting raw medical terminology to it's abbreviated equivalent. What is needed is a blackbox that will do the job with no equivocation. In computer terms, what is needed is an algorithm. Algorithm design and selection can be a hairy topic and best covered in the Docle90 book - see references. A rough and good test is that the algorithm comes out with different results for 'AUSTRALIA' and 'AUSTRIA'. It is quite straightforward to write a computer program to implement this algorithm. Having made this leap to open all medical terms to abbreviations, the next step was to define some special characters that bind these abbreviations together to empower the notation. For instance the expression :

LBP>LEG.RHS/COUG@/\SLRreads as 'low back pain radiating to leg on right hand side, aggravated by cough, associated with abnormal straight leg raising.

The Algorithm

The algorithm itself is of general applicability. An argument can be put forward for the use of the Docle algorithm as a good programming aid - as a naming convention of program variables, procedures, functions and computer commands. A Docle type exercise may even be relevant in the general print media beyond the medical domain. At the flight information displays, one can imagine MELBN, SYDN and NEWY to mean Melbourne, Sydney and New York. The abbreviations for Austria and Australia would be AUSR and AUSRL respectively.

The Docle Algorithm Description

Definitions:

  1. The output of the Docle Algorithm is called a Docle word.
  2. Each Docle word is comprised of the printable ASCII characters.
  3. Each Docle word is unique in its meaning.
  4. No spaces are allowed in a Docle word.
  5. The characters a e i o u y are defined as Docle vowels. The rest of the character set are consonants by definition.
  6. Docle words are case insensitive.
  7. Suggestions:

i) Docle words referring to the same medical entity are avoided. Where possible, combine existing Docle words to refer to a new medical entity rather than create a new Docle word.

ii) Eponymous entities are avoided if possible.

Transformation Rules:

The medical expression is subjected to the initial process of fragmentation. A word that contains a hyphen is broken up into two words. Words that are constituted from multiple subwords are fragmented into the original source words, e.g. electrocardiography is fragmented into electro cardio graphy, X-ray is X ray. After fragmentation the word(s) are treated according to the three different possibilities of the one word, two words or three and more words situations.

Medical terminology comprised of a single word

  1. A word that has four or less characters is left unaltered.
  2. The Docle word is derived by concatenating the derived prefix to the derived suffix.
  3. iii) The prefix is defined as that contiguous group of initial characters of the source word. Scanning from the left, the prefix starts from the first character and ends at the first consonant after any vowel. In the event of the prefix being less than three characters long, the first three characters is made the prefix.
  4. iii) The source word minus its prefix is termed the remainder.
  5. The suffix is a group of characters derived by scanning the remainder from right to left. The immediate consonant after any vowel that is encountered is added to the group, until the whole remainder is scanned. The consonants thus obtained are placed in the order as found in the source word. In the event that this scanning process yields zero character, and that the prefix is only three characters long, then the first character of the remainder is made the prefix.
  6. In the event that the result of concatenating the suffix to the prefix is six or more characters long, this result is passed through the algorithm again.
  7. Medical terminology comprising two words

The first word is subjected to the same treatment as the single word terminology case. The product is then concatenated to the first character of the second word. Any mapping conflict is resolved by using alternate medical expressions as the source.

Medical terminology - three or more words

The first characters of each and every word are concatenated together.Any mapping conflict is resolved by using alternate medical expressions as the source.

Medical terminology - two or more words where part of the expression can be represented by a Docle word

The first word of the expression is replaced by its Docle word. The first characters of the remaining words are concatenated with the Docle word.

Comment: The design goal of aiming at preponderance of four character length Docle words is met. The second pass mechanism sets an upper bound of six character for single word derivatives.

Examples:

ONE WORD (Get the prefix, then the suffix and then concatenate.)

MIGRAINE -> MIG_RAINE -> MIGRN ABDOMEN -> ABD_OMEN -> ABDM FEMUR -> FEM_UR -> FEMU HUMERUS -> HUM_ERUS -> HUMR PAIN -> PAIN -> PAIN CONJUCTIVITIS -> CON_JUNCTIVITIS -> CONJTVT -> CONJ (2nd pass).

TWO WORDS

(First word subjected to transformation as one word case. The product is concatenated with the first character of the second word.)

DIABETES MELLITU -> DIAB_ETES M_ELLITUS -> DIABTM BENIGN NEOPLASM -> BEN_IGN N_EOPLASM-> BENIN X RAY -> X R_AY -> XR HERPES ZOSTER -> HER_PES Z_OSTER -> HERPZ PERSONAL COMPUTER -> PER_SONAL C_OMPUTER -> PERSNC

THREE OR MORE WORDS

(The first characters of each word is concatenated together.)

ACQUIRED IMMUNO DEFICIENCY SYNDROME -> AIDS ELECTRO CARDIO GRAPH -> ECG TRANSIENT ISCHAEMIC ATTACK -> TIA

Operators and Symbolic Shorthand

The Docle operators are designed to give the Docle language the power of expression not seen with numeric coding of data. The operators allow the user to combine two or more Docle words together to form Docle expressions. There is provision for more operators as there are more than a dozen printable ASCII characters still unused. There are, as currently defined, a dozen operators and several shorthand symbols:


Operator Meaning Example of Use<space> LOCATED AT TUBC.KIDN is read as tuberculosis located at kidney.> LEADING TO TUBC>OSTEM is read as tuberculosis leading to osteomyelitis.< DUE TO OSTEM<STAPHCA is read as osteomyelitis due to staphylococcus aureus.@ APROPOS TUBC@PREG is read as tuberculosis apropos pregnancy. @ can also be read as 'associated'.: DESCRIBED AS PAIN:DULL is read as pain described as dull.$ DATE #.HUMR$2-2-87 is read as fracture of humerus date being 2-2-87.% QUANTIFICATION LUMP.BREAS.RHS%2CM means lump at right breast 2 cm in size.() PROBABILITY MALR(0.7) means the probability of malaria is rated at 0.7. (-1) implies a certainty that it is not the diagnosis., PAUSE Use akin to comma, PAIN.ABDM@FREQC,XY,PASTH@GOUT - pain in abdomen, frequency, male with past history of gout.# FRACTURE #.FEMU.RHS is fracture of right femur./ INCREASED WCC/ reads as white cell count elevated. / is mildly elevated. // is moderately elevated. /// is grossly elevated.\ DECREASED HEMOG\ reads as hemoglobin decreased. \is mildly decreased. \\ is moderately decreased. \\\ is grossly decreased.= NORMAL WCC= reads as white cell count is within normal range. ~ INCIDENCE Distribution of parameter over time. Useful in pain description./ AGGRAVATION Accentuating factor in pain analysis. Also denotes increased tendency.See above, example of operator loading.\ ALLEVIATION Attenuating factor in pain analysis. Also denotes decreased tendency./\ ABNORMAL Plays on the idiom 'up and down'. FBE/\ implies an abnormal FBE.

Docle Notation - Style, Usage and Random Thoughts

Docle is in essence a natural language without the disadvantages of a natural language. With a Docle type standard, spelling and semantic checks can be built into programs to weed out illegal input. Alternatively because the computer representation is in Docle, the operator can verify the veracity of the data.

Laterality Construct

It is important in medicine to describe the side of the lesion or illness. The convention proposed uses the ' . ' operator which is read as 'LOCATED AT' and the Docle words RHS (Right Hand Side), LHS (Left Hand Side) and BIL (Bilateral).

#.FEMU.RHS means fracture located at femur located at right.

Denoting Sex - Phenotypic and Chromosomal Options

Instead of the Docle words MALE and FEML, it is proposed to use the sex chromosome complement to describe sexual status. This use of biological sex resolves the dilemma caused by sex change operation and sex chromosome disorders. XX denotes female and XY male.

The Analysis of Pain

The analysis of pain is of great importance in the diagnostic process. The elements of this analysis are represented with the Docle notation.

  1. site PAIN.ABDM
  2. Pain located at abdomen
  3. character PAIN:ACUT
  4. Pain described as acute
  5. radiation PAIN.ABDM>LEG.RHS
  6. Pain radiating to right leg
  7. duration PAIN %3DAYS
  8. Pain quantitatively 3 days
  9. incidence PAIN~SUDD
  10. Pain with sudden onset PAIN~CONT
  11. Pain is constant.
  12. aggravation PAIN/COUG
  13. Pain aggravated by cough
  14. alleviation PAIN\FOOD
  15. Pain alleviated by food
  16. association PAIN@SEXUI
  17. Pain associated with sexual intercourse
The case history of a 34 year old male with abdominal pain. The pain was described as steady, epigastric in location, had been present for 3 weeks. This was associated with nausea, anorexia and vomiting. Admitted to heavy drinking, had lost 10 kilograms during illness. On examination he had low fever, icterus and epigastric mass. Urine and full blood examinations were normal. Bilirubin was elevated, amylase was grossly elevated. The Docle equivalent for above is:

XY%34,PAIN.ABDM.EPIG%3week:CONT@MASS@NAUS@VOMT@ANORX@WEIG\%10KILG@MASS.EPIG@AMYL///@BILRB//@FBE=@URIN=@ALCH@ICTR

If the above is input into a system that understands Docle, the computer would list pseudocyst of the Pancreas as a strong possibility! A typical GP consult can be written: RFE<URTI,AMOX

Let's Clean Up Prescription Writing

Instead of the notation based on antiquity such as BD and the beloved TDS, a uniform approach based on time intervals is proposed.

SI reads 'seconds interval'

MI reads 'minutes interval'

HI reads 'hours interval'

DI reads 'days interval'

WI reads 'weeks interval'

This expanded scale copes with the micro doses used in insulin pump infusion to monthly Modecate injections. Hence the beloved TDS is 8HI in Docle. An injection to be given every alternate day is written as 2DI. Rather than a dogmatic proposal, this is to suggest a form of standardization and simplification is needed sooner or later with the need to computerize. The Amoxil script can be scribbled as: AMOX%250MG,8HI,20

A Systematic Tumour Nomenclature

The Docle proposal would classify neoplasms as malignant neoplasm (MALNN) or benign neoplasm (BENIN) and carcinoma-in-situ (CIS).

MALNN.PANR.HEAD:ADENC reads as malignant neoplasm of pancreas located at head described as adenocarcinoma.The hierarchical nature of the anatomic location is illustrated by the zooming effect of using more ' . ' operators.

MALNN.PANR.HEAD

MALNN.PANR.DUCT

MALNN.PANR.DUCT.WIRS - malignant neoplasm located at pancreas, located at the duct of Wirsung as opposed to Santorini (SANT).

A Systematized Nomenclature For Infections

The Docle approach uses the generic Docle word INFT which reads as infection and the operator ' < ' meaning 'due to'. Hence AIDS can be referenced as INFT<HIV, which means 'Infection due to human immunodeficiency virus'.

Quantification

A proposal to incorporate quantification in SNOMED has been made [Henkind l986]. The operator ' % ' allows one to specify the quantity in a particular situation.

SMOK%20 reads as smoke 20 cigarettes per day.

LUMP.UPPEA.RHS%2CM reads lump at upper arm at right is 2cm.

Quo Vadis

Docle is ideal for computer retrieval, hence its future role in the computerization of medicine. Certainly the two to six letter size of Docle words would make them easier to key in with fewer attendant mistakes. Because Docle is designed like a high level computer language, it is easier to construct syntax and semantic validation programs. To wait for an English language interface, one has to wait ten, twenty or even a hundred years before we can have a reliable natural language system that can be entrusted with life and death clinical decision making. Even then, most would like to communicate in some condensed precise jargon for efficiency reasons. The best can often be the enemy of the good. We can wait and procrastinate forever by waiting for some foreign invention. By plunging in and experimentation, medicos in Australia have a unique contribution to make in the field of medical communication internationally. [Copies of the computer algorithm (DOS/Mac) and Docle word listing is available, rrp valued at one standard consultation on (03) 763 8935]

References

Henkind S J et al, Medinfo 86, Pt 2, 858. North Holland 1986.

Oon, Y K, The Docle-90 medical notation. Structured Language Resources 1990.


Previous Article Next article Title Contents Your Comments
HTML by Jon Hilton SCON Solutions, +61-3-94823794
Conversion greatly assisted by rtftoweb.
Copyright © HISA Inc.
Revised Tue Sep 12 12:17:34 1995