Computer Predicting Complications after Acute Myocardiac Infarction

1.Kats Mark Davidovich, 1937, specialist in applied mathematics , Dr. Sc., Academician of Ukrainian Process Engineering Academy and Academy of Ecological Sciences, Corresponding Member of International Academy of Computer Sciences and Systems. Address: Donetskaya St. 37-24, Severodonetsk, Lugansk Reg., 349940, Ukraine. Tel. (380-, 6452)-3-08-75, E-mail: (Kats M.D.).

2. The title of the project.

"DEVELOPMENT OF THE INTELLIGENT SYSTEM FOR COMPUTER PREDICTING COMPLICATIONS AFTER ACUTE MYOCARDIAC INFARCTION".

3. Field of application.

Medicine. After the project has been completely developed the problem of reliable predicting the complications after myocardiac infarction will be solved on base of the data obtained on the very first day of a patient's stay in the hospital.

4. Offer description.

The problem of correct predicting the complications after myocardiac infarction in acute stage has not been solved yet.

Jointly with medical centers specializing in cardiac diseases on base of presented by them information we are able to construct a mathematical model predicting the complications after myocardiac infarction in acute stage.

After clinical testing and correction of the model the package " THE INTELLIGENT SYSTEM FOR COMPUTER PREDICTING THE COMPLICATIONS AFTER ACUTE MYOCARDIAC INFARCTION" will be developed which after patenting will represent end-use software product ready for commercialization in all the countries.

Mathematical representation of constructing the model for predicting the complications after myocardiac infarction is the following:

GIVEN: The table of experimental information M=XxY (X={Xij}, i=1,m, j=1,n; Y={Yil}, l=1,k) with each line containing information on parameters of the symptoms {Xij} and a versatile complication in terms of remote consequences Yil for i-patient, where m - number of lines (patients) in table M; n- number of columns (symptoms) in table M; K- number of complications to be differentiated.

IT IS NEEDED: To construct on base of table M using formalized procedures a mathematical model reliably distinguishing between each of complications to be differentiated.

When the table basing on investigation results has been obtained with each line containing parameters of symptoms and verified diagnosis of the complication for a particular patient a mathematical model for predicting complications can be made by method of mosaic portrait using the following formalized procedures:

- converting the table M of initial experimental data to table M' by transferring from continuum scales in which the symptoms are measured to discrete scales. In doing this each subrange of each symptom parameter is assigned its own code;

- selecting code combinations characterizing patients having a particular complication which is not found with any patient having other complications.

Such combinations are interpreted as differential syndromes of corresponding complications. The mosaic model thus obtained consists of syndrome subranges K, each containing differential syndromes belonging only to one from K of disease complications to be differentiated. The number of differential syndromes obtained while constructing a mosaic model is quite large and in fact almost all of them are unknown in cardiology.

5. Advantages compared to analogs

At present there are no mathematical methods for construction of differential syndromes within reasonable time for each daisies under differentiation even with 15 symptoms available.

6. Potential users.

The potential users of the package " THE INTELLIGENT SYSTEM FOR COMPUTER PREDICTING THE COMPLICATIONS AFTER ACUTE MYOCARDIAC INFARCTION" can be cardiological centers.

7. Status of the development.

AS A RESULT OF STRATEGIC INVESTIGATIONS AIMED AT SEARCHING NEW APPROACHES SOLVING THE PROBLEM OF CORRECT IDENTIFICATION OF OBJECTS WHICH INFORMATIVE CHARACTERISTICS ARE CLASSIFIED AMONG CLASS OF "LARGE-SCALE" SYSTEMS, WITH THE HELP OF EXPERIMENTAL INFORMATION OBTAINED IN OBSERVING REGIME, THE METHOD OF MOSAIC PORTRAIT BASED ON IDEAS OF ARTIFICIAL INTELLIGENCE HAS BEEN DEVELOPED, TESTED AND EXPERIMENTALLY PROVED IN SOLVING SCIENTIFIC, RESEARCH AND APPLIED PROBLEMS.

METHOD OF MOSAIC PORTRAIT ALLOWS CONSTRUCTING INFORMATIVE MATHEMATICAL MODELS OF OBJECTS OF DIFFERENT NATURE WITHOUT AN EXPERT IN THE FIELD CONCERNED BASING ON THE TABLE OF EXPERIMENTAL DATA. THE MODELS CONTAIN SIGNIFICANT CONTENT OF NON-TRIVIAL, UNKNOWN BEFORE RELATIONSHIPS CONNECTING INPUT AND OUTPUT VARIABLES.

As applied to predicting complications after myocardiac infarction in acute stage the mosaic portrait method allows to obtain for each complication its own set of differential syndromes, most being unknown before.

The software program is available for realization of mosaic portrait method which basing on corresponding experimental data allows to construct the model for predicting complications of different diseases.

The structure of the package is available where a model of any differential diagnostics or predicting can be input.

The diagnostic package is already in action realizing intelligent system "PREDICTING THE COMPLICATIONS AFTER ACUTE MYOCARDIAC INFARCTION" with mathematical models developed jointly with Military Medical Academy (St. Petersburg, Russia).

The following problems are solved with the package:

- predicting one or more complications after myocardiac infection basing on patient's data obtained at the first day in hospital;

- in case of predicting several complications , the probability of each is realized;

- recommendations on preventive therapy for the complication(s) predicted and applicable symptomatic therapy considering compatibility of medicines and treatment procedures;

- providing every 48 hours a new prediction adjusted for treatment results and corresponding recommendations on preventive and symptomatic therapy;

- presenting information on differential syndromes on which the prediction has been based, when required;

- input to data base of patient's data, prediction on complication, syndromes on which the prediction has been based and the treatment recommended.

Initial information on patient's state including anamnesis, examination, cardiograph diagram (39 criteria in total) is realized in push-button mode as answers to questions.

Experimental testing of the model at the cardiological clinic of Military Medical Academy (St.Petersburg), 23rd Moscow clinical hospital, 20th St. Petersburg city hospital, 42nd St. Petersburg city hospital showed that accuracy of prediction was 85-88%, the preventive therapy applied reduced lethal outcomes from myocardiac infarction by 36-45%.

Improvements of the package on the basis of medical traditions (list of required symptoms ) and requirements to interface and execution meeting standards are needed.

8. Patenting status of the offer.

Method of mosaic portrait is not patented, but main part of the algorithm for its realization providing polynomial dependence of computer time from dimension of the task has not been published; the computer program for its realization is kept by the author.

Models for differential diagnostics and predicting the complications after myocardiac infarction included in the packages concerned are not patented.

9. Ownership of the offer belongs to Kats Mark Davidovich.

Information supplied by the Author October 1999.  Page last updated: January 19, 2004