ISPRAS Logo
Compiler Technologies Team
Intro Background Projects Offers Publications Downloads Staff
ISPRAS Description Logo
Home   R&D Groups   Projects   Grants   Publications   Downloads

The system of information support of treatment based upon modified method of dynamic cluster analysis

       The system "Doctor's Partner" is organized in two levels. The lower level (kernel) called The Universal Quantitative Qualifier performs an abstract task of pattern recognition. The top level is a shell. It addresses a wide range of problems in real-time intelligent monitoring and informational support for treatment:

  • Continuous acquisition and interpretation of patient data.
  • Diagnosis.
  • Selection of treatment.
  • The outcome forecast including alarming of unexpected disorders and complications.
       In training doctor enters actual or simulated cases of individual patient-s data including signs, diseases, treatment and outcome. To use the system as diagnostic software, type in patient-s signs and it will offer some possible diagnoses. The differential diagnosis allows for more than one disease to be present. One can add a disease by his experience or by comparing patient-s data with disease-s one in system directories. Then the system controls an adequacy of patient-s signs for each disease and transmits the data into a kernel for an estimation of probability that patient has a particular disease.

       In selecting a treatment the doctor operates with signs, diseases, stages and the selected variant of treatment as the features. The possible outcomes will constitute a differential set. The patient belongs to the class "outcome".

       The modified method of cluster analysis makes it possible to classify under the incomplete set of signs, particularly in the absence of specific manifestations of identifying diseases. After estimation, if there is more then one likely disease in the differential set, additional signs or new tests will help to differ the diseases. If those no, usage of weights will help to select the most probable disease. As for planning of treatment and forecast, the weights of signs will help us to establish the probability of outcomes in the differential set. After estimation we can get an alarm of unexpected disorders and complications as a most likely outcome.

       In contrast with the majority of intelligence and expert systems, the system is not limited to a separate area of medicine. The system helps the doctor to make a decision for or against the selection, not imposing the selection.

       At its most basic level, it functions as an electronic textbook of medicine. It helps the doctor to mare a decision under the incomplete set of signs, particularly in the absence of specific manifestations of identifying diseases.

       The developed system of analogues helps the doctor to use his own experience in terms of preceding patient cases (signs, disease, stage, treatment, outcome).

       The system is ideal for practicing your own skills, through randomly generated patient scenarios, viewing references.

       Finally, it functions as a teaching tool for students and trainees.

 


Copyright © 2002 ISP RAS