<<
>>

Sampling of a mathematical apparatus for an estimation of efficiency of the gained results and an estimation of quality of work of mathematical models of forecasting of a heart attack of a myocardium in the rehabilitation period

For an estimation of efficiency of the gained results it is used ROC - curves (ROC - Receiver operating characteristic) which, being the schedule, allow to state an estimation to quality of model on separation of two class-rooms.

ROC - the curve shows dependence of quantity of truly classified positive instances - diagnostic sensitivity (DCH) from quantity of incorrectly classified negative instances: 1 diagnostic specificity (DS). Thus it is supposed, that the qualifier has some parametre, varying which, was possiblly to vary a relationship between DCH and DS. Parametre name a threshold or a point of cutting off of one diagnosed class-room from another. This point the drawing most close had to the overhead left angle or it is maximum udalenna from the diagonal straight line.

For an estimation of quality of model with heterogeneous qualifiers on forecasting by IT in the rehabilitation period according to structure (drawing 4.2) us in polyclinics of a city of Kursk it has been selected 230 patients who have transferred acute IT. Sampling of volume training and control vyborok was carried out under recommendations [62]. Using the gained
The various parametres characterising risk of emersion PIM as scales for construction of bar graphs of distribution of class-rooms ω0и shch = soim, have been defined the classification thresholds, allowed to solve a problem ROC - the analysis.

In the capacity of a prototype for the heterogeneous qualifier scale ASCORE [95] has been chosen. In table 4.3 figures of merit of forecasting PIM on control sample in model of the heterogeneous qualifier with use of all heterogeneous space IP which are compared with the analogous results gained on scale ASCORE are resulted.

Table 4.3 - Experimental data on forecasting PIM for control sample by the heterogeneous qualifier builted with use of all heterogeneous space of informative signs, and on scale ASCORE

The surveyed Model of the heterogeneous qualifier Scale ASCORE
DCH DS DE DCH DS DE
‰ = 100 90 % 86 % 88 % 75 % 71 % 73 %
= 60 86 % 90 % 71 % 75 %

For construction of ROC-curves we will take advantage of bar graphs of heterogeneous qualifiers, the instance of one of which is presented in drawing 4.18.

We are set by a range of change KUPor - a parametre on an exit of the heterogeneous qualifier above which the unknown sample will belong to the class ω1. After that, having chosen a suitable step of change KUPor. We will define figures of merit of the heterogeneous qualifier.

Such images, in this case the ROC-curve for each heterogeneous qualifier is functional dependence of figures of merit of classification DCH and DS from KUPor. Considering character
Approximations of bar graphs Hωoи Hω1 presented in drawing 4.18, the step of change KUPor is chosen 0,05, and a dynamic range 0,5...

0,8.

Drawing 4.18 - the Typical bar graph of distribution of class-rooms ω0и ω1по

To magnitude Kyω1

In tables 4.4... 4.8 results of the ROC-analysis for the builted heterogeneous qualifiers are resulted.

Table 4.4 - Results ROC - the analysis for the heterogeneous qualifier builted on all space of informative signs

KUpor 0,50 0,55 0,60 0,65 0,70 0,75 0,80
DCH 0,95 0,94 0,92 0,90 0,83 0,75 0,62
DS 0,74 0,78 0,84 0,86 0,90 0,94 0,99

Table 4.5 - Results ROC - the analysis for heterogeneous

The qualifier builted on space of informative signs without virtual streams

KUpor 0,50 0,55 0,60 0,65 0,70 0,75 0,80
DCH 0,94 0,91 0,88 0,84 0,78 0,72 0,65
DS 0,72 0,78 0,82 0,86 0,90 0,93 0,96

Table 4.6 - Results ROC - the analysis for heterogeneous

The qualifier builted on heterogeneous space of informative signs with exclusion of virtual streams and bioimpedance researches

KUPor 0,50 0,55 0,60 0,65 0,70 0,75 0,80
DCH 0,92 0,88 0,84 0,78 0,74 0,69 0,65
DS 0,72 0,80 0,84 0,88 0,90 0,92 0,94

Table 4.7 - Results ROC - the analysis for the heterogeneous qualifier builted on heterogeneous space of informative signs with exclusion of virtual streams, bioimpedance researches and the psychoemotional pressure analysis

KUPor 0,50 0,55 0,60 0,65 0,70 0,75 0,80
DCH 0,80 0,79 0,78 0,76 0,69 0,60 0,5
DS 0,67 0,71 0,77 0,79 0,86 0,90 0,93

Table 4.8 - Results ROC - the analysis for heterogeneous

The qualifier builted on heterogeneous space of informative signs with exclusion of virtual streams, bioimpedance researches, the analysis of a psychoemotional pressure and parametres perekisnogo oxidations lipidov and antioxidative activity

KUPor 0,50 0,55 0,60 0,65 0,70 0,75 0,80
DCH 0,77 0,77 0,76 0,72 0,65 0,58 0,45
DS 0,66 0,70 0,75 0,80 0,84 0,87 0,9

The analysis of results of use of all six making spaces of the informative signs offered in-process in a final solving rule, shows, that is thus provided
The highest quality of classification though is required an additional time and means.

Drawing 4.19 illustrates comparative characteristics of heterogeneous qualifiers on a ROC-plane. The analysis of these schedules allows to size up the contribution of this or that group of informative signs to the general prognostichesky process.

Drawing 4.19 - ROC-curves of six heterogeneous qualifiers

So, to the signs gained without application of difficult special hardware components and methods, by standard diagnostic study of the patient, possible in an ordinary polyclinic, we gain the summary figures of merit resulted in table 4.8.

The analysis of the gained results shows satisfactory coincidence of results of an expert sizing up and quality check of the work, the synthesised heterogeneous qualifiers on control sample on parametres DCH and DS. Thus, results of check on control sample of an overall performance of rules of forecasting PIM in rehabilitation on the basis of the complex account of informative signs show quality of the solution comprehensible to practice put in
To work of the purposes and solved problems.

During generalisations of the gained results of observations we had been analysed the various parametres characterising a condition of the investigated class of patients without application of developed methods and means and with their application (work with use of intellectual system of a support of decision-making of the doctor-cardiologist).

4.5

<< | >>
Scientific source Kiselyov Alexey Viktorovich. HETEROGENEOUS QUALIFIERS With VIRTUAL STREAMS FOR INTELLECTUAL SYSTEMS of FORECASTING of the REPEATED HEART ATTACK of the MYOCARDIUM. The dissertation on competition of a scientific degree of a Cand.Tech.Sci. Kursk - 2019. 2019

Other medical related information Sampling of a mathematical apparatus for an estimation of efficiency of the gained results and an estimation of quality of work of mathematical models of forecasting of a heart attack of a myocardium in the rehabilitation period:

  1. application TP the HORN for an estimation of efficiency of therapeutic treatment
  2. application TP the HORN for an estimation of efficiency of surgical treatment
  3. survey of mathematical methods of forecasting, feature of use of illegible logic of decision-making at urolithic sickness
  4. Synthesis of mathematical models of forecasting of origination MKB
  5. Synthesis of mathematical models of forecasting of relapse MKB
  6. the analysis of efficiency of mathematical model on origination forecasting kamneobrazovanija at urolithic sickness
  7. 4.3 Analysis of efficiency of mathematical model of forecasting of relapse MKB
  8. 4.4. The analysis of efficiency of the device for metafilaktiki MKB
  9. Synthesis of hybrid indistinct models of forecasting of occurrence and relapses of a gangrene of the bottom extremities
  10. 4.3 Analysis of quality of work of mathematical models of forecasting of occurrence and relapse of a gangrene of the bottom extremities
  11. Main principles of construction of systems of support of the decision-making, the indicators of health intended for an estimation and professional suitability of workers of extreme trades