COMPARISON OF METHODS FOR DETECTING ABNORMAL EMISSIONS IN THE ORIGINAL DATA AND THEIR APPLICATION IN PROCESSING ARTIFICIAL VENTILATORY DATA

The article considers the problem of detecting abnormal outliers and mitigating their negative impact on the estimates of the allocated characteristics of the calculated indicators. To solve the problem, the article considers various robust methods used in practice and computational schemes for detecting abnormal outliers in the values ​​of the studied indicators based on them. A comparison is made of the efficiency of detecting abnormal outliers by various methods for samples of random variables with a normal distribution law for various options for the number and location of abnormal outliers in relation to non-anomalous values ​​using standard random variable sensors. The robust methods and schemes studied in the article are used to detect abnormal outliers in patient readings and ventilation flow parameters during artificial lung ventilation (ALV). The numerical experiments conducted, the results of which are presented in this article, showed that the most effective method for detecting abnormal outliers with an unknown variance of the main part of non-anomalous data is the modified Huber method developed at MEPhI. This method allows us to effectively identify abnormal emissions from the formed database of clinical experience in treating patients on ventilators, which makes it possible to use this method to create a stable scheme for selecting optimal values ​​of ventilation flow indicators depending on the values ​​of the indicators of the patient's current condition.

1. Smolyak S.A., Titarenko B.P. Ustojchivye metody ocenivaniya [Robust estimation methods]. Moscow, Statistika Publ., 1980. 254 р.

2. Huber P. Robastnost' v statistike [Robustness in statistics]. Moscow, Mir Publ., 1984. 303 р.

3. Arsenin V.Ya., Kryanev A.V., Tsupko-Sitnikov M.V. Primenenie robastnyh metodov dlya resheniya nekorrektnyh zadach [Application of robust methods for solving ill-posed problems]. JVМMF, 1989, vol.29, no.5, pp. 653-661.(in Russian)

4. Hampel M., Ronchetti E., Rausseu P., Stael W. Robastnost' v statistike. Podhod na osnove funkcij vliyaniya [Robustness in statistics. An approach based on influence functions]. Moscow, Mir Publ., 1989. 512 p.

5. Tsypkin Ya. Z. Informacionnaya teoriya identifikacii [Information theory of identification]. Moscow, Nauka Publ., 1995. 336 р.

6. Shurygin A.M. Prikladnaya statistika: robastnost', ocenivanie, prognoz [Applied statistics: robustness, estimation, forecast]. Moscow, Finansy i statistika Publ., 2000. 223 р.

7. Kryanev AV, Lukin GV, Udumyan DK. Metricheskij analiz i obrabotka dannyh [Metric analysis and data processing]. Moscow, Fizmatlit Publ., 2012. 308 p.

8. Ananyev E.P., Polupan A.A, Matskovsky I.V. [et al.]. Ispol'zovanie rezhima IntelliVent-ASV dlya podderzhaniya celevogo diapazona EtCO2 u pacientov s tyazheloj CHMT. [Using the IntelliVent-ASV mode to maintain the target EtCO2 range in patients with severe TBI]. Zhurnal «Voprosy nejrohirurgii» imeni N.N. Burdenko, 2017. Vol. 81 (5). Pр. 63 -68. (in Russian)

9. Artificial intelligence and machine learning show promise in cancer diagnosis and treatment. Medical Xpress. Available at: https://medicalxpress.com/news/2022-03-artificial-intelligence-machine-cancer-diagnosis.html. (accessed 20.11.2024)

10. Kryanev A.V., Udumyan D.K. Metric analysis, properties and applications as a tool for interpolation. International Journal of Mathematical Analysis, 2014. Vol. 8. No. 45. Pр. 2221-2228.

11. Kryanev A.V., Udumyan D.K. Metric Analysis, Properties and Applications as a Tool for Forecasting. International Journal of Mathematical Analysis. Vol. 8. No. 60. 2014. Pр. 2971 – 2978.

12. Ivanov V.V., Kryanev A.V., Udumyan D.K., Lukin G.V. Metric Analysis Approach for Interpolation and Forecasting of Time Processes. Applied Mathematical Sciences, 2014. Vol. 8, No. 22. Pp. 1053 – 1060.

13. Klimanov S. G., Kryanev A. V., Trikozova V.A., Tsareva D.D. Vychislitel'nye skhemy vyyavleniya anomal'nyh vybrosov v znacheniyah tekushchih pokazanij pacienta pri iskusstvennoj ventilyacii legkih. [Computational Schemes for Detecting Abnormal Outliers in the Values of Current Patient Readings during Artificial Lung Ventilation]. Vestnik NIYaU MIFI, 2024, Vol. 13, No. 2, рp. 76-82. (in Russian)