FORECASTING THE DEVELOPMENT OF RECURRENCES IN PATIENTS WITH PRIMARY AND POSTOPERATIVE VENTRICULAR HERNIAS WITH APPLICATION OF MULTIPARAMETER NEURO NETWORK CLUSTERIZATION

  • V. I. Piatnochka SHEE I. Horbachevsky Ternopil State Medical University of the Ministry of Health of Ukraine
  • I. Ya. Dzyubanovskyi SHEE I. Horbachevsky Ternopil State Medical University of the Ministry of Health of Ukraine
  • P. R. Selskyi SHEE I. Horbachevsky Ternopil State Medical University of the Ministry of Health of Ukraine
Keywords: primary hernia, postoperative ventral hernia, relapse, neural network clasterization, forecasting

Abstract

Background. Surgical treatment of patients with primary and postoperative ventral hernia remains one of the important problems in the surgery of abdominal cavity. About 60 % of patients are of working age. The results of surgical treatment of postoperative ventral hernia indicate a significant percentage of recurrence, which is 4.3-46 %. This problem is relevant especially in cases where the inappropriate method of alloplasty and mesh implant is selected and in the early postoperative period a number of general complications arise and lead to recurrence. The latter require a detailed analysis of the reasons for their occurrence and the development of individualized prescriptions before choosing the method of surgery.

Modern information technology greatly facilitates solving prediction problems. Currently, neural networks are widely used for early diagnosis.

Materials and methods. Between 2001 and 2017, an in-depth comprehensive clinical-instrumental and laboratory examination of 1419 patients with primary ventral hernia (PVH) and postoperative ventral hernia (POVH) was carried out, with further evaluation of the nature of complications in the early and late postoperative period. For more in-depth analysis and clustering of the surveyed under study, a neural network approach was used with the help of NeuroXL Classifier add-in for Microsoft Excel.

Results. In order to determine the value of the combination of changes in certain parameters for the prediction of the development of relapses of ventral hernias, a neural network clusterization of the results of the patient's examination based on the parameters of relapse, age, gender, blood group, concomitant pathology, etc. was performed. For the algorithm of neural network clustering the parameters proposed by the same program were selected. The largest number of patients with the first relapse was referred to the cluster 1, which was 8.4 %, where the highest number of patients with heart failure, chronic obstructive pulmonary disease and diabetes was noted. Most patients with relapses were found in the cluster 4, which was 27.6 %. The most commonly encountered patients with undifferentiated connective tissue dysplasia and the second blood group were also included in this cluster. At the same time, the average age in this group was the lowest and amounted 49.4 years. Analyzing the results of clustering of patients with relapses of ventral hernias by types of surgical interventions, it was determined that this cluster had the highest incidence of patients with primary ventral hernias, who underwent an onlay type surgery using lightweight mesh. The highest number of relapses (R1-R4) fell on the cluster 2 (19.1 %), which showed the highest number of women compared to the clusters 1 and 3-5.

Conclusions. Analysis of cluster portraits revealed that the risk group for relapse included of women with the first-degree obesity, as well as patients with heart failure, chronic obstructive pulmonary disease and diabetes. The presence of the second blood group, the relatively younger age (< 49.4 years), and the diagnosed syndrome of undifferentiated connective tissue dysplasia gives a reason to predict a higher risk of relapse after ventral hernia recurrences in operated patients.

References

Helgstrand, F. (2016). National results after ventral hernia repair. Dan Med J., 63(7), B5258.

Israelsson, L. A., & Millbourn, D. (2013). Prevention of incisional hernias: How to close a midline incision. Surg Clin North Am., 93(5), 1027-1040. doi: 10.1016 / j.suc.2013.06.009

Kaoutzanis, C., Leichtle, S., Mouawad, N., Welch, K. B., Lampman, R. M., Wahl, W. L., & Cleary, R. K. (2015). Risk factors for postoperative wound infections and prolonged hospitalization after ventral/incisional hernia repair. Hernia, 19(1), 113-123. doi: 10.1007/ s10029-013-1155-y.

Skipworth, J. A., Vyas, S., Uppal, L., Floyd, D., & Shankar, A. (2014). Improved outcomes in the management of high-risk incisional hernias utilizing biologic mesh and soft-tissue reconstruction: a single center experience. World J Surg., 38, 1026-1034. doi: 10.1007/s00268-013-2442-6.

Patel, P. V., & Merchant, A. M. (2014). Ventral hernia repair in the morbidly obese patient: A review of medical and surgical approaches in the literature. Bariatr Surg Pract Patient Care, 9(2), 61-65. doi.org/10.1089/ bari.2014.0008.

Bishop, C. M. (1995). Neural Networks for Pattern Recognition. Oxford : Oxford University Press, p. 504.

Martsenyuk, V. (2017). On an indirect method of exponential estimation for a neural network model with discretely distributed delays. Electronic Journal of Qualitative Theory of Differential Equations, 23, 1-16. doi: 10.14232/ejqtde.2017.1.23

Martsenyuk, V. (2017). Indirect method of exponential convergence estimation for neural network with discrete and distributed delays. Electronic Journal of Differential Equations 2017; 246: 1-12. URL: http://ejde.math. txstate.edu or http://ejde.math.unt.edu

Shepherd, A. J. (1997). Second-Order Methods for Neural Networks : Fast and Reliable Training Methods for Multi-Layer Perceptrons. Shepherd. London : Springer, p. 342.

Martsenyuk, V. P., Semenets, A. V., & Stakhanska, O. O. (2010). Neyromerezheve prohnozuvannia skladannia studentamy-medykamy litsenziinoho intehrovanoho ispytu "Krok 1" na osnovi rezultativ potochnoi uspishnosti ta semestrovoho kompleksnoho testovoho ispytu [Neuronetwork Prediction Of License Integrated Exam «Step 1» Passing Based On Current Results And Semester Complex Test Exam For Medical Students]. Medychna informatyka ta inzheneriya -Medical Informatics and engeneering, 2, 57-62 [in Ukrainian].

Published
2019-02-19
How to Cite
Piatnochka, V. I., Dzyubanovskyi, I. Y., & Selskyi, P. R. (2019). FORECASTING THE DEVELOPMENT OF RECURRENCES IN PATIENTS WITH PRIMARY AND POSTOPERATIVE VENTRICULAR HERNIAS WITH APPLICATION OF MULTIPARAMETER NEURO NETWORK CLUSTERIZATION. Medical Informatics and Engineering, (4), 41-45. https://doi.org/10.11603/mie.1996-1960.2018.4.9843
Section
Articles