repository
인공신경망을 이용한 위암 수술후 합병증의 예측
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | 한상욱 | - |
| dc.contributor.author | 최룡해 | - |
| dc.date.accessioned | 2019-10-21T07:30:21Z | - |
| dc.date.available | 2019-10-21T07:30:21Z | - |
| dc.date.issued | 2017-08 | - |
| dc.identifier.other | 25816 | - |
| dc.identifier.uri | https://dspace.ajou.ac.kr/handle/2018.oak/19054 | - |
| dc.description | 학위논문(박사)--아주대학교 일반대학원 :의학과,2017. 8 | - |
| dc.description.tableofcontents | Ⅰ. INTRODUCTION 1 Ⅱ. METHODS 3 A. Patients’ inclusion criteria and characteristics 3 B. Data correction 4 C. Prognosis factor selection 6 D. Selection of artificial intelligence algorithms 6 E. Interval validation 8 F. Statistical analysis 8 Ⅲ. RESULTS 9 Ⅳ. DISCUSSION 20 Ⅴ. CONCLUSION 24 REFERENCES 25 | - |
| dc.language.iso | eng | - |
| dc.publisher | The Graduate School, Ajou University | - |
| dc.rights | 아주대학교 논문은 저작권에 의해 보호받습니다. | - |
| dc.title | 인공신경망을 이용한 위암 수술후 합병증의 예측 | - |
| dc.title.alternative | Prediction of postoperative complication after gastric cancer surgery using artificial neural network | - |
| dc.type | Thesis | - |
| dc.contributor.affiliation | 아주대학교 일반대학원 | - |
| dc.contributor.alternativeName | CUI LONGHAI | - |
| dc.contributor.department | 일반대학원 의학과 | - |
| dc.date.awarded | 2017. 8 | - |
| dc.description.degree | Doctoral | - |
| dc.identifier.localId | 788480 | - |
| dc.identifier.url | http://dcoll.ajou.ac.kr:9080/dcollection/jsp/common/DcLoOrgPer.jsp?sItemId=000000025816 | - |
| dc.subject.keyword | Artificial neuronal network | - |
| dc.subject.keyword | Gastric carcinoma | - |
| dc.subject.keyword | Postoperative complication | - |
| dc.subject.keyword | Gastrectomy | - |
| dc.description.alternativeAbstract | Background: Incidence of gastric cancer is very high in Asia, and then gastrectomy and perigastric lymphadenectomy is commonly performed surgical procedure. An artificial neuronal network (ANN) had been applied to predict survivals or postoperative complications in some cancers and it showed a higher level of accuracy compared with the conventional linear regression analysis. However, it has not been validated whether this prediction model can be applied in the far and away time periods from the learning time, thus this study was conducted to evaluate the feasibility of previously developed ANN model for predicting severe postoperative complications after gastric cancer surgery in the recent year data. Methods: From March to September 2010, the data of 186 patients who underwent curative gastric cancer surgery were prospectively collected for 42 prognostic variables for postoperative complications. A three-layer ANN model was constructed with Clementine 12.0 and trained by a three-layer feed forward neural network using back-propagation learning algorithm. Then, the trained model was applied to the data between 2011 (n=389), 2012 (n=313), 2013 (n=356), 2014 (n=350) and 2015 (n=329) for validating its accuracy in prediction of severe postoperative complications. Results: For the development of an ANN model, 70% of 2010 data were used in training and 30% of 2010 data were used in the internal validation. In this model, the postoperative complication could be predicted by 20 top scored variables with an accuracy of 83.8% and an area under the curve (AUC) of 0.873, which was higher than that of POSSUM scoring system (AUC of 0.637). In the interval validation using the recent year data, the accuracy rate was 82.9% in 2011, 79.5% in 2012, 75.8% in 2013, 67.8% in 2014 data and 86.7% in 2015 data; however, AUC was 0.600 in 2011, 0.650 in 2012, 0.580 in 2013, 0.620 in 2014 data and 0.540 in 2015 data, respectively; it tended to decrease over the years. Conclusions: The ANN model for predicting severe postoperative complication after gastric cancer surgery is a fair tool with a high level of accuracy and diagnostic value compared to that of linear regression analysis. The ANN model would be new predictive model in the field of upper gastrointestinal surgery. | - |
- Appears in Collections:
- Graduate School of Ajou University > Department of Medicine > 4. Theses(Ph.D)
- Files in This Item:
- There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.
-
- License
- STATISTICS
- Total Visit :3,551,967
- Total Download :1,795
- Today View :14,332
