• Nuclear Engineering and Technology
• /
• 제54권8호
• /
• pp.3027-3033
• /
• 2022

Background: In this study, various types of deep-learning models for predicting in vitro radiosensitivity from gene-expression profiling were compared. Methods: The clonogenic surviving fractions at 2 Gy from previous publications and microarray gene-expression data from the National Cancer Institute-60 cell lines were used to measure the radiosensitivity. Seven different prediction models including three distinct multi-layered perceptrons (MLP), four different convolutional neural networks (CNN) were compared. Folded cross-validation was applied to train and evaluate model performance. The criteria for correct prediction were absolute error < 0.02 or relative error < 10%. The models were compared in terms of prediction accuracy, training time per epoch, training fluctuations, and required calculation resources. Results: The strength of MLP-based models was their fast initial convergence and short training time per epoch. They represented significantly different prediction accuracy depending on the model configuration. The CNN-based models showed relatively high prediction accuracy, low training fluctuations, and a relatively small increase in the memory requirement as the model deepens. Conclusion: Our findings suggest that a CNN-based model with moderate depth would be appropriate when the prediction accuracy is important, and a shallow MLP-based model can be recommended when either the training resources or time are limited.

• Asian Pacific Journal of Cancer Prevention
• /
• 제16권17호
• /
• pp.7463-7466
• /
• 2015

Aims: To analyse the predictors of recurrence, disease free survival and overall survival in cases with endometrial cancer. Materials and Methods: A total of 152 women diagnosed with endometrial cancer were screened using a prospectively collected database including age, smoking history, menopausal status, body mass index, CA125, systemic disorders, tumor histology, tumor grade, lymphovascular space invasion, tumor diameter, cervical involvement, myometrial invasion, adnexal metastases, positive cytology, serosal involvement, other pelvic metastases, type of surgery, fertility sparing approach to assess their ability to predict recurrence, disease free survival and overall survival. Results: In ROC analyses tumor diameter was a significant predictor of recurrence (AUC:0.771, P<0.001). The optimal cut off value was 3.75 with 82% sensitivity and 63% specificity. In correlation analyses tumor grade (r=0.267, p=0.001), tumor diameter (r=0.297, p<0.001) and the serosal involvement (r=0.464, p<0.001) were found to significantly correlate with the recurrence. In Cox regression analyses when some different combinations of variables included in the model which are found to be significantly associated with the presence of recurrence, tumor diameter was found to be a significant confounder for disease free survival (OR=1.2(95 CI,1.016-1.394, P=0.031). On Cox regression for overall survival only serosal involvement was found to be a significant predictor (OR=20.8 (95 % CI 2.4-179.2, P=0.006). In univariate analysis of tumor diameter > 3.75 cm and the recurrence, there was 14 (21.9 %) cases with recurrence in group with high tumor diameter where as only 3 (3.4 %) cases group with smaller tumor size (Odds ratio:7.9 (95 %CI 2.2-28.9, p<0.001). Conclusions: Although most of the significantly correlated variables are part of the FIGO staging, tumor diameter was also found to be predictor for recurrence with higher values than generally accepted.

• Journal of the Korean Data and Information Science Society
• /
• 제20권1호
• /
• pp.203-209
• /
• 2009

Frailty models are useful for the analysis of correlated and/or heterogeneous survival data. However, the inferences of fixed parameters, rather than random effects, have been mainly studied. The prediction (or estimation) of random effects is also practically useful to investigate the heterogeneity of the hospital or patient effects. In this paper we propose how to extend the prediction method for random effects in HGLMs (hierarchical generalized linear models) to log-normal semiparametric frailty models with nonparametric baseline hazard. The proposed method is demonstrated by a simulation study.

• Nuclear Engineering and Technology
• /
• 제54권4호
• /
• pp.1439-1448
• /
• 2022

Background: We investigated the feasibility of in vitro radiosensitivity prediction with gene expression using deep learning. Methods: A microarray gene expression of the National Cancer Institute-60 (NCI-60) panel was acquired from the Gene Expression Omnibus. The clonogenic surviving fractions at an absorbed dose of 2 Gy (SF2) from previous publications were used to measure in vitro radiosensitivity. The radiosensitivity prediction model was based on the convolutional neural network. The 6-fold cross-validation (CV) was applied to train and validate the model. Then, the leave-one-out cross-validation (LOOCV) was applied by using the large-errored samples as a validation set, to determine whether the error was from the high bias of the folded CV. The criteria for correct prediction were defined as an absolute error<0.01 or a relative error<10%. Results: Of the 174 triplicated samples of NCI-60, 171 samples were correctly predicted with the folded CV. Through an additional LOOCV, one more sample was correctly predicted, representing a prediction accuracy of 98.85% (172 out of 174 samples). The average relative error and absolute errors of 172 correctly predicted samples were 1.351±1.875% and 0.00596±0.00638, respectively. Conclusion: We demonstrated the feasibility of a deep learning-based in vitro radiosensitivity prediction using gene expression.

• 한국국방경영분석학회지
• /
• 제17권1호
• /
• pp.159-176
• /
• 1991

The probability of target survival is the most important factor in the target assignment, Most of the studies about it have assumed the case of one target and ane weapon type. Therefore, they can not be applied to the real situation. In this paper. the quantity and type of enemy assets of the friendly force are considered simultaneously. Considered defense type is the coordinated defense with no impact point prediction. The objective function is to minimize the expected total survival value of targets which are scattered in the defense area. The rules of aircraft assignment are as follows : first, classify targets into several groups, each of those has the same desired damage level secondly. select the critical group which has the least survival value in accordance with the additional aircraft assignment, and finally. assign the same number of attack assets against each target in the critical group. In this paper, the attack assets, the escort assets, and the defense assets are considered. The model is useful to not only the simple aircraft assignment problem but also the complicated wargame models.

• 디지털산업정보학회논문지
• /
• 제17권1호
• /
• pp.109-115
• /
• 2021

The number of freshmen at universities is decreasing due to the recent decline in the school-age population, and the survival of many universities is threatened. To overcome this situation, universities are seeking ways to use big data within the school to improve the quality of education. A study on the prediction of dropout students is a representative case of using big data in universities. The dropout prediction can prepare a systematic management plan by identifying students who will drop out of school due to reasons such as dropout or expulsion. In the case of actual on-campus data, a large number of missing values are included because it is collected and managed by various departments. For this reason, it is necessary to construct a model by effectively reflecting the missing values. In this study, we propose a university student dropout prediction model based on eXtreme Gradient Boost that can be applied to data with many missing values and shows high performance. In order to examine the practical applicability of the proposed model, an experiment was performed using data from C University in Chungbuk. As a result of the experiment, the prediction performance of the proposed model was found to be excellent. The management strategy of dropout students can be established through the prediction results of the model proposed in this paper.

• Asian Pacific Journal of Cancer Prevention
• /
• 제16권12호
• /
• pp.5081-5084
• /
• 2015

Background: Breast cancer is a malignant tumor that starts from cells of the breast and is seen mainly in women. It's the most common cancer in women worldwide and is a major threat to health. The purpose of this study was to fit a Cox proportional hazards model for prediction and determination of years of survival in Iranian patients. Materials and Methods: A total of 366 patients with breast cancer in the Cancer Research Center were included in the study. A Cox proportional hazard model was used with variables such as tumor grade, number of removed positive lymph nodes, human epidermal growth factor receptor 2 (HER2) expression and several other variables. Kaplan-Meier curves were plotted and multi-years of survival were evaluated. Results: The mean age of patients was 48.1 years. Consumption of fatty foods (p=0.033), recurrence (p<0.001), tumor grade (p=0.046) and age (p=0.017) were significant variables. The overall 1- year, 3-year and 5-year survival rates were found to be 93%, 75% and 52%. Conclusions: Use of covariates and the Cox proportional hazard model are effective in predicting the survival of individuals and this model distinguished 4 effective factors in the survival of patients.

• Asian Pacific Journal of Cancer Prevention
• /
• 제13권11호
• /
• pp.5715-5719
• /
• 2012

The relation between cyclooxygenase enzymes and E-cadherin, along with the roles of these markers in the prediction of survival in optimally cytoreduced serous ovarian cancer patients was investigated. Individuals who underwent primary staging surgery and achieved optimal cytoreduction (largest residual tumor volume <1 cm) constituted the study population. Specimens of 32 cases were immunohistochemically examined for cyclooxygenase-1, cyclooxygenase-2, and E-cadherin. Two could not be evaluated for E-cadherin and cyclooxygenase-1. Overall, 14/30, 19/30, and 15/32 cases were positive for E-cadherin, cyclooxygenase-1, and cyclooxygenase-2, respectively. The expressions of E-cadherin and cyclooxygenase-2 were inversely correlated (p:0.02). E-cadherin expression was related with favorable survival (p<0.001). The relation between the expression of cyclooxygenase enzymes and poor survival did not reach statistical significance. On multivariate analysis, E-cadherin appeared as an independent prognostic factor for survival. In conclusion, E-cadherin expression is strongly linked with favorable survival. E-cadherin and cyclooxygenase 2 may interact with each other during the carcinogenesis-invasion process. Further studies clarifying the relation between E-cadherin and cyclooxygenase enzymes may lead to new preventive and therapeutic targets in ovarian cancer.

• Asian Pacific Journal of Cancer Prevention
• /
• 제14권5호
• /
• pp.3331-3335
• /
• 2013

Background: The incidence of prostate cancer, one of the most common cancers in elderly men, is increasing annually in Thailand. Matrix metalloproteinase 11 (MMP-11) is a member of the extracellular matrix metalloproteases which has been associated with human tumor progression and clinical outcome. Aim: To quantify MMP-11 expression in prostatic adenocarcinoma tissues and to determine whether its overexpression correlates with survival outcome, and to assess its potential as a new prognostic marker. Materials and Methods: Expression of MMP-11 was analyzed using immunohistochemistry in 103 Thai patients with prostatic adenocarcinoma. Overall survival was analyzed using Kaplan-Meier methods and Cox regression models. Results: Immunoreactivity of MMP-11 was seen in the stroma of prostatic adenocarcinoma tissue samples, high expression being significantly correlated with poor differentiation in Gleason grading, pathologic tumor stage 4 (pT4), and positive-bone metastasis (p<0.05), but not age and prostatic-specific antigen (PSA) level. Patients with high levels of MMP-11 expression demonstrated significantly shorter survival (p<0.001) when compared to those with low levels. Multivariate analysis showed that MMP-11 expression and pT stage were related with survival in prostatic adenocarcinoma [hazard ratio (HR)=0.448, 95% confidence interval (95%CI)=0.212-0.946, HR=0.333, 95%CI=0.15-0.74, respectively]. Conclusions: Expression of MMP-11 is significantly associated with survival in prostatic adenocarcinoma. High levels may potentially be used for prediction of a poor prognosis.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 추계학술대회
• /
• pp.297-299
• /
• 2022

악성 뇌교종은 예후가 매우 나쁜 질병으로 평균 생존 기간은 6개월에서 14개월 사이로 보고되어 있다. 따라서 악성 뇌교종을 가진 환자들에게는 정확한 예후 예측이 요구된다. 본 논문에서는 악성 뇌교종을 가진 환자의 예후와 연령을 동시에 예측하는 합성곱 신경망 모델을 제안한다. 악성 뇌교종의 영상 특성을 효과적으로 파악할 수 있는 네 가지 자기공명영상인 T1, T1-contrast enhanced, T2, fluid-attenuated inversion recovery 영상을 입력 데이터로 이용하였다. 예후 예측에 가장 중요한 환자의 연령을 고려함으로써 신경망 모델의 예후 예측 성능이 높아질 것으로 기대된다. 학습된 모델을 검증 데이터에 적용한 결과 환자의 예후와 연령의 피어슨 상관계수가 각각 0.1748, 0.3056으로 나타난 것을 확인하였다.