• Korean Journal of Radiology
• /
• 제22권10호
• /
• pp.1697-1707
• /
• 2021

The recent introduction of various high-dimensional modeling methods, such as radiomics and deep learning, has created a much greater diversity in modeling approaches for survival prediction (or, more generally, time-to-event prediction). The newness of the recent modeling approaches and unfamiliarity with the model outputs may confuse some researchers and practitioners about the evaluation of the performance of such models. Methodological literacy to critically appraise the performance evaluation of the models and, ideally, the ability to conduct such an evaluation would be needed for those who want to develop models or apply them in practice. This article intends to provide intuitive, conceptual, and practical explanations of the statistical methods for evaluating the performance of survival prediction models with minimal usage of mathematical descriptions. It covers from conventional to deep learning methods, and emphasis has been placed on recent modeling approaches. This review article includes straightforward explanations of C indices (Harrell's C index, etc.), time-dependent receiver operating characteristic curve analysis, calibration plot, other methods for evaluating the calibration performance, and Brier score.

• 벤처창업연구
• /
• 제19권5호
• /
• pp.1-11
• /
• 2024

온라인 리뷰는 시장 내에서의 기업의 가치를 평가하는 데 있어 중요한 역할을 하며, 기업의 수익에 큰 영향을 미치는 요인 중 하나이다. 따라서 온라인 리뷰의 감성 분석 지표는 사업의 성공을 예측할 수 있는 중요한 지표 중 하나이다. 본 연구에서는 대표적인 온라인 리뷰 플랫폼 중의 하나인 Yelp 플랫폼에 있는 레스토랑 리뷰 텍스트를 연구대상으로 선정하였고, Yelp Open Dataset에서 제공하는 리뷰 데이터 세트를 활용하였다. 본 연구에서는 레스토랑 리뷰의 Polarity Prediction을 위해 Logistic Regression, SVM, Random Forest, Gradient Boosting Machine(GBM), XGBoost, LightGBM 총 6가지 머신러닝 알고리즘을 사용하여 연구를 진행하였다. 각 모델의 성능평가 결과, Logistic Regression, SVM, LightGBM 알고리즘이 0.91로 가장 정확도가 높게 나타났다. 본 연구는 비정형화된 형태로 작성된 텍스트의 리뷰 데이터를 정량화하여 평점으로 예측할 수 있도록 하여 스타트업을 포함한 기업이 고객 피드백을 효과적으로 분석할 수 있도록 한다는 점에서 공헌점이 있다, 나아가 비즈니스 운영자들이 소비자 행동을 예측하고, 마케팅 전략 수립에 활용할 수 있는 유용한 인사이트를 제공할 수 있을 것으로 기대된다.

• 정보처리학회논문지:컴퓨터 및 통신 시스템
• /
• 제9권1호
• /
• pp.17-24
• /
• 2020

본 논문에서는 합성곱 신경망 기반의 영화 평점 예측 구조를 제안한다. 제안하는 구조는 문장 분류을 위하 고안된 TextCNN를 세 가지 측면에서 확장하였다. 첫 번째로 문자 임베딩을 이용하여 단어의 다양한 변형들을 처리할 수 있다. 두 번째로 주목 메커니즘을 적용하여 중요한 특징을 더욱 부각하였다. 세 번째로 활성 함수의 출력을 1-10 사이의 평점으로 만드는 점수 함수를 제안하였다. 제안하는 영화 평점 예측 구조를 평가하기 위해서 영화 리뷰 데이터를 이용하여 평가해 본 결과 기존의 방법을 사용했을 때보다 더욱 낮은 MSE를 확인하였다. 이는 제안하는 영화 평점 예측 구조의 우수성을 보여 주었다.

• 디지털산업정보학회논문지
• /
• 제15권1호
• /
• pp.87-97
• /
• 2019

Reaction of people is importantly considered about specific case as a social network service grows. In the previous research on analysis of social network service, they predicted tendency of interesting topic by giving scores to sentences written by user. Based on previous study we proceeded research of sentiment analysis for social network service's sentences, which predict the result as positive or negative for movie reviews. In this study, we used movie review to get high accuracy. We classify the movie review into positive or negative based on the score for learning. Also, we performed embedding and morpheme analysis on movie review. We could predict learning result as positive or negative with a number 0 and 1 by applying the model based on learning result to social network service. Experimental result show accuracy of about 80% in predicting sentence as positive or negative.

• 한국컴퓨터정보학회논문지
• /
• 제28권12호
• /
• pp.259-266
• /
• 2023

본 논문에서는 속성기반 오피니언 마이닝(ABOM)을 적용한 협업 필터링의 정확도 성능을 개선할 수 있는 알고리즘을 제안한다. 실험을 위해 국내 스마트폰 사용자의 스마트폰 앱에 대한 총 1,227건의 온라인 소비자 리뷰 데이터가 분석에 사용되었다. KKMA(꼬꼬마)분석기를 이용하여 형태소 분석 및 KOSAC를 사용하여 감성어 분석 후 LDA 토픽 모델링을 사용하여 속성 추출한 가중치 값을 부여한 리뷰별로 토픽 모델링 결과를 이용하여 협업필터링의 평점과 감성스코어의 평점을 합산한 평균값 정확도 오차를 계산한 통계모형 성능 평가인 MAE, MAPE, RMSE를 사용하였다. 실험을 통해 추천 알고리즘 중 전통적인 협업필터링과 LDA 속성 추출과 감성분석을 결합한 속성기반 오피니언 마이닝(Aspect-Based Opinion Mining, ABOM) 기법을 결합하여 온라인 고객의 앱 평점(APP_Score) 대한 정확도를 예측하였다. 분석 결과 전통적인 협업필터링을 구현한 평점의 정확도 보다 속성기반 오피니언 마이닝 CF를 적용한 평점의 예측 정확도가 더 우수한 것으로 나타났다.

• 대한안전경영과학회지
• /
• 제15권2호
• /
• pp.243-253
• /
• 2013

This research paper introduces the application and implementation of medical decision metrics that classifies medical decision-making into four different metrics using statistical diagnostic tools, such as confusion matrix, normal distribution, Bayesian prediction and Receiver Operating Curve(ROC). In this study, the metrics are developed based on cross-section study, cohort study and case-control study done by systematic literature review and reformulated the structure of type I error, type II error, confidence level and power of detection. The study proposed implementation strategies for 10 quality improvement activities via 14 medical decision metrics which consider specificity and sensitivity in terms of ${\alpha}$ and ${\beta}$. Examples of ROC implication are depicted in this paper with a useful guidelines to implement a continuous quality improvement, not only in a variable acceptance sampling in Quality Control(QC) but also in a supplier grading score chart in Supplier Chain Management(SCM) quality. This research paper is the first to apply and implement medical decision-making tools as quality improvement activities. These proposed models will help quality practitioners to enhance the process and product quality level.

• Journal of Korean Neurosurgical Society
• /
• 제30권8호
• /
• pp.970-975
• /
• 2001

Objective : By conducing a review of clinical outcomes for patients with aneurysm treated using current microneurosurgical techniques and intensive care unit management, we speculated that grading systems based only on clinical condition or CT finding after admission failed to provide a significant stratification of outcome between individual grades of patients, because these systems did not include the factor for postoperative vasospasm. We hypothesized that postoperative blood flow velocity could have a significant impact on outcome prediction for patients surgically treated for intracranial aneurysms. Methods : We conducted a analysis on patient- and lesion-specific factors that might have been associated with outcome in a series of 55 aneurysm operations performed with measurements of blood-flow velocity with transcranial Doppler ultrasonography(TCD). In the new scale(NS) score system, 1 point is assigned additionally for the case with Hunt and Hess(H-H)/World Federation of Neurological Surgeons(WFNS) Grade IV or V, Fisher Scale(FS) score 3 or 4, aneurysm size greater than 10mm, patient age older than 60 years, blood-flow velocity higher than 120cm/sec, and posterior circulation lesion. By adding the total points, a 6-point scale score(score 0-6) is obtained. Results : Age of patient, size of aneurysm, clinical condition(H-H grade and WFNS), FS score, and blood flow velocity(TCD 1day after operation) were independently and strongly associated with long-term outcome. When NS scores were applied to 55 patients with at least 6 months follow-up, the correlation of individual scores with outcome was strongly validated the retrospective findings. Conclusion : It was speculated that TCD could be used to assess postoperative vasospasm and to monitor noninvasively the patients with aneurysmal SAH. This NS score system is easy to apply, divide patients into groups with different outcome, and is comprehensive, allowing for more accurate prediction of surgical outcome.

• 한국IT서비스학회지
• /
• 제20권1호
• /
• pp.131-143
• /
• 2021

Commercial transactions, one of the pillars of the capitalist economy, are occurring countless times every day, especially small and medium-sized businesses. However, small and medium-sized enterprises are bound to be the legal underdogs in contracts for commercial transactions and do not receive legal support for contracts for fair and legitimate commercial transactions. When subcontracting contracts are concluded among small and medium-sized enterprises, 58.2% of them do not apply standard contracts and sign contracts that have not undergone legal review. In order to support small and medium-sized enterprises' fair and legitimate contracts, small and medium-sized enterprises can be protected from legal threats if they can reduce the risk of signing contracts by analyzing various risks in the contract and analyzing and informing them of toxic clauses and omitted contracts in advance. We propose a risk prediction model for the machine reading-based legal contract to minimize legal damage to small and medium-sized business owners in the legal blind spots. We have established our own set of legal questions and answers based on the legal data disclosed for the purpose of building a model specialized in legal contracts. Quantitative verification was carried out through indicators such as EM and F1 Score by applying pine tuning and hostile learning to pre-learned machine reading models. The highest F1 score was 87.93, with an EM value of 72.41.

• 대한화상학회지
• /
• 제23권1호
• /
• pp.1-6
• /
• 2020

Purpose: PL, creatinine and urine output are biomarkers of the suitability and prognosis of fluid therapy in severe burn patients. The purpose of this study is to evaluate the usefulness of predicting mortality by biomarkers and its change during initial fluid therapy for severe burn patients. Methods: A retrograde review was performed on 733 patients from January 2014 to December 2018 who were admitted as severe burn patients to our burn intensive care unit (BICU). Plasma lactate, serum creatinine and urine output were measured at the time of admission to the BICU and after 48 hours. ABSI score, Hangang score, APACHEII, revised Baux index and TBSA were collected after admission. Results: 733 patients were enrolled. PL was the most useful indicators for predicting mortality in burn patients at the time of admission (AUC: 0.813) and after 48 hours (AUC: 0.698). On the other hand, mortality prediction from initial fluid therapy for 48 hours showed different results. Only creatinine showed statistical differences (P<0.05) in mortality prediction. But there were no statistical differences in mortality prediction with PL and UO (P>0.05). Conclusion: In this study, PL was most useful predictor among biomarkers for predicting mortality. Improvement in creatinine levels during the first 48 hours is associated with improved mortality. Therefore, efforts are needed to improve creatinine levels.

• Journal of Hospice and Palliative Care
• /
• 제26권1호
• /
• pp.1-6
• /
• 2023

This paper aimed to summarize the current situation of prognostication for patients with an expected survival of weeks or months, and to clarify future research priorities. Prognostic information is essential for patients, their families, and medical professionals to make end-of-life decisions. The clinician's prediction of survival is often used, but this may be inaccurate and optimistic. Many prognostic tools, such as the Palliative Performance Scale, Palliative Prognostic Index, Palliative Prognostic Score, and Prognosis in Palliative Care Study, have been developed and validated to reduce the inaccuracy of the clinician's prediction of survival. To date, there is no consensus on the most appropriate method of comparing tools that use different formats to predict survival. Therefore, the feasibility of using prognostic scales in clinical practice and the information wanted by the end users can determine the appropriate prognostic tool to use. We propose four major themes for further prognostication research: (1) functional prognosis, (2) outcomes of prognostic communication, (3) artificial intelligence, and (4) education for clinicians.