Quantitative Risk Estimation Model

2021 ◽  
Vol 9 (5) ◽  
pp. 538
Author(s):  
Jinwan Park ◽  
Jung-Sik Jeong

According to the statistics of maritime collision accidents over the last five years (2016–2020), 95% of the total maritime collision accidents are caused by human factors. Machine learning algorithms are an emerging approach in judging the risk of collision among vessels and supporting reliable decision-making prior to any behaviors for collision avoidance. As the result, it can be a good method to reduce errors caused by navigators’ carelessness. This article aims to propose an enhanced machine learning method to estimate ship collision risk and to support more reliable decision-making for ship collision risk. In order to estimate the ship collision risk, the conventional support vector machine (SVM) was applied. Regardless of the advantage of the SVM to resolve the uncertainty problem by using the collected ships’ parameters, it has inherent weak points. In this study, the relevance vector machine (RVM), which can present reliable probabilistic results based on Bayesian theory, was applied to estimate the collision risk. The proposed method was compared with the results of applying the SVM. It showed that the estimation model using RVM is more accurate and efficient than the model using SVM. We expect to support the reasonable decision-making of the navigator through more accurate risk estimation, thus allowing early evasive actions.


2019 ◽  
Vol 40 (Supplement_1) ◽  
Author(s):  
H Odesjo ◽  
S Bjorck ◽  
P Hjerpe ◽  
K Manhem ◽  
A Rosengren ◽  
...  

Abstract Introduction The preventive effect of lipid lowering treatment in secondary prevention after coronary heart disease (CHD) is well documented. In 2015, regional guidelines recommend an LDL cholesterol of ≤1.8 mmol/L for patients with established CHD but the adherence to these guidelines is low. Purpose Our aim was to predict potential reductions in cardiovascular disease (CVD) events defined as acute myocardial infarction or stroke if patients: 1) with low-dose/less potent or no statin were treated with Atorvastatin 80 mg, or 2) all reached LDL ≤1.8 mmol/L. Methods In total, 37 120 patients with established CHD in a primary care regional register 2015 were studied. Predicted number of CVD events were calculated with actual treatment, with improved treatment and with lowered LDL. For risk estimation we used data from a Cox Proportional Hazards risk estimation model based on patients from 2010 (n=52 042) in combination with data from the literature on effect of statin treatment and LDL reduction. A risk reduction of 22% for CVD events per 1 mmol/L reduction in LDL was used in our model. The risk prediction model included age, sex, diabetes mellitus, a history of heart failure and/or atrial fibrillations, treatment with acetylic salicylic acid and stroke or AMI past year. Smoking and BMI were excluded due to missing data but sensitivity analysis has shown only small differences in results. Results In total, 18% of included patients reached LDL ≤1.8 mmol/L and 32% had no statin treatment. Based on actual LDL levels and treatments, the predicted number of CVD events over 5 years was 9209/37120. If all patients with no statin or less potent statin treatment had been given atorvastatin 80 mg this would lead to a reduction of CVD events by 14% (7901 vs 9209). The largest gain, 33% reduction, occurred when adding statins to patients without previous treatment (1970 vs 2937). Furthermore, if all patients were to reach LDL ≤1.8 mmol/L the predicted number of events would be reduced by 18% (7577 vs 9209). Conclusion There is a substantial potential to reduce the number of CVD events in the large population of patients with established CHD in primary care by improved adherence to lipid treatment guidelines. Acknowledgement/Funding Närhälsan R&D Health Care, R&D Centre Gothenburg and Södra Bohuslän. the Swedish state under the Agreement concerning research and education of doctor


2012 ◽  
Vol 94 (6) ◽  
pp. 2091-2095 ◽  
Author(s):  
Andrzej Kansy ◽  
Jeffrey P. Jacobs ◽  
Andrzej Pastuszko ◽  
Małgorzata Mirkowicz-Małek ◽  
Małgorzata Manowska ◽  
...  

2015 ◽  
Vol 61 (2) ◽  
pp. 507-516 ◽  
Author(s):  
Hyunsoo Chung ◽  
Young Eun Chon ◽  
Seung Up Kim ◽  
Sang Kil Lee ◽  
Kyu Sik Jung ◽  
...  

2021 ◽  
Vol 87 (11) ◽  
pp. 70-80
Author(s):  
A. I. Orlov

We define risk as an unwanted opportunity and divide risk theory into three stages — risk analysis, risk estimation, risk management. Safety and risk are directly related to each other, being like a «mirror image» of each other which necessitates developing both the general theory of risk and particular theories of risk in specific areas. General risk theory allows for a uniform approach to the analysis, estimation and management of risks in specific situations. Currently, three main approaches to accounting for the uncertainty and describing risks are used — probabilistic and statistical approach, fuzzy sets, and the approach based on interval mathematics. The methods of risk estimation primarily based on probabilistic and statistical models are considered. The mathematical apparatus for estimating and managing risks is based on nonparametric formulations, limit relations, and multi-criteria optimization. Asymptotic nonparametric point estimates and confidence limits for the probability of a risk event are constructed on the base of binomial distribution and the Poisson distribution. Rules for testing statistical hypotheses regarding the equality (or difference) of two probabilities of risk events are proposed. An additive-multiplicative risk estimation model based on a hierarchical risk system based on a three-level risk system has become widespread: private risks — group risks — final risk. For this model, the role of expert estimation is revealed. The prospects of using (in the future) the theory of fuzzy sets are shown. The article deals with the main components of the mathematical apparatus of the theory of risks, in particular, the mathematical support of private theories of risks related to the quality management, innovations and investments. The simplest risk assessment in a probabilistic-statistical model is the product of the probability of a risk event and the mathematical expectation of the accidental damage. Mathematical and instrumental methods for studying global economic and environmental risks are discussed.


2012 ◽  
Vol 187 (4S) ◽  
Author(s):  
Ryo Takata ◽  
Shusuke Akamatsu ◽  
Hidewaki Nakagawa ◽  
Atsushi Takahashi ◽  
Nguyen Ha ◽  
...  

Sign in / Sign up

Export Citation Format

Share Document