Overestimation Of Risk Research Articles

Detecting underestimates of risk in VaR models

Abstract This paper explores avenues to improve VaR backtesting against violations of unconditional coverage, with emphasis on detecting underestimates of risk. I show that existing tests for unconditional coverage have an undesirable asymmetry in power, making them less (more) effective at detecting underestimates (overestimates) of risk. A set of Lagrange multiplier (LM) tests is proposed, leading to improved power against the more dangerous underestimates. Also, the use of one-sided coverage tests is revisited, with a one-sided version of the new LM test performing best. Results are supported by simulations, showing the effectiveness of the new tests in finite samples. The same issues of asymmetric power are shown to carry over to the case of conditional coverage, though it appears that finite sample issues affecting the independence component of the tests can adversely impact performance. A simple weighted test statistic is shown to help alleviate this problem, providing an avenue to exploit the advantages of the LM approach in a conditional coverage setting.

Model Risk Measures: A Review and New Proposals on Risk Forecasting

In financial decisions, model risk has been recognized as an important source of uncertainty. The revision of the Basel II suggests that financial institutions quantify and manage their model risk. Focusing on risk forecasting literature, we identify two main approaches to quantify model risk: the worst-case and loss function. The first approach includes measures with a similar structure to deviation measures, which are applied to a set of forecasts obtained by different models. For the second approach, monetary risk measures are employed under a loss or error function, from some forecasting procedure. Based on the untapped features of model risk for both approaches we suggest new proposals, which include measures to quantify upside and downside model risk, and average costs associated with risk overestimation and underestimation. We also conduct an empirical assessment of model risk measures using Value at Risk (VaR) and Expected Shortfall (ES) forecasting. We verify, using worst-case measures, that model risk tends to increase before crisis period and, according to loss function measures, model risk values increase in the crisis period. We also conclude that a model with good performance to risk forecasting (lower realized loss) does not indicate this model has lower model risk. Furthermore, we highlight insights into future research directions regarding this topic.

Probabilistic modeling for an integrated temporary acquired immunity with norovirus epidemiological data.

Integration of acquired immunity into microbial risk assessment for illness incidence is of no doubt essential for the study of susceptibility to illness. In this study, a probabilistic model was set up as dose response for infection and a mathematical derivation was carried out by integrating immunity to obtain probability of illness models. Temporary acquire immunity from epidemiology studies which includes six different Norovirus transmission scenarios such as symptomatic individuals infectious, pre- and post-symptomatic infectiousness (low and high), innate genetic resistance, genogroup 2 type 4 and those with no immune boosting by asymptomatic infection were evaluated.Simulated results on illness inflation factor as a function of dose and exposure indicated that high frequency exposures had immense immunity build up even at high dose levels; hence minimized the probability of illness. Using Norovirus transmission dynamics data, results showed, and immunity included models had a reduction of 2–6 logs of magnitude difference in disease burden for both population and individual probable illness incidence. Additionally, the magnitude order of illness for each dose response remained largely the same for all transmission scenarios; symptomatic infectiousness and no immune boosting after asymptomatic infectiousness also remained the same throughout. With integration of epidemiological data on acquired immunity into the risk assessment, more realistic results were achieved signifying an overestimation of probable risk of illness when epidemiological immunity data are not included. This finding supported the call for rigorous integration of temporary acquired immunity in dose-response in all microbial risk assessments.

The Impact of GE Lunar vs ICMR Database in Diagnosis of Osteoporosis among South Indian Subjects.

Objective:To study the effect of choosing ICMR reference values on the classification of bone mineral density in Indian patients.Design:Retrospective analysis of Dual Energy X-ray absorptiometry (DEXA) and clinical data.Patients:Totally, 316 patients aged more than 65 years attending a tertiary care hospital in South India who underwent DEXA scan were included in the study.Measurements:DEXA scan at femoral neck and lumbar spine.Results:A total of 316 patients were studied. The mean age was 61.98 ± 7.66 years. There were 46.84% females and 53.16% males. The average BMI was 26.37 ± 4.51. Of these patients, 46 had history of hip fracture (14.55%). The adoption of the ICMR normative data resulted in a significant increase in T scores in both the hip (+0.51, P < 0.05) and the spine (+1.64, P < 0.01). The adoption of ICMR normative values, resulted in reduction of osteoporosis prevalence from 26.58% to 5.06%.Conclusions:There is a clinically significant reduction in diagnosis of osteoporosis with the adoption of ICMR reference standard. Clinicians should be recommended to use raw BMD values in gm/cm2 in FRAX calculation and avoid the use of T scores, to avoid overestimation of fracture risk. If our results are replicated, the implications are enormous – Osteoporosis is currently being over diagnosed.

Prospective validation of the NCI Breast Cancer Risk Assessment Tool (Gail Model) on 40,000 Australian women

BackgroundThere is a growing interest in delivering more personalised, risk-based breast cancer screening protocols. This requires population-level validation of practical models that can stratify women into breast cancer risk groups. Few studies have evaluated the Gail model (NCI Breast Cancer Risk Assessment Tool) in a population screening setting; we validated this tool in a large, screened population.MethodsWe used data from 40,158 women aged 50–69 years (via the lifepool cohort) participating in Australia’s BreastScreen programme. We investigated the association between Gail scores and future invasive breast cancer, comparing observed and expected outcomes by Gail score ranked groups. We also used machine learning to rank Gail model input variables by importance and then assessed the incremental benefit in risk prediction obtained by adding variables in order of diminishing importance.ResultsOver a median of 4.3 years, the Gail model predicted 612 invasive breast cancers compared with 564 observed cancers (expected/observed (E/O) = 1.09, 95% confidence interval (CI) 1.00–1.18). There was good agreement across decile groups of Gail scores (χ2 = 7.1, p = 0.6) although there was some overestimation of cancer risk in the top decile of our study group (E/O = 1.65, 95% CI 1.33–2.07). Women in the highest quintile (Q5) of Gail scores had a 2.28-fold increased risk of breast cancer (95% CI 1.73–3.02, p < 0.0001) compared with the lowest quintile (Q1). Compared with the median quintile, women in Q5 had a 34% increased risk (95% CI 1.06–1.70, p = 0.014) and those in Q1 had a 41% reduced risk (95% CI 0.44–0.79, p < 0.0001). Similar patterns were observed separately for women aged 50–59 and 60–69 years. The model’s overall discrimination was modest (area under the curve (AUC) 0.59, 95% CI 0.56–0.61). A reduced Gail model excluding information on ethnicity and hyperplasia was comparable to the full Gail model in terms of correctly stratifying women into risk groups.ConclusionsThis study confirms that the Gail model (or a reduced model excluding information on hyperplasia and ethnicity) can effectively stratify a screened population aged 50–69 years according to the risk of future invasive breast cancer. This information has the potential to enable more personalised, risk-based screening strategies that aim to improve the balance of the benefits and harms of screening.

"1-in-X" bias: "1-in-X" format causes overestimation of health-related risks.

According to the "1-in-X" effect, "1-in-X" ratios (e.g., 1 in 12) trigger a higher subjective probability than do numerically equivalent "N-in-X*N" ratios (e.g., 3 in 36). Here we tested the following: (a) the effect on objective measures, (b) its consequences for decision-making, (c) whether this effect is a form of bias by measuring probability accuracy, and (d) its amplification in people with lower health literacy and numeracy. In parallel-designed experiments, 975 participants from the general adult population participated in 1 of 5 experiments following a 2(format: "1-in-X" or "N-in-X*N") × 4(scenarios) mixed design. Participants assessed the risk of contracting a disease on either a verbal probability scale (Experiment 1) or a numerical probability/frequency scale with immediate (Experiments 2-3) or delayed presentation (Experiments 4-5). Participants also made a health-related decision and completed a health literacy and numeracy scale. The "1-in-X" ratios yielded higher probability perceptions than did the "N-in-X*N" ratios and affected relevant decisions. Critically, the "1-in-X" ratios led to a larger objective overestimation of numerical probabilities than did the "N-in-X*N" ratios. People with lower levels of health literacy and numeracy were not more sensitive to the bias. Health professionals should use "1-in-X" ratios with great caution when communicating to patients, because they overestimate health risks. (PsycINFO Database Record (c) 2018 APA, all rights reserved).

Integrating Clinical, Morphological, and Molecular Data to Assess Prognosis in Patients with Primary Myelofibrosis: A Practical Approach

Current prognostic scoring systems are not able to easily integrate clinical, histological and molecular data in primary myelofibrosis (PMF) as they required additional information, mainly concerning non-driver mutations, which are available only in a very limited number of research laboratories.The present multicenter study developed a simple method to integrate, at diagnosis, the prognostic information from two well-defined prognostic scoring systems [IPSS and bone marrow fibrosis grade (GBMF)] with the molecular status (MS) concerning the so-called driver mutations to obtain an overall prognostic stratification of PMF patients applicable worldwide.The study included 401 PMF patients who were diagnosed at three Italian Hematological Centers (Milan, Monza, and Bologna) between November 1983 and December 2016 and were followed up for a minimum of 12 months. Follow-up information was updated in December 2017.An integrated score (ISC) was developed based on the sum of the scores attributed to the following variables: IPSS risk categories (low risk = 0 point; intermediate-1 risk = 1 point; intermediate-2 and high risk = 2 points), GBMF (pre-fibrotic stage = 0 point; overt fibrotic stage = 1 point), and MS (CALR type 1 = 0 point; CALR type 2 or other / JAK2V617F / MPL = 1 point; triple-negative = 3 points).An ISC ranging from 0 to 6 was obtained using the following formula: ISC = IPSS + GBMF score + MS score.Based on the ISC results, patients were divided into four prognostic categories: ISC-low risk, score = 0-1 (126 patients); ISC-intermediate-1 risk, score = 2 (142 patients); ISC-intermediate-2 risk, score = 3 (70 patients); and ISC-high risk, score = 4-6 (63 patients).Considering patients with an ISC-low risk as the reference group, the odds of death was 3.5 times higher (95% CI, 1.7-7.0) for ISC-intermediate-1 risk, 5.6 times higher (95% CI, 2.6-12.1) for ISC-intermediate-2 risk, and 10.5 times higher (95% CI, 4.8-22.6) for ISC-high risk patients.Interestingly, the ISC was significantly associated (p<0.004) with constitutional symptoms, splenomegaly, and transfusion need, and a higher frequency of cytoreductive therapy.In addition, median time from diagnosis to the development of an index event (including thrombotic events and leukemic evolution) was shorter in ISC-high risk patients than in ISC-low risk patients: 2.8 vs. 6.8 years for thrombotic events, and 3.5 vs. 4.8 years for leukemic evolution.As reported in Figure 1, the median overall survival (Kaplan-Meier estimate) was 11.9 years in ISC-intermediate-1 risk, 8.8 years in ISC-intermediate-2 risk, and 6.4 years in ISC-high risk patients (log-rank test <0.0001).The results were confirmed using the multivariate Cox model: the Hazard Ratio (HR) for death using the ISC-low risk group as reference was 4.3 (95% CI, 2.2-8.3) for the ISC-intermediate-1 risk group, 7.0 (95% CI, 3.5-14.0) for the ISC-intermediate-2 risk group, and 13.9 (95% CI, 7.1-27.3) for the ISC-high risk group.When patients were instead stratified according to the IPSS, the median overall survival was 10.8 years in IPSS-intermediate-1 risk, 6.9 years in IPSS-intermediate-2 risk, and only 4 years in IPSS-high risk patients. The HR for death was 13.0-times higher for intermediate-2 risk and 35.3-times higher for high risk than for the low IPSS risk categories. Accordingly, these results suggest that the prognostic stratification of risk according to the IPSS can be modulated by applying the ISC score, thereby preventing the overestimation of the real risk of death.The present study had two main limitations related to its retrospective nature and the inability to evaluate the impact of new drugs such as JAK1/2 inhibitors on the natural history of this disease, mainly because of the wide range of the time of diagnosis. Therefore, additional prospective studies involving other Hematological Centers and a larger number of patients are needed to draw definite conclusions.In conclusion, the comprehensive approach proposed in the present study is an improved tool for predicting mortality in PMF patients. The proposed model will allow clinicians to evaluate the mutual interactions between IPSS, GBMF, and MS to identify high-risk patients with a poor prognosis who may benefit from more aggressive treatments. In addition, the present model is applicable worldwide in the real-life clinical practice. [Display omitted] DisclosuresPalandri:Novartis: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau. Cortelezzi:roche: Consultancy; novartis: Consultancy; abbvie: Consultancy; janssen: Consultancy.

Estimating institutional physician turnover attributable to self-reported burnout and associated financial burden: a case study

BackgroundAwareness of the economic cost of physician attrition due to burnout in academic medical centers may help motivate organizational level efforts to improve physician wellbeing and reduce turnover. Our objectives are: 1) to use a recent longitudinal data as a case example to examine the associations between physician self-reported burnout, intent to leave (ITL) and actual turnover within two years, and 2) to estimate the cost of physician turnover attributable to burnout.MethodsWe used de-identified data from 472 physicians who completed a quality improvement survey conducted in 2013 at two Stanford University affiliated hospitals to assess physician wellness. To maintain the confidentially of survey responders, potentially identifiable demographic variables were not used in this analysis. A third party custodian of the data compiled turnover data in 2015 using medical staff roster. We used logistic regression to adjust for potentially confounding factors.ResultsAt baseline, 26% of physicians reported experiencing burnout and 28% reported ITL within the next 2 years. Two years later, 13% of surveyed physicians had actually left. Those who reported ITL were more than three times as likely to have left. Physicians who reported experiencing burnout were more than twice as likely to have left the institution within the two-year period (Relative Risk (RR) = 2.1; 95% CI = 1.3–3.3). After adjusting for surgical specialty, work hour categories, sleep-related impairment, anxiety, and depression in a logistic regression model, physicians who experienced burnout in 2013 had 168% higher odds (Odds Ratio = 2.68, 95% CI: 1.34–5.38) of leaving Stanford by 2015 compared to those who did not experience burnout. The estimated two-year recruitment cost incurred due to departure attributable to burnout was between $15,544,000 and $55,506,000. Risk of ITL attributable to burnout was 3.7 times risk of actual turnover attributable to burnout.ConclusionsInstitutions interested in the economic cost of turnover attributable to burnout can readily calculate this parameter using survey data linked to a subsequent indicator of departure from the institution. ITL data in cross-sectional studies can also be used with an adjustment factor to correct for overestimation of risk of intent to leave attributable to burnout.

Incidence of cardiovascular disease in familial combined hyperlipidemia: A 15-year follow-up study

Background and aimsFamilial combined hyperlipidemia (FCHL) is a complex dyslipidemia associated with premature cardiovascular disease (CVD). The present study was conducted to 1) determine the incidence of CVD in FCHL in this era of protocolled, primary prevention; and 2) examine whether cardiovascular risk estimation based on the Systemic Coronary Risk Estimation (SCORE) chart, as proposed in the 2016 ESC/EAS guidelines for the management of dyslipidemia, is justified in FCHL. MethodsFCHL patients, their normolipidemic (NL) relatives and spouses originally included in our baseline cohort in 1998–2005 (n = 596) were invited for a follow-up visit to determine the incidence of CVD, defined as (non-)fatal coronary artery disease, ischemic stroke and peripheral artery disease requiring invasive treatment. ResultsFollow-up data (median: 15 years) was acquired for 85% of the original cohort. The cumulative incidence of CVD was significantly higher in FCHL patients than in spouses (23.6% versus 4.7%; hazard ratio (HR): 5.4, 95%CI: 2.0–14.6; HR after adjustment for risk factors included in SCORE: 4.7, 95%CI: 1.6–13.8), but not in NL relatives compared to spouses (5.8% versus 4.7%). The SCORE chart tended to overestimate CVD risk in the spouses (observed [O]/expected [E] ratio:0.2, p = 0.01), but not in FCHL patients (O/E:1.3, p = 0.50). ConclusionsRisk of primary CVD is still substantially increased in FCHL patients, despite preventive measures. The overestimation of CVD risk by the SCORE chart – a nowadays frequently observed phenomenon thanks to improved primary prevention – was not seen in FCHL. These results suggest that more aggressive treatment is justified to avoid excessive CVD in FCHL.

Short-term to long-term extrapolation of lethal effects of an herbicide on the marine mysid shrimp Americamysis Bahia by use of the General Unified Threshold Model of Survival (GUTS).

Risk assessments for plant protection products and their active ingredients that are based on standard laboratory tests performed under constant exposure conditions may result in an overestimation of risks because exposure in the environment is often characterized by a few short peaks. Here, the General Unified Threshold Model of Survival (GUTS) was used to conduct a refined risk assessment for the herbicide tembotrione and its effects on the marine invertebrate Americamysis bahia, for which the standard chronic effect assessment failed. The GUTS model was first calibrated with time-to-effect and concentration-response data from 2 independent acute experiments with A. bahia. Model parameters for both toxicodynamic assumptions of stochastic death (SD) and individual tolerance (IT) were estimated with the reduced GUTS model (GUTS-RED) using the scaled internal concentration as a dose metric. Both the calibrated GUTS-RED-SD and GUTS-RED-IT models described survival dynamics well. Model validation using datasets of 2 independent chronic tests yielded robust predictions of long-term toxicity of tembotrione on A. bahia, with GUTS-RED-IT being more reliable than GUTS-RED-SD. The validated model was subsequently used to predict survival from time-variable exposure profiles, as derived from the FOrum for Co-ordination of pesticide fate models and their USe (FOCUS). Because ecotoxicological independence of peaks had not been empirically verified, the link between exposure and effects was assessed with complete exposure profiles. Effect thresholds resulting from different peak exposure concentrations and durations were determined with GUTS and directly compared with the exposure concentrations from the FOCUS surface water scenarios. The derived values were higher than the predicted FOCUS critical concentrations. Additionally, comparing the areas under the curve (AUCs) derived with GUTS for multiple peak exposure profiles to those from FOCUS revealed significant additional safety margins, demonstrating that only unrealistically high exposure concentrations would produce significant effects. In conclusion, no unacceptable effects of tembotrione on aquatic invertebrates under realistic environmental exposure conditions are expected. Integr Environ Assess Manag 2019;15:29-39. © 2018 SETAC.

Conditional Heteroskedasticity in Crypto-Asset Returns

In a recent contribution to the financial econometrics literature, Chu et al. (2017) provide the first examination of the time-series price behaviour of the most popular cryptocurrencies. However, insufficient attention was paid to correctly diagnosing the distribution of GARCH innovations. When these data issues are controlled for, their results lack robustness and may lead to either underestimation or overestimation of future risks. The main aim of this paper therefore is to provide an improved econometric specification. Particular attention is paid to correctly diagnosing the distribution of GARCH innovations by means of Kolmogorov type non-parametric tests and Khmaladze's martingale transformation. Numerical computation is carried out by implementing a Gauss-Kronrod quadrature. Parameters of GARCH models are estimated using maximum likelihood. For calculating P-values, the parametric bootstrap method is used. Further reference is made to the merits and demerits of statistical techniques presented in the related and recently published literature.

English

In a recent contribution to the financial econometrics literature, Chu et al. (2017) provide the first examination of the time-series price behaviour of the most popular cryptocurrencies. However, insufficient attention was paid to correctly diagnosing the distribution of GARCH innovations. When these data issues are controlled for, their results lack robustness and may lead to either underestimation or overestimation of future risks. The main aim of this paper therefore is to provide an improved econometric specification. Particular attention is paid to correctly diagnosing the distribution of GARCH innovations by means of Kolmogorov type non-parametric tests and Khmaladze's martingale transformation. Numerical computation is carried out by implementing a Gauss-Kronrod quadrature. Parameters of GARCH models are estimated using maximum likelihood. For calculating P-values, the parametric bootstrap method is used. Further reference is made to the merits and demerits of statistical techniques presented in the related and recently published literature.

Perceived Risk, Anxiety and Alexithymia in Sisters of Breast Cancer Patients Veena Shukla Mishra and Dhananjaya Saranath

Background: Breast cancer is the most common cancer in Indian women with an annual incidence of 144,937 cases and mortality of 70,218. The perception of cancer risk has consistently been mentioned as major factor influencing the women, who are at increased risk of inherited breast and ovarian cancer. The overestimation of cancer risk has been associated with many negative outcomes like anxiety and distress for one's self and family. Study suggests that women those who have higher perceived breast cancer risk experience greater worry and disengage coping. Aim: The purpose of the study was to examine the perception of risk of breast cancer in sisters of newly diagnosed breast cancer patients and further examine the association with anxiety and alexithymia, as compared with healthy controls without a family history of breast cancer. Methods: The participants were requested to complete questionnaires including demographic and risk perception of breast cancer. The association of risk perception was evaluated by using the State Trait Anxiety Inventory Scale and Toronto Alexithymia Scale in sisters of breast cancer patients and healthy controls. Results: Sisters of 111 breast cancer patients and 123 healthy controls completed the questionnaire. Sisters with high perceived risk showed high scores on anxiety and alexithymia scale as compared with healthy control. Regression analysis showed significant association between perceived risk and anxiety (t=2.023, P &lt; .05) and alexithymia factor difficulty in identifying feelings and total alexithymia score (t=6.787, P &lt; .000 and 3.726, P &lt; .000). Conclusion: Sisters of breast cancer patients showed significantly higher perceived risk, anxiety and alexithymia than their healthy counterparts. The sister's perception of breast cancer risk influences the anxiety and emotional experience. Our data emphasizes that medical professional should discuss risk appraisals to anxiety and emotional concerns in both breast cancer patients and their sisters to help them in coping with breast cancer and concerns in the family.

Discordance between 10-year cardiovascular risk estimates using the ACC/AHA 2013 estimator and coronary artery calcium in individuals from 5 racial/ethnic groups: Comparing MASALA and MESA

Background and aimsSouth Asian (SA) individuals are thought to represent a group that is at high-risk for atherosclerotic cardiovascular disease (ASCVD). However, the performance of the Pooled Cohort Equations (PCE) remains uncertain in SAs living in the US. We aimed to study the interplay between predicted 10-year ASCVD risk and coronary artery calcium (CAC) in SAs compared to other racial/ethnic groups. MethodsWe studied 536 SAs from the Mediators of Atherosclerosis in South Asians Living in America (MASALA) study, and 2073 Non-Hispanic Whites (NHWs), 1514 African Americans (AAs), 1254 Hispanics, and 671 Chinese Americans (CAs) from the Multi-Ethnic Study of Atherosclerosis (MESA) who were not currently on statins. We used logistic regression models to assess the association between race/ethnicity and CAC within each ASCVD risk stratum. ResultsSAs at low and at intermediate estimated ASCVD risk were more likely to have CAC = 0 compared to NHWs, while SAs at high risk had a similar CAC burden to NHWs. For example, intermediate-risk SAs had a 73% higher odds of CAC = 0 compared to NHWs (95% 1.00–2.99), while high-risk SAs were equally likely to have CAC = 0 (OR 0.95, 95% CI 0.65–1.38) and CAC >100 (OR 0.86, 95% CI 0.61–1.22). ConclusionsOur results suggest that the extent of ASCVD risk overestimation using the PCEs may be even greater among SAs considered at low and intermediate risk than among NHWs. Studies with incident ASCVD events are required to validate and/or recalibrate current ASCVD risk prediction tools in this group.

An Insurance Model for Risk Management of Process Facilities.

Most existing risk management models for process industries do not consider the effect of insurance coverage, which results in an overestimation of overall risk. A model is presented in this article to study the effect of insurance coverage of health, safety, environmental, and business risks. The effect of insurance recovery is modeled through the application of adjustment factors by considering the stochastic factors affecting insurance recovery. The insurance contract's conditions, deductibles, and policy limits are considered in developing the insurance recovery adjustment factors. Copula functions and Monte Carlo simulations are used to develop the distribution of the aggregate loss by considering the dependence among loss classes. A case study is used to demonstrate both the practical application of the proposed insurance model to improve management decisions, and the mitigating effect of insurance to minimize the residual risk.

Incidence, Dependence Structure of Disease, and Rate Making for Health Insurance

In order to analyze the two goals under the national strategy of “Healthy China”, this paper attempts to solve the problem of coverage rate and guarantee level of health insurance, as well as the rational allocation of full life cycle health insurance resources. This paper uses pair copula to model the dependence of different disease incidence and proposes an actuarial model for rate making in health insurance based on the dependence captured by pair copula. These are far more accurate than any other model and more proper for covering a basket of several different diseases. The data for the paper was drawn from the experience incidence table of major diseases (malignant tumor, acute myocardial infarction, and stroke sequelae) from the ages 0-65 years in the Chinese life insurance industry. Extending the hypothesis of independence in actuarial modeling, the authors comprehensively use a hierarchical copula theory to extract the dependence structure of risk variable in insurance. The classification rate making technology and survival analysis method in traditional actuarial pricing were also considered. This paper applied the generalized linear model, which is commonly used in nonlife insurance pricing for empirical study of health insurance rate making. The authors discovered that the incidence of major diseases and the single premium rate calculated by the generalized linear model under HAC dependence structure were both significantly different from that calculated by the Manchester United method without dependency. The authors also stated that the rate based on the generalized linear model under HAC dependence structure was a bit different from that without dependency but both were generally the same as that of Care Expert in PICC Health. The underestimation or overestimation of systematic risks and the distortion of the rate system can be eliminated if we combine risk dependence into modeling.

An Optimized DBN-based Coronary Heart Disease Risk Prediction

Coronary Heart Disease (CHD) is the world’s leading cause of death according to a World Health Organization (WHO) report. Despite the evolution of modern medical technology, the mortality rate of CHD has increased. Nevertheless, patients often do not realize they have CHD until their condition is serious due to the complexity, high cost, and the side effects of the diagnosis process. Thus, research on predicting CHD risk has been conducted. The Framingham study is a widely-accepted study in this field. However, one of its limitations is its overestimation of risk, which threatens its accuracy. Therefore, this study suggests a more advanced CHD risk prediction algorithm based on Optimized-DBN (Deep Belief Network). Optimized- DBN is an algorithm to improve performance by overcoming the limitations of the existing DBN. DBN does not have the global optimum values for number of layers and nodes, which affects research results. We overcame this limitation by combining with a genetic algorithm. The result of genetic algorithm for deriving the number of layers and nodes of Optimized-DBN for CHD prediction was 2 layers, 5 and 7 nodes to each layers. The accuracy of the CHD prediction algorithm based on Optimized- DBN which is developed by applying results of genetic algorithm was 0.8924, which is better than Framingham’s 0.5015 and DBN’s 0.7507. In the case of specificity, Optimized-DBN based CHD prediction was 0.7440, which was slightly lower than 0.8208 of existing DBN, but better than Framingham’s 0.65. In the case of sensitivity, Optimized-DBN is 0.8549, which is better than Framingham 0.4429 and DBN 0.7468. AUC of suggesting algorithm was 0.762, which was much better than Framingham 0.547 and DBN 0.570.

Evaluation of the Pooled Cohort Risk Equations for Cardiovascular Risk Prediction in a Multiethnic Cohort From the Women’s Health Initiative

Atherosclerotic cardiovascular disease (ASCVD) kills approximately 1 in every 3 US women. Current cholesterol, hypertension, and aspirin guidelines recommend calculating 10-year risk of ASCVD using the 2013 Pooled Cohort Equations (PCE). However, numerous studies have reported apparent overestimation of risk with the PCE, and reasons for overestimation are unclear. We evaluated the predictive accuracy of the PCE in the Women's Health Initiative (WHI), a multiethnic cohort of contemporary US postmenopausal women. We evaluated the effects of time-varying treatments such as aspirin and statins, and ascertainment of additional ASCVD events by linkage with the Centers for Medicare and Medicaid Services (CMS) claims. The WHI recruited the largest number of US women (n = 161 808) with the racial/ethnic, geographic, and age diversity of the general population (1993-1998). For this study, we included women aged 50 to 79 (n = 19 995) participating in the WHI with data on the risk equation variables at baseline and who met the guideline inclusion and exclusion criteria. Median follow-up was 10 years. For this study, ASCVD was defined as myocardial infarction, stroke, or cardiovascular death. Among the 19 995 women (mean [SD] age, 64 [7.3] years; 8305 [41.5%] white, 7688 [38.5%] black, 3491 [17.5%] Hispanic, 103 [0.5%] American Indian, 321 [1.6%] Asian/Pacific Islander, and 87 [0.4%] other/unknown), a total of 1236 ASCVD events occurred in 10 years and were adjudicated through medical record review by WHI investigators. The WHI-adjudicated observed risks were lower than predicted. The observed (predicted) risks for baseline 10-year risk categories less than 5%, 5% to less than 7.5%, 7.5% to less than 10%, and 10% or more were 1.7 (2.8), 4.4 (6.2), 5.3 (8.7), and 12.4 (18.2), respectively. Small changes were noted after adjusting for time-dependent changes in statin and aspirin use. Among women 65 years or older enrolled in Medicare, WHI-adjudicated risks were also lower than predicted, but observed (predicted) risks became aligned after including events ascertained by linkage with CMS for additional surveillance for events: 3.8 (4.3), 7.1 (6.4), 8.3 (8.7), and 18.9 (18.7), respectively. Similar results were seen across ethnic/racial groups. Overall, the equations discriminated risk well (C statistic, 0.726; 95% CI, 0.714-0.738). Without including surveillance for ASCVD events using CMS, observed risks in the WHI were lower than predicted by PCE as noted in several other US cohorts, but risks were better aligned after including CMS events. ClinicalTrials.gov identifier: NCT00000611.

Prediction Models for Clinical Outcome After a Carotid Revascularization Procedure.

Background and Purpose- Prediction models may help physicians to stratify patients with high and low risk for periprocedural complications or long-term stroke risk after carotid artery stenting or carotid endarterectomy. We aimed to evaluate external performance of previously published prediction models for short- and long-term outcome after carotid revascularization in patients with symptomatic carotid artery stenosis. Methods- From a literature review, we selected all prediction models that used only readily available patient characteristics known before procedure initiation. Follow-up data from 2184 carotid artery stenting and 2261 carotid endarterectomy patients from 4 randomized trials (EVA-3S [Endarterectomy Versus Angioplasty in Patients With Symptomatic Severe Carotid Stenosis], SPACE [Stent-Protected Angioplasty Versus Carotid Endarterectomy], ICSS [International Carotid Stenting Study], and CREST [Carotid Revascularization Endarterectomy Versus Stenting Trial]) were used to validate 23 short-term outcome models to estimate stroke or death risk ≤30 days after the procedure and the original outcome measure for which the model was developed. Additionally, we validated 7 long-term outcome models for the original outcome measure. Predictive performance of the models was assessed with C statistics and calibration plots. Results- Stroke or death ≤30 days after the procedure occurred in 158 (7.2%) patients after carotid artery stenting and in 84 (3.7%) patients after carotid endarterectomy. Most models for short-term outcome after carotid artery stenting (n=4) or carotid endarterectomy (n=19) had poor discriminative performance (C statistics ranging from 0.49-0.64) and poor calibration with small absolute risk differences between the lowest and highest risk groups and overestimation of risk in the highest risk groups. Long-term outcome models (n=7) had a slightly better performance with C statistics ranging from 0.59 to 0.67 and reasonable calibration. Conclusions- Current models did not reliably predict outcome after carotid revascularization in a trial population of patients with symptomatic carotid stenosis. In particular, prediction of short-term outcome seemed to be difficult. Further external validation of existing prediction models or development of new prediction models is needed before such models can be used to support treatment decisions in individual patients.

Crash risk by driver age, gender, and time of day using a new exposure methodology

Introduction: Concerns have been raised that the nonlinear relation between crashes and travel exposure invalidates the conventional use of crash rates to control for exposure. A new metric of exposure that bears a linear association to crashes was used as basis for calculating unbiased crash risks. This study compared the two methods – conventional crash rates and new adjusted crash risk – for assessing the effect of driver age, gender, and time of day on the risk of crash involvement and crash fatality. Method: We used police reports of single-car and multi-car crashes with fatal and nonfatal driver injuries that occurred during 2002–2012 in Great Britain. Results: Conventional crash rates were highest in the youngest age group and declined steeply until age 60–69 years. The adjusted crash risk instead peaked at age 21–29 years and reduced gradually with age. The risk of nighttime driving, especially among teenage drivers, was much smaller when based on adjusted crash risks. Finally, the adjusted fatality risk incurred by elderly drivers remained constant across time of day, suggesting that their risk of sustaining a fatal injury due to a crash is more attributable to excess fragility than to crash seriousness. Conclusions: Our findings demonstrate a biasing effect of low travel exposure on conventional crash rates. This implies that conventional methods do not yield meaningful comparisons of crash risk between driver groups and driving conditions of varying exposure to risk. The excess crash rates typically associated with teenage and elderly drivers as well as nighttime driving are attributed in part to overestimation of risk at low travel exposure. Practical Applications: Greater attention should be directed toward crash involvement among drivers in their 20s and 30s as well as younger drivers. Countermeasures should focus on the role of physical vulnerability in fatality risk of elderly drivers.