Abstract 153: Novel Machine Learning Proves Stroke Risk is Not Linear

Stroke ◽  
2020 ◽  
Vol 51 (Suppl_1) ◽  
Author(s):  
Agni Orfanoudaki ◽  
Amre M Nouh ◽  
Emma Chesley ◽  
Christian Cadisch ◽  
Barry Stein ◽  
...  
Keyword(s):  
Machine Learning ◽  
Risk Factors ◽  
Risk Score ◽  
Stroke Risk ◽  
Assessment Tools ◽  
Risk Calculator ◽  
Stroke Incidence ◽  
Non Linear ◽  
User Friendly ◽  
The Impact

Background: Current stroke risk assessment tools presume the impact of risk factors is linear and cumulative. However, both novel risk factors and their interplay influencing stroke incidence are difficult to reveal using traditional linear models. Objective: To improve upon the Revised-Framingham Stroke Risk Score and design an interactive non-linear Stroke Risk Score (NSRS). Our work aimed at increasing the accuracy of event prediction and uncovering new relationships in an interpretable user-friendly fashion. Methods: A two phase approach was used to develop our stroke risk score predictor. First, clinical examinations of the Framingham offspring cohort were utilized as the training dataset for the predictive model consisting of 14,196 samples where each clinical examination was considered an independent observation. Optimal Classification Trees (OCT) were used to train a model to predict 10-year stroke risk. Second, this model was validated with 17,527 observations from the Boston Medical Center. The NSRS was developed into an online user friendly application in the form of a questionnaire (http://www.mit.edu/~agniorf/files/questionnaire_Cohort2.html). Results: The algorithm suggests a key dichotomy between patients with or without history of cardiovascular disease. While the model agrees with known findings, it also identified 23 unique stroke risk profiles and introduced new non-linear relationships; such as the role of T-wave abnormality on electrocardiography and hematocrit levels in a patient’s risk profile. Our results in both the training and validation populations suggested that the non-linear approach significantly improves upon the existing revised Framingham stroke risk calculator in the c-statistic (training 87.43% (CI 0.85-0.90) vs. 73.74% (CI 0.70-0.76); validation 75.29% (CI 0.74-0.76) vs 65.93% (CI 0.64-0.67), even in multi-ethnicity populations. Conclusions: We constructed a highly predictive, interpretable and user-friendly stroke risk calculator using novel machine-learning uncovering new risk factors, interactions and unique profiles. The clinical implications include prioritization of risk factor modification and personalized care improving targeted intervention for stroke prevention.

Abstract TP196: New Zealand Workforce Stroke Incidence: Urban and Rural Risk Factor Evaluation

Stroke ◽  
2013 ◽  
Vol 44 (suppl_1) ◽  
Author(s):  
Hope E Buell ◽  
Patricia Metcalf ◽  
Daniel Exeter
Keyword(s):  
Risk Factors ◽  
New Zealand ◽  
Stroke Risk ◽  
Proportional Hazards ◽  
Proportional Hazards Model ◽  
Smoking Status ◽  
Prior History ◽  
Stroke Incidence ◽  
Stroke Outcomes ◽  
The Impact

This analysis aims to assess the impact of urban and rural risk factors on a model of stroke incidence in a New Zealand workforce population. The New Zealand study consisted of 4,926 subjects prospectively enrolled at 46 worksites. The subjects were aged 40-78 years at baseline and had no prior history of stroke. This prospective study defines stroke events experienced by the study subjects during follow-up between 1988 and 2012 based on hospital admission coding. Proportional hazards regression models were fit using baseline characteristics. The difference in stroke outcomes for urban and rural worksites was also evaluated. Results demonstrate that baseline demographic, physical exam, and behavioural measures impact stroke outcomes. While the baseline distribution of stroke risk factors such as Pacific Island ethnicity, smoking status, and increased blood pressure indicates a potentially higher risk of stroke in the rural population, the proportional hazards model does not identify increased stroke risk for rural workers. Additional analysis of the diet, exercise and Quality of Life measures for these subjects may provide further information into the stroke risk profiles of individuals working in different locales.

Abstract 183: Depressive Symptoms and Incident Stroke in Black and White Participants in the REGARDS Study

Stroke ◽  
2012 ◽  
Vol 43 (suppl_1) ◽  
Author(s):  
Audrey L Austin ◽  
Michael G Crowe ◽  
Martha R Crowther ◽  
Virginia J Howard ◽  
Abraham J Letter ◽  
...  
Keyword(s):  
Risk Factors ◽  
Risk Factor ◽  
Depressive Symptoms ◽  
Racial Differences ◽  
Stroke Risk ◽  
Depression Scale ◽  
Stroke Incidence ◽  
Blacks And Whites ◽  
The Impact

Background and Purpose: Research suggests that depression may contribute to stroke risk independent of other known risk factors. Most studies examining the impact of depression on stroke have been conducted with predominantly white cohorts, though blacks are known to have higher stroke incidence than whites. The purpose of this study was to examine depressive symptoms as a risk factor for incident stroke in blacks and whites, and determine whether depressive symptomatology was differentially predictive of stroke among blacks and whites. Methods: The REasons for Geographic and Racial Differences in Stroke (REGARDS), is a national, population-based longitudinal study designed to examine risk factors associated with black-white and regional disparities in stroke incidence. Among 30,239 participants (42% black) accrued from 2003-2007, excluding those lacking follow-up or data on depressive symptoms, 27,557 were stroke-free at baseline. As of the January 2011 data closure, over an average follow-up of 4.6 years, 548 incident stroke cases were verified by study physicians based on medical records review. The association between baseline depressive symptoms (assessed via the Center for Epidemiological Studies Depression scale, 4-item version) and incident stroke was analyzed with Cox proportional hazards models adjusted for demographic factors (age, race, and sex), stroke risk factors (hypertension, diabetes, smoking, atrial fibrillation, and history of heart disease), and social factors (education, income, and social network). Results: For the total sample, depressive symptoms were predictive of incident stroke. The association between depressive symptoms and stroke did not differ significantly based on race (Wald X 2 = 2.38, p = .1229). However, race-stratified analyses indicated that the association between depressive symptoms and stroke was stronger among whites and non-significant among blacks. Conclusions: Depressive symptoms were an independent risk factor for incident stroke among a national sample of blacks and whites. These findings suggest that assessment of depressive symptoms may warrant inclusion in stroke risk scales. The potential for a stronger association in whites than blacks requires further study.

scholarly journals Interpretable machine learning prediction of all-cause mortality

2021 ◽  
Author(s):  
Wei Qiu ◽  
Hugh Chen ◽  
Ayse Berceste Dincer ◽  
Su-In Lee
Keyword(s):  
Machine Learning ◽  
Risk Factors ◽  
Prediction Accuracy ◽  
Linear Models ◽  
Age Groups ◽  
Mortality Prediction ◽  
Linear Relationships ◽  
All Cause Mortality ◽  
Non Linear ◽  
The Impact

AbstractExplainable artificial intelligence provides an opportunity to improve prediction accuracy over standard linear models using “black box” machine learning (ML) models while still revealing insights into a complex outcome such as all-cause mortality. We propose the IMPACT (Interpretable Machine learning Prediction of All-Cause morTality) framework that implements and explains complex, non-linear ML models in epidemiological research, by combining a tree ensemble mortality prediction model and an explainability method. We use 133 variables from NHANES 1999–2014 datasets (number of samples: n = 47, 261) to predict all-cause mortality. To explain our model, we extract local (i.e., per-sample) explanations to verify well-studied mortality risk factors, and make new discoveries. We present major factors for predicting x-year mortality (x = 1, 3, 5) across different age groups and their individualized impact on mortality prediction. Moreover, we highlight interactions between risk factors associated with mortality prediction, which leads to findings that linear models do not reveal. We demonstrate that compared with traditional linear models, tree-based models have unique strengths such as: (1) improving prediction power, (2) making no distribution assumptions, (3) capturing non-linear relationships and important thresholds, (4) identifying feature interactions, and (5) detecting different non-linear relationships between models. Given the popularity of complex ML models in prognostic research, combining these models with explainability methods has implications for further applications of ML in medical fields. To our knowledge, this is the first study that combines complex ML models and state-of-the-art feature attributions to explain mortality prediction, which enables us to achieve higher prediction accuracy and gain new insights into the effect of risk factors on mortality.

Secular and longitudinal trends in cardiovascular risk in a general population using a national risk model: The Tromsø Study

2019 ◽  
Vol 26 (17) ◽  
pp. 1852-1861
Author(s):  
Amalie Nilsen ◽  
Tove A Hanssen ◽  
Knut T Lappegård ◽  
Anne E Eggen ◽  
Maja-Lisa Løchen ◽  
...  
Keyword(s):  
Risk Factors ◽  
Blood Pressure ◽  
Cardiovascular Risk ◽  
Systolic Blood Pressure ◽  
Birth Cohort ◽  
Risk Score ◽  
Risk Model ◽  
Assessment Tools ◽  
Cohort Differences ◽  
The Impact

Background Primary prevention guidelines promote the use of risk assessment tools to estimate total cardiovascular risk. We aimed to study trends in cardiovascular risk and contribution of single risk factors, using the newly developed NORRISK 2 risk score, which estimates 10-year risk of fatal and non-fatal cardiovascular events. Design Prospective population-based study. Methods We included women and men aged 45–74 years attending the sixth and seventh survey of the Tromsø Study (Tromsø 6, 2007–2008, n = 7284 and Tromsø 7, 2015–2016, n = 14,858) to study secular trends in NORRISK 2 score. To study longitudinal trends, we followed participants born 1941–1962 attending both surveys ( n = 4534). We calculated NORRISK 2 score and used linear regression models to study the relative contribution (% R2) of each single risk factor to the total score. Results Mean NORRISK 2 score decreased and distribution in risk categories moved from higher to lower risk in both sexes and all age-groups between the first and second surveys ( p < 0.001). In birth cohorts, when age was set to baseline in NORRISK 2 calculations, risk score decreased during follow-up. Main contributors to NORRISK 2 were systolic blood pressure, smoking and total cholesterol, with some sex, age and birth cohort differences. Conclusion We found significant favourable secular and longitudinal trends in total cardiovascular risk and single risk factors during the last decade. Change in systolic blood pressure, smoking and cholesterol were the main contributors to risk score change; however, the impact of single risk factors on the total score differed by sex, age and birth cohort.

scholarly journals Machine learning provides evidence that stroke risk is not linear: The non-linear Framingham stroke risk score

PLoS ONE ◽  
2020 ◽  
Vol 15 (5) ◽  
pp. e0232414 ◽  
Author(s):  
Agni Orfanoudaki ◽  
Emma Chesley ◽  
Christian Cadisch ◽  
Barry Stein ◽  
Amre Nouh ◽  
...  
Keyword(s):  
Machine Learning ◽  
Risk Score ◽  
Stroke Risk ◽  
Non Linear

Stress, anxiety, depression and burnout in frontline health care workers during COVID-19 pandemic: a brief systematic review and new data from Russia

2021 ◽  
Author(s):  
Ekaterina Mosolova ◽  
Dmitry Sosin ◽  
Sergey Mosolov
Keyword(s):  
Risk Factors ◽  
Systematic Review ◽  
Female Gender ◽  
Assessment Tools ◽  
Care Workers ◽  
Younger Age ◽  
Wide Range ◽  
Associated Risk Factors ◽  
Anxiety Depression ◽  
The Impact

During the COVID-19 pandemic, healthcare workers (HCWs) have been subject to increased workload while also exposed to many psychosocial stressors. In a systematic review we analyze the impact that the pandemic has had on HCWs mental state and associated risk factors. Most studies reported high levels of depression and anxiety among HCWs worldwide, however, due to a wide range of assessment tools, cut-off scores, and number of frontline participants in the studies, results were difficult to compare. Our study is based on two online surveys of 2195 HCWs from different regions of Russia during spring and autumn epidemic outbreaks revealed the rates of anxiety, stress, depression, emotional exhaustion and depersonalization and perceived stress as 32.3%, 31.1%, 45.5%, 74.2%, 37.7% ,67.8%, respectively. Moreover, 2.4% of HCWs reported suicidal thoughts. The most common risk factors include: female gender, nurse as an occupation, younger age, working for over 6 months, chronic diseases, smoking, high working demands, lack of personal protective equipment, low salary, lack of social support, isolation from families, the fear of relatives getting infected. These results demonstrate the need for urgent supportive programs for HCWs fighting COVID-19 that fall into higher risk factors groups.

scholarly journals Body Mass Index, Multi-Morbidity, and COVID-19 Risk Factors as Predictors of Severe COVID-19 Outcomes

2021 ◽  
Vol 12 ◽  
pp. 215013272110185
Author(s):  
Sanjeev Nanda ◽  
Audry S. Chacin Suarez ◽  
Loren Toussaint ◽  
Ann Vincent ◽  
Karen M. Fischer ◽  
...  
Keyword(s):  
Risk Factors ◽  
Body Mass Index ◽  
Hospital Admission ◽  
Body Mass ◽  
Risk Score ◽  
The Body ◽  
Risk Scores ◽  
Nasopharyngeal Swab ◽  
Morbidity Score ◽  
The Impact

Purpose The purpose of the present study was to investigate body mass index, multi-morbidity, and COVID-19 Risk Score as predictors of severe COVID-19 outcomes. Patients Patients from this study are from a well-characterized patient cohort collected at Mayo Clinic between January 1, 2020 and May 23, 2020; with confirmed COVID-19 diagnosis defined as a positive result on reverse-transcriptase-polymerase-chain-reaction (RT-PCR) assays from nasopharyngeal swab specimens. Measures Demographic and clinical data were extracted from the electronic medical record. The data included: date of birth, gender, ethnicity, race, marital status, medications (active COVID-19 agents), weight and height (from which the Body Mass Index (BMI) was calculated, history of smoking, and comorbid conditions to calculate the Charlson Comorbidity Index (CCI) and the U.S Department of Health and Human Services (DHHS) multi-morbidity score. An additional COVID-19 Risk Score was also included. Outcomes included hospital admission, ICU admission, and death. Results Cox proportional hazards models were used to determine the impact on mortality or hospital admission. Age, sex, and race (white/Latino, white/non-Latino, other, did not disclose) were adjusted for in the model. Patients with higher COVID-19 Risk Scores had a significantly higher likelihood of being at least admitted to the hospital (HR = 1.80; 95% CI = 1.30, 2.50; P < .001), or experiencing death or inpatient admission (includes ICU admissions) (HR = 1.20; 95% CI = 1.02, 1.42; P = .028). Age was the only statistically significant demographic predictor, but obesity was not a significant predictor of any of the outcomes. Conclusion Age and COVID-19 Risk Scores were significant predictors of severe COVID-19 outcomes. Further work should examine the properties of the COVID-19 Risk Factors Scale.

scholarly journals A Narrative Review of the Risk Factors for Cancer and the Preventive Opportunities: Current Status, Future Perspectives, and Implications for India

2021 ◽  
Author(s):  
Vinod K. Ramani ◽  
Ganesha D. V. ◽  
Radheshyam Naik
Keyword(s):  
Risk Factors ◽  
Lung Cancer ◽  
Air Pollution ◽  
Cancer Prevention ◽  
Assessment Tools ◽  
Biologically Active ◽  
Pubmed Database ◽  
Risk Of Cancer ◽  
Google Search ◽  
The Impact

Abstract Introduction Clinical cancer can arise from heterogenous pathways through various genetic mutations. Although we cannot predict the timeline by which an individual will develop cancer, certain risk assessment tools can be used among high-risk groups for focusing the preventive activities. As primary level of cancer prevention, healthy lifestyle approach is being promoted. The etiological factors for lung cancer include by-products of industrialization and air pollution. We need to factor the increase in household air pollution as well. Methods “PubMed” database and Google search engines were used for searching the relevant articles. Search terms with Boolean operators used include “Cancer prevention,” “Missed opportunities in cancer causation,” and “incidence of risk factors.” This review includes 20 studies and other relevant literature that address the opportunities for cancer prevention. Body The narrative describes the association between many of the risk factors and development of cancer. This includes tobacco, alcohol, infections, air pollution, physical inactivity, diet, obesity, screening and preventive strategies, chemoprevention, biomarkers of carcinogenesis, and factors that prolong the diagnosis of cancer. Discussion Reports from basic science research provide evidence on the potential of biologically active food components and pharmacological agents for mitigating the risk of cancer and its progression. However, some reports from observational studies and randomized trials have been inconsistent. We need to recognize the impact of sociodemographic factors such as age, sex, ethnicity, culture, and comorbid illness on preventive interventions. Spiral computed tomographic scan is a robust tool for early detection of lung cancer. Conclusion Infectious etiology for specific cancers provides opportunities for prevention and treatment. The complex interplay between man and microbial flora needs to be dissected, for understanding the pathogenesis of relevant malignancies. For reducing the morbidity of cancer, we need to focus on prevention as a priority strategy and intervene early during the carcinogenic process.

The Burden of Stroke in Pakistan

2008 ◽  
Vol 3 (4) ◽  
pp. 293-296 ◽  
Author(s):  
Bhojo A. Khealani ◽  
Mohammad Wasay
Keyword(s):  
Risk Factors ◽  
Stroke Risk ◽  
Case Series ◽  
Stroke Care ◽  
Poor Control ◽  
Stroke Incidence ◽  
Treatment Practices ◽  
Stroke Risk Factors ◽  
The Subject ◽  
Artery Disease

Epidemiologic literature on stroke burden, patterns of stroke is almost non existent from Pakistan. However, several hospital-based case series on the subject are available, mainly published in local medical journals. Despite the fact that true stroke incidence and prevalence of stroke in Pakistan is not known, the burden is assumed to be high because of highly prevalent stroke risk factors (hypertension, diabetes mellitus, coronary artery disease, dyslipidemia and smoking) in the community. High burden of these conventional stroke risk factors is further compounded by lack of awareness, poor compliance hence poor control, and inappropriate management/treatment practices. In addition certain risk factors like rheumatic valvular heart disease may be more prevalent in Pakistan. We reviewed the existing literature on stroke risk factors in community, the risk factor prevalence among stroke patients, patterns of stroke, out come of stroke, availability of diagnostic services/facilities related to stroke and resources for stroke care in Pakistan.

Sign in / Sign up

Export Citation Format

Share Document