Materiały źródłowe

• #1

  https://www.who.int/news-room/fact-sheets/detail/cardiovascular-diseases-(cvds)

  Cardiovascular diseases (CVDs) are the leading cause of death globally. […] It is important to detect cardiovascular disease as early as possible so that management with counselling and medicines can begin. […] These intermediate risks factors can be measured in primary care facilities and indicate an increased risk of heart attack, stroke, heart failure and other complications. […] In addition, drug treatment of hypertension, diabetes and high blood lipids are necessary to reduce cardiovascular risk and prevent heart attacks and strokes among people with these conditions. […] At least three-quarters of the world’s deaths from CVDs occur in low- and middle-income countries. […] The key to cardiovascular disease reduction lies in the inclusion of cardiovascular disease management interventions in universal health coverage packages, although in a high number of countries health systems require significant investment and reorientation to effectively manage CVDs.

• #1 Deep learning-based artificial intelligence for predicting risk a

  https://www.openaccessjournals.com/articles/deep-learningbased-artificial-intelligence-for-predicting-risk-and-prognosis-in-patients-with-cardiovascular-disease-14269.html

  Although these conventional models based on regression were useful in clinical practice, these statistical methods use a limited number of predictive factors that operate in the same manner for all patients. […] Therefore, these models have a problem of over-generalization and are not applicable to all individual patients. […] Predicting the risk and prognosis of CVD is a complex task with many factors to consider and needs time-consuming human operations to analyze that. Deep learning based AI has excellent ability to solve problems automatically by analyzing these complex factors. […] The deep learning model was developed using data from the Korea OHCA Registry (KOHCAR) in South Korea, in which 36,190 patients with OHCA from 712 emergency departments (EDs) were enrolled. […] Therefore, deep learning based AI model accurately predicted the neurologic recovery and survival to discharge of OHCA patients, outperforming the conventional method.

• #1 Prediction models for cardiovascular disease risk in the general population: systematic review | The BMJ

  https://www.bmj.com/content/353/bmj.i2416

  Most models (66%) were based on a common set of predictors, consisting of age, smoking, blood pressure, and cholesterol levels. […] The current literature is overwhelmed with models for predicting the risk of cardiovascular outcomes in the general population. Most, however, have not been externally validated or directly compared on their relative predictive performance, making them currently of yet unknown value for practitioners, policy makers, and guideline developers.

• #1 Cardiovascular Event Prediction by Machine Learning: The Multi-Ethnic Study of Atherosclerosis

  https://pmc.ncbi.nlm.nih.gov/articles/PMC5640485/

  The RF based method of risk prediction provided better event prediction over standard risk scores. […] The results also suggest the importance of deep phenotyping using subclinical markers defined by imaging, electrocardiographic and blood biochemistry, as revealed by their prominent presence on the lists of top-20 predictors, for cardiovascular disease event prediction. […] For all outcomes of interest, the RF model with all 735 covariates showed a very high C-index and low BS. […] The prediction ability of conventional risk scores for heart failure (MESA HF risk score), cardiovascular disease (AHA/ASCVD risk score) and coronary heart disease (Framingham CHD risk score) are also shown (yellow curve). […] The performance of the RF-20 model for incident CHD prediction was better than the Framingham CHD risk score (C-index: 069, BS: 0072) with a higher C-index and lower BS. […] In an extensively phenotyped population free of CV disease at baseline, using random forests, we show efficient cardiovascular risk prediction for specific outcomes including death, stroke, CV events, incident heart failure and atrial fibrillation.

• #1 Cardiovascular Event Prediction by Machine Learning: The Multi-Ethnic Study of Atherosclerosis

  https://pmc.ncbi.nlm.nih.gov/articles/PMC5640485/

  Machine learning may be useful to characterize cardiovascular risk, predict outcomes and identify biomarkers in population studies. […] To test the ability of random survival forests (RF), a machine learning technique, to predict six cardiovascular outcomes in comparison to standard cardiovascular risk scores. […] We included participants from the Multi-Ethnic Study of Atherosclerosis (MESA). […] The RF technique performed better than established risk scores with increased prediction accuracy (decreased Brier score by 1025%). […] Machine learning in conjunction with deep phenotyping improve prediction accuracy in cardiovascular event prediction in an initially asymptomatic population. […] The results of this study suggest that machine learning methods are well-suited for meaningful risk prediction in extensively phenotyped large-scale epidemiological studies.

• #1 Prediction model for cardiovascular disease in patients with diabetes using machine learning derived and validated in two independent Korean cohorts | Scientific Reports

  https://www.nature.com/articles/s41598-024-63798-y

  The RF model displayed impressive performance on the validation set, exhibiting an AUROC of 0.830. […] Consequently, considering its superior and consistent results, the RF model emerged as the most effective predictor of CVD onset within a 3-year timeframe in patients with diabetes. […] The findings of this study can potentially improve patient outcomes by facilitating timely interventions, enhancing the understanding of contributing variables, and reducing the burden of cardiovascular complications in patients with diabetes. […] Our model outperformed traditional risk assessment tools, such as the Framingham risk score, which has shown limitations in their applicability to patients with diabetes. […] Highlighting the importance of creatinine and HbA1c level variabilities, the study illustrates how well-developed ML models can predict CVD risk across diverse populations, using routine clinical data to enhance risk assessment accessibility.

• #1

  https://link.springer.com/article/10.1007/s10916-024-02087-7

  The most frequently selected ML models (based on their prediction performance) were ensemble methods (RSF and survival GBM). […] As for DL models, in almost all studies, DeepSurv was selected as the best performing model. […] Consistent with studies that have examined both ML and DL models in the context of predicting the survival of cancer patients, our study found that DL models surpass ML models in predicting time to CVD occurrence. […] Despite recent studies revealing the major role of SDoH in CVD, only a handful of prediction models incorporated a wide range of SDoH variables. […] In 80% of the studies, gender-stratified prediction was overlooked despite gender playing a role in CVD presentation, diagnosis, and survival. […] AI-based risk prediction models have an increased discrimination ability and accuracy as compared to the conventional multivariable models. […] To improve CVD risk prediction and inform clinicians decision-making future studies need to assess SDoH, in addition to the traditional factors and other emerging risk factors. […] Therefore, it is crucial to consider gender differences when it comes to predicting and managing CVD risks.

• #1 Risk prediction of cardiovascular disease using machine learning classifiers

  https://www.degruyter.com/document/doi/10.1515/med-2022-0508/html?lang=en

  Cardiovascular disease (CVD) makes our heart and blood vessels dysfunctional and often leads to death or physical paralysis. Therefore, early and automatic detection of CVD can save many human lives. […] Early diagnosis of CVD can potentially cure patients and save innumerable lives. Diagnosis and treatment of patients at early stages by cardiologists remain a challenge. […] The goal of this study is to determine if ML can enhance cardiovascular risk prediction accuracy in population primary care at large and find out which ML algorithm result had fairly high brevity. […] This study presents a comparison of two ML techniques for CVD prediction: K-NN and MLP. Between these algorithms, MLP provides better accuracy (82.47%) than K-NN with an accuracy of 73.77%. The diagnosis rate was found to be 86.41 and 86.21% for the MLP and K-NN algorithms, respectively.

• #1 XAI Framework for Cardiovascular Disease Prediction Using Classification Techniques

  https://www.mdpi.com/2079-9292/11/24/4086

  XAI Framework for Cardiovascular Disease Prediction Using Classification Techniques […] Machine intelligence models are robust in classifying the datasets for data analytics and for predicting the insights that would assist in making clinical decisions. The models would assist in the disease prognosis and preliminary disease investigation, which is crucial for effective treatment. […] The efficiency of the XAI-based classification models is reasonably fair, compared to the other state-of-the-art models, which are assessed using the various evaluation metrics, such as area under curve (AUC), receiver operating characteristic (ROC), sensitivity, specificity, and the F1-score. The performances of the XAI-driven SVM, LR, and naive Bayes are robust, with an accuracy of 89%, which is assumed to be reasonably fair, compared to the existing models. […] The performance of the proposed XAI-based prediction of cardiovascular disease using the ensemble classifiers is being assessed over various evaluation metrics, such as the AUC, accuracy, true positive rate (TPR), recall, and precision. The XAI-driven feature selection and weight optimization have enhanced the model’s performance and made the internal evaluation of the model better interpretable, to make the decisions trustworthy. […] The comparative results obtained from the dataset using six ensemble models, i.e., SVM, KNN, AdaBoost, bagging, logistic regression, and the Gaussian naive Bayes algorithms, are shown in Table 4. The accuracy obtained by the SVM after training on the dataset is 82.5%, the highest accuracy among the other algorithms. […] The ROC curve train categorizes a patient’s disease condition as either positive or negative, based on the test results and perceives the ideal cut-off value with the best symptomatic performance. […] The efficiency of the proposed XAI-driven ensemble classifiers is better than the conventional classification models. Moreover, the XAI framework is more interpretable, and the decisions are trustworthy. […] The SVM algorithm has exhibited a better performance with an accuracy of 82.5%, among all of the classifiers used in heart disease classification.

• #1 Cardiovascular risk prediction in type 2 diabetes: a comparison of 22 risk scores in primary care setting | medRxiv

  https://www.medrxiv.org/content/10.1101/2020.10.08.20209015v3.full-text

  Risk scores performed badly in people with pre-existing CVD (c-statistic 0.55). […] Scores with more predictors did not perform better: for CVD+ QRISK3 (19 variables) c-statistic 0.69 (95%CI 0.68;0.69), compared to CHD Basic (8 variables) 0.71 (95%CI 0.70; 0.71). […] Conclusions CVD risk prediction scores performed well in T2DM, irrespective of derivation population and of original predicted outcome. […] Scores performed poorly in patients with established CVD. […] Complex scores with multiple variables did not outperform simple scores. […] A simple population specific recalibration markedly improved score performance and is recommended for future use.

• #1 A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease | npj Digital Medicine

  https://www.nature.com/articles/s41746-023-00916-6

  The electrocardiogram (ECG) is the most frequently performed cardiovascular diagnostic test, but it is unclear how much information resting ECGs contain about long term cardiovascular risk. Here we report that a deep convolutional neural network can accurately predict the long-term risk of cardiovascular mortality and disease based on a resting ECG alone. […] SEER predicts 5-year cardiovascular mortality with an area under the receiver operator characteristic curve (AUC) of 0.83 in a held-out test set at Stanford, and with AUCs of 0.78 and 0.83 respectively when independently evaluated at Cedars-Sinai Medical Center and Columbia University Irving Medical Center. […] SEER can also predict several other cardiovascular conditions such as heart failure and atrial fibrillation. […] SEER is trained on Stanford ECGs to predict 5-year cardiovascular mortality, but can accurately predict a range of cardiovascular disease across an array of time-scales.

• #1 Predicting cardiovascular disease risk using photoplethysmography and deep learning | PLOS Global Public Health

  https://journals.plos.org/globalpublichealth/article?id=10.1371/journal.pgph.0003204

  Cardiovascular diseases (CVDs) are responsible for a large proportion of premature deaths in low- and middle-income countries. Early CVD detection and intervention is critical in these populations, yet many existing CVD risk scores require a physical examination or lab measurements, which can be challenging in such health systems due to limited accessibility. We investigated the potential to use photoplethysmography (PPG), a sensing technology available on most smartphones that can potentially enable large-scale screening at low cost, for CVD risk prediction. We developed a deep learning PPG-based CVD risk score (DLS) to predict the probability of having major adverse cardiovascular events (MACE: non-fatal myocardial infarction, stroke, and cardiovascular death) within ten years, given only age, sex, smoking status and PPG as predictors. […] DLS predicts ten-year MACE risk comparable with the office-based refit-WHO score. Our study provides a proof-of-concept and suggests the potential of a PPG-based approach strategies for community-based primary prevention in resource-limited regions.

• #1 Predicting cardiovascular disease risk using photoplethysmography and deep learning | PLOS Global Public Health

  https://journals.plos.org/globalpublichealth/article?id=10.1371/journal.pgph.0003204

  We developed a deep learning PPG-based CVD risk score, DLS, to predict ten-year MACE risk using age, sex, smoking status, heart rate and deep learning-derived PPG features. Without requiring any vital signs or laboratory measurement, DLS demonstrated non-inferior performance compared to the office-based refit-WHO score with coefficients re-estimated on the same cohort. Results were consistent between metrics (C-statistic, NRI, cfNRI, sensitivity, specificity, calibration slope), and in various subgroups. Improved cfNRI and NRI also indicate the capability of DLS to reclassify cases better than the office-based refit-WHO score. Additionally, if available, adding office-based features (BMI, SBP) on top of DLS further improved the model performance.

• #1 Risk stratification and prognosis prediction using cardiac biomarkers in COVID-19: a single-centre retrospective cohort study | BMJ Open

  https://bmjopen.bmj.com/content/14/4/e082220

  Objective There is a need for a robust tool to stratify the patients risk with COVID-19. We assessed the prognostic values of cardiac biomarkers for COVID-19 patients. […] The adjusted risk for all-cause death remained significant for each threshold of cardiac biomarkers. […] Elevation of cardiac biomarkers was associated with poor prognosis of COVID-19 patients. […] Each cardiac biomarker, namely hsTnI and NT-proBNP, could well stratify the mortality risk of COVID-19 patients requiring admission. […] Current analysis revealed the high ability for risk stratification for mortality in COVID-19 patients. […] Elevation of hsTnI, NT-proBNP, CK or CK-MB at the time of admission was associated with a poor prognosis in the current relatively severely ill COVID-19 patients. Measurement of cardiac biomarkers can be an attractive option for risk stratification and deciding appropriate management in patients with COVID-19.

• #1 Predictive Value of IL-6 and PDGF-AA for Short-term Mortality Risk in | IJGM

  https://www.dovepress.com/predictive-value-of-il-6-and-pdgf-aa-for-28-day-mortality-risk-in-crit-peer-reviewed-fulltext-article-IJGM

  In conclusion, IL-6 and PDGF-AA levels are independent risk factors associated with the 28-day mortality in critically ill patients. This study highlights the importance of monitoring serum levels of IL-6 and PDGF-AA in critically ill patients. Compared with the marker alone, combinations with other conventional risk factors have better predictive values. […] The logistic regression analysis revealed that IL-6 and PDGF-AA were both independent risk factors associated with 28-day mortality (OR=1.003, 95% CI (1.0011.005), OR=1.002, 95% CI (1.0011.003)). […] The AUROCs for IL-6, PDGF-AA, SOFA and APACHE II scores were 0.736, 0.596, 0.856, and 0.767, respectively. The AUROC curves of IL-6 and SOFA were significantly different from those of IL-6, SOFA, and APACHE II scores. The diagnostic value of IL-6 combined SOFA was better than that of SOFA alone, with an AUROC of 0.892, a sensitivity of 91.4%, and a specificity of 71.7%. Further combined use of IL-6, PDGF-AA and SOFA can increase the AUROC to 0.905 (95% CI (0.864,0.946) and specificity to 91.5%. These data indicate that the addition of IL-6 and PDGF-AA to SOFA can effectively improve the sensitivity or specificity for predicting short-term mortality in critically ill patients.

• #1 CardioRiskNet: A Hybrid AI-Based Model for Explainable Risk Prediction and Prognosis in Cardiovascular Disease

  https://www.mdpi.com/2306-5354/11/8/822

  This paper addresses this critical need by introducing CardioRiskNet, a novel hybrid AI model for CVD risk assessment and prognosis. CardioRiskNet surpasses the traditional models by harnessing the power of XAI, active learning, and attention mechanisms. […] Integrating XAI, active learning, and attention mechanisms in risk prediction and prognosis models holds great promise for improving the accuracy, interpretability, and clinical relevance of CVD assessments. […] The traditional models have played a vital role in this domain, providing valuable insights. However, their limitations in handling complex and diverse patient data, along with the growing need for interpretable AI in clinical settings, necessitate the exploration of more advanced techniques. […] AI algorithms can examine many different kinds of patient data, like their medical history, imaging results, and genetic information, to find hidden patterns and links that may initiate CVDs and cause them to become worse.

• #1 CardioRiskNet: A Hybrid AI-Based Model for Explainable Risk Prediction and Prognosis in Cardiovascular Disease

  https://www.mdpi.com/2306-5354/11/8/822

  Cardiovascular diseases (CVDs) continue to pose a significant global health challenge, accounting for a substantial burden of morbidity and mortality. Timely risk prediction and prognosis play a critical role in identifying individuals at high risk of developing CVDs, enabling early intervention and personalized treatment plans. […] The global prevalence of cardiovascular diseases (CVDs) as a leading cause of death highlights the imperative need for refined risk assessment and prognostication methods. […] The proposed CardioRiskNet consists of seven parts: data preprocessing, feature selection and encoding, eXplainable AI (XAI) integration, active learning, attention mechanisms, risk prediction and prognosis, evaluation and validation, and deployment and integration. […] According to the experimental results, CardioRiskNet demonstrates superior performance in terms of accuracy, sensitivity, specificity, and F1-Score, with values of 98.7%, 98.7%, 99%, and 98.7%, respectively. These findings show that CardioRiskNet can accurately assess and prognosticate the CVD risk, demonstrating the power of active learning and AI to surpass the conventional methods.

• #1 Prediction models for cardiovascular disease risk in the general population: systematic review | The BMJ

  https://www.bmj.com/content/353/bmj.i2416

  Objective To provide an overview of prediction models for risk of cardiovascular disease (CVD) in the general population. […] Conclusions There is an excess of models predicting incident CVD in the general population. The usefulness of most of the models remains unclear owing to methodological shortcomings, incomplete presentation, and lack of external validation and model impact studies. […] This review shows that there is an abundance of cardiovascular risk prediction models for the general population. […] Clearly, the array of studies describing the development of new risk prediction models for cardiovascular disease (CVD) in the general population is overwhelming, whereas there is a paucity of external validation studies for most of these developed models. […] Healthcare professionals and policymakers are already in great doubt about which CVD prediction model to use or advocate in their specific setting or population.

• #1 Modelling of longitudinal data to predict cardiovascular disease risk: a methodological review | BMC Medical Research Methodology | Full Text

  https://bmcmedresmethodol.biomedcentral.com/articles/10.1186/s12874-021-01472-x

  Single stage models are still heavily utilized by many CVD risk prediction studies for modelling longitudinal data. Future studies should fully utilize available longitudinal data when analyzing CVD risk by employing two-stage and joint approaches which can often better utilize the available data. […] This review has identified a multitude of methods to analyze the risk of CVD using longitudinally repeated data. There has been an increase in the complexity of methodology used over the past two decades, with an increasing proportion of studies applying more efficient approaches such as two-stage and joint models over time. However, many studies only used simple analysis based on one time-point, even when more data were available. […] The use of two-stage and joint models is a critical part of understanding the relationship between the longitudinal risk factors and CVD. Many studies still employ single stage approaches which often underutilize available longitudinal data when modelling cardiovascular risk. Further studies should aim to optimize the use of longitudinal data by using two-stage and joint models whenever possible for a more accurate estimation of cardiovascular risk.

• #2 Deep learning-based artificial intelligence for predicting risk a

  https://www.openaccessjournals.com/articles/deep-learningbased-artificial-intelligence-for-predicting-risk-and-prognosis-in-patients-with-cardiovascular-disease-14269.html

  Although these conventional models based on regression were useful in clinical practice, these statistical methods use a limited number of predictive factors that operate in the same manner for all patients. […] Therefore, these models have a problem of over-generalization and are not applicable to all individual patients. […] Predicting the risk and prognosis of CVD is a complex task with many factors to consider and needs time-consuming human operations to analyze that. Deep learning based AI has excellent ability to solve problems automatically by analyzing these complex factors. […] The deep learning model was developed using data from the Korea OHCA Registry (KOHCAR) in South Korea, in which 36,190 patients with OHCA from 712 emergency departments (EDs) were enrolled. […] Therefore, deep learning based AI model accurately predicted the neurologic recovery and survival to discharge of OHCA patients, outperforming the conventional method.

• #2 Prediction models for cardiovascular disease risk in the general population: systematic review | The BMJ

  https://www.bmj.com/content/353/bmj.i2416

  Most models (66%) were based on a common set of predictors, consisting of age, smoking, blood pressure, and cholesterol levels. […] The current literature is overwhelmed with models for predicting the risk of cardiovascular outcomes in the general population. Most, however, have not been externally validated or directly compared on their relative predictive performance, making them currently of yet unknown value for practitioners, policy makers, and guideline developers.

• #2 Machine learning-based long-term outcome prediction in patients undergoing percutaneous coronary intervention – Liu – Cardiovascular Diagnosis and Therapy

  https://cdt.amegroups.org/article/view/69917/html

  Traditional prognostic risk assessment in patients with coronary artery disease undergoing percutaneous coronary intervention (PCI) is based on a limited selection of clinical and imaging findings. […] Machine learning (ML) can consider a higher number and complexity of variables and may be useful for characterising cardiovascular risk, predicting outcomes, and identifying biomarkers in large population studies. […] The random forest model (RF-PCI) exhibited the best performance on predicting all-cause mortality. […] ML models improved the prediction of long-term all-cause mortality in patients with coronary artery disease before PCI. […] The performance of the RF model was better than that of the other models, providing a meaningful stratification. […] Prognosis assessment is the key to coronary heart disease diagnosis and treatment.

• #2 Prediction model for cardiovascular disease in patients with diabetes using machine learning derived and validated in two independent Korean cohorts | Scientific Reports

  https://www.nature.com/articles/s41598-024-63798-y

  The RF model displayed impressive performance on the validation set, exhibiting an AUROC of 0.830. […] Consequently, considering its superior and consistent results, the RF model emerged as the most effective predictor of CVD onset within a 3-year timeframe in patients with diabetes. […] The findings of this study can potentially improve patient outcomes by facilitating timely interventions, enhancing the understanding of contributing variables, and reducing the burden of cardiovascular complications in patients with diabetes. […] Our model outperformed traditional risk assessment tools, such as the Framingham risk score, which has shown limitations in their applicability to patients with diabetes. […] Highlighting the importance of creatinine and HbA1c level variabilities, the study illustrates how well-developed ML models can predict CVD risk across diverse populations, using routine clinical data to enhance risk assessment accessibility.

• #2 Risk prediction of cardiovascular disease using machine learning classifiers

  https://www.degruyterbrill.com/document/doi/10.1515/med-2022-0508/html?srsltid=AfmBOopOja3gOpNN7voqQF_rV1WZyt0bKVlUUQ7QmrMpQT26TvRYlg6z

  The comparison of results indicates that the MLP model has a higher prediction accuracy of 82.47%, followed by the K-NN model with an accuracy value of 73.77%. […] The constructed MLP model offers consistent accuracy compared to other techniques mentioned and is also capable of predicting other diseases. […] The proposed method can also be used for the classification of other chronic diseases such as breast cancer, liver disease, diabetes mellitus, and thyroid. […] The application of ML techniques will vastly aid in preventing fatalities and supplement the efforts of doctors in fighting CVD occurrence among all patient categories of different age groups, genders, and socio-economic backgrounds.

• #2

  https://link.springer.com/article/10.1007/s10916-024-02087-7

  The most frequently selected ML models (based on their prediction performance) were ensemble methods (RSF and survival GBM). […] As for DL models, in almost all studies, DeepSurv was selected as the best performing model. […] Consistent with studies that have examined both ML and DL models in the context of predicting the survival of cancer patients, our study found that DL models surpass ML models in predicting time to CVD occurrence. […] Despite recent studies revealing the major role of SDoH in CVD, only a handful of prediction models incorporated a wide range of SDoH variables. […] In 80% of the studies, gender-stratified prediction was overlooked despite gender playing a role in CVD presentation, diagnosis, and survival. […] AI-based risk prediction models have an increased discrimination ability and accuracy as compared to the conventional multivariable models. […] To improve CVD risk prediction and inform clinicians decision-making future studies need to assess SDoH, in addition to the traditional factors and other emerging risk factors. […] Therefore, it is crucial to consider gender differences when it comes to predicting and managing CVD risks.

• #2 A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease | npj Digital Medicine

  https://www.nature.com/articles/s41746-023-00916-6

  The electrocardiogram (ECG) is the most frequently performed cardiovascular diagnostic test, but it is unclear how much information resting ECGs contain about long term cardiovascular risk. Here we report that a deep convolutional neural network can accurately predict the long-term risk of cardiovascular mortality and disease based on a resting ECG alone. […] SEER predicts 5-year cardiovascular mortality with an area under the receiver operator characteristic curve (AUC) of 0.83 in a held-out test set at Stanford, and with AUCs of 0.78 and 0.83 respectively when independently evaluated at Cedars-Sinai Medical Center and Columbia University Irving Medical Center. […] SEER can also predict several other cardiovascular conditions such as heart failure and atrial fibrillation. […] SEER is trained on Stanford ECGs to predict 5-year cardiovascular mortality, but can accurately predict a range of cardiovascular disease across an array of time-scales.

• #2 Prediction models for cardiovascular disease risk in the general population: systematic review | The BMJ

  https://www.bmj.com/content/353/bmj.i2416

  Objective To provide an overview of prediction models for risk of cardiovascular disease (CVD) in the general population. […] Conclusions There is an excess of models predicting incident CVD in the general population. The usefulness of most of the models remains unclear owing to methodological shortcomings, incomplete presentation, and lack of external validation and model impact studies. […] This review shows that there is an abundance of cardiovascular risk prediction models for the general population. […] Clearly, the array of studies describing the development of new risk prediction models for cardiovascular disease (CVD) in the general population is overwhelming, whereas there is a paucity of external validation studies for most of these developed models. […] Healthcare professionals and policymakers are already in great doubt about which CVD prediction model to use or advocate in their specific setting or population.

• #3

  https://link.springer.com/article/10.1007/s10916-024-02087-7

  The most frequently selected ML models (based on their prediction performance) were ensemble methods (RSF and survival GBM). […] As for DL models, in almost all studies, DeepSurv was selected as the best performing model. […] Consistent with studies that have examined both ML and DL models in the context of predicting the survival of cancer patients, our study found that DL models surpass ML models in predicting time to CVD occurrence. […] Despite recent studies revealing the major role of SDoH in CVD, only a handful of prediction models incorporated a wide range of SDoH variables. […] In 80% of the studies, gender-stratified prediction was overlooked despite gender playing a role in CVD presentation, diagnosis, and survival. […] AI-based risk prediction models have an increased discrimination ability and accuracy as compared to the conventional multivariable models. […] To improve CVD risk prediction and inform clinicians decision-making future studies need to assess SDoH, in addition to the traditional factors and other emerging risk factors. […] Therefore, it is crucial to consider gender differences when it comes to predicting and managing CVD risks.

• #3 A deep learning-based electrocardiogram risk score for long term cardiovascular death and disease | npj Digital Medicine

  https://www.nature.com/articles/s41746-023-00916-6

  We envision SEER being used alongside the PCE score to evaluate cardiovascular risk in ambulatory settings including in outpatient clinics and on wearable devices. […] SEER predicted 5-year cardiovascular mortality with areas under the receiver operator characteristic curve (AUC) of 0.83 (95% CI: 0.810.85), 0.78 (0.770.80), and 0.83 (0.820.83) at Stanford, Cedars-Sinai, and Columbia respectively. […] Patients in the top third of the SEER score were at higher risk for developing a range of incident cardiovascular diseases. […] SEER therefore reclassified 16% of patients classified as low risk by PCE into a moderate risk category, identifying additional patients who may benefit from statin therapy. […] Given the advantages of ECGs over CAC scanslower expense, lack of radiation, clinical ubiquity, and presence in resource-limited environments SEER or similar models could potentially present an attractive alternative after further validation. […] Our risk score, SEER, identifies groups of patients to be up-risked and successfully stratifies intermediate risk patients for risk of ASCVD and cardiovascular mortality.

Zobacz więcej