Search

Background: In suspected myocardial infarction (MI), guidelines recommend using high-sensitivity cardiac troponin (hscTn)- based approaches. These require fixed assay-specific thresholds and timepoints, without directly integrating clinical information. Using machine-learning techniques including hs-cTn and clinical routine variables, we aimed to build a digital tool to directly estimate the individual probability of MI, allowing for numerous hs-cTn assays. Methods: In 2,575 patients presenting to the emergency department with suspected MI, two ensembles of machine-learning models using single or serial concentrations of six different hs-cTn assays were derived to estimate the individual MI probability ( ARTEMIS model). Discriminative performance of the models was assessed using area under the receiver operating characteristic curve (AUC) and logLoss. Model performance was validated in an external cohort with 1688 patients and tested for global generalizability in 13 international cohorts with 23,411 patients. Results: Eleven routinely available variables including age, sex, cardiovascular risk factors, electrocardiography, and hs-cTn were included in the ARTEMIS models. In the validation and generalization cohorts, excellent discriminative performance was confirmed, superior to hs-cTn only. For the serial hs-cTn measurement model, AUC ranged from 0.92 to 0.98. Good calibration was observed. Using a single hs-cTn measurement, the ARTEMIS model allowed direct rule-out of MI with very high and similar safety but up to tripled efficiency compared to the guideline- recommended strategy. Conclusion We developed and validated diagnostic models to accurately estimate the individual probability of MI, which allow for variable hs-cTn use and flexible timing of resampling. Their digital application may provide rapid, safe and efficient personalized patient care.

Background: In suspected myocardial infarction (MI), guidelines recommend using high-sensitivity cardiac troponin (hscTn)- based approaches. These require fixed assay-specific thresholds and timepoints, without directly integrating clinical information. Using machine-learning techniques including hs-cTn and clinical routine variables, we aimed to build a digital tool to directly estimate the individual probability of MI, allowing for numerous hs-cTn assays. Methods: In 2,575 patients presenting to the emergency department with suspected MI, two ensembles of machine-learning models using single or serial concentrations of six different hs-cTn assays were derived to estimate the individual MI probability ( ARTEMIS model). Discriminative performance of the models was assessed using area under the receiver operating characteristic curve (AUC) and logLoss. Model performance was validated in an external cohort with 1688 patients and tested for global generalizability in 13 international cohorts with 23,411 patients. Results: Eleven routinely available variables including age, sex, cardiovascular risk factors, electrocardiography, and hs-cTn were included in the ARTEMIS models. In the validation and generalization cohorts, excellent discriminative performance was confirmed, superior to hs-cTn only. For the serial hs-cTn measurement model, AUC ranged from 0.92 to 0.98. Good calibration was observed. Using a single hs-cTn measurement, the ARTEMIS model allowed direct rule-out of MI with very high and similar safety but up to tripled efficiency compared to the guideline- recommended strategy. Conclusion We developed and validated diagnostic models to accurately estimate the individual probability of MI, which allow for variable hs-cTn use and flexible timing of resampling. Their digital application may provide rapid, safe and efficient personalized patient care.

Background: In suspected myocardial infarction (MI), guidelines recommend using high-sensitivity cardiac troponin (hscTn)- based approaches. These require fixed assay-specific thresholds and timepoints, without directly integrating clinical information. Using machine-learning techniques including hs-cTn and clinical routine variables, we aimed to build a digital tool to directly estimate the individual probability of MI, allowing for numerous hs-cTn assays. Methods: In 2,575 patients presenting to the emergency department with suspected MI, two ensembles of machine-learning models using single or serial concentrations of six different hs-cTn assays were derived to estimate the individual MI probability ( ARTEMIS model). Discriminative performance of the models was assessed using area under the receiver operating characteristic curve (AUC) and logLoss. Model performance was validated in an external cohort with 1688 patients and tested for global generalizability in 13 international cohorts with 23,411 patients. Results: Eleven routinely available variables including age, sex, cardiovascular risk factors, electrocardiography, and hs-cTn were included in the ARTEMIS models. In the validation and generalization cohorts, excellent discriminative performance was confirmed, superior to hs-cTn only. For the serial hs-cTn measurement model, AUC ranged from 0.92 to 0.98. Good calibration was observed. Using a single hs-cTn measurement, the ARTEMIS model allowed direct rule-out of MI with very high and similar safety but up to tripled efficiency compared to the guideline- recommended strategy. Conclusion We developed and validated diagnostic models to accurately estimate the individual probability of MI, which allow for variable hs-cTn use and flexible timing of resampling. Their digital application may provide rapid, safe and efficient personalized patient care.

Background: In suspected myocardial infarction (MI), guidelines recommend using high-sensitivity cardiac troponin (hscTn)- based approaches. These require fixed assay-specific thresholds and timepoints, without directly integrating clinical information. Using machine-learning techniques including hs-cTn and clinical routine variables, we aimed to build a digital tool to directly estimate the individual probability of MI, allowing for numerous hs-cTn assays. Methods: In 2,575 patients presenting to the emergency department with suspected MI, two ensembles of machine-learning models using single or serial concentrations of six different hs-cTn assays were derived to estimate the individual MI probability ( ARTEMIS model). Discriminative performance of the models was assessed using area under the receiver operating characteristic curve (AUC) and logLoss. Model performance was validated in an external cohort with 1688 patients and tested for global generalizability in 13 international cohorts with 23,411 patients. Results: Eleven routinely available variables including age, sex, cardiovascular risk factors, electrocardiography, and hs-cTn were included in the ARTEMIS models. In the validation and generalization cohorts, excellent discriminative performance was confirmed, superior to hs-cTn only. For the serial hs-cTn measurement model, AUC ranged from 0.92 to 0.98. Good calibration was observed. Using a single hs-cTn measurement, the ARTEMIS model allowed direct rule-out of MI with very high and similar safety but up to tripled efficiency compared to the guideline- recommended strategy. Conclusion We developed and validated diagnostic models to accurately estimate the individual probability of MI, which allow for variable hs-cTn use and flexible timing of resampling. Their digital application may provide rapid, safe and efficient personalized patient care.

Background: In suspected myocardial infarction (MI), guidelines recommend using high-sensitivity cardiac troponin (hscTn)- based approaches. These require fixed assay-specific thresholds and timepoints, without directly integrating clinical information. Using machine-learning techniques including hs-cTn and clinical routine variables, we aimed to build a digital tool to directly estimate the individual probability of MI, allowing for numerous hs-cTn assays. Methods: In 2,575 patients presenting to the emergency department with suspected MI, two ensembles of machine-learning models using single or serial concentrations of six different hs-cTn assays were derived to estimate the individual MI probability ( ARTEMIS model). Discriminative performance of the models was assessed using area under the receiver operating characteristic curve (AUC) and logLoss. Model performance was validated in an external cohort with 1688 patients and tested for global generalizability in 13 international cohorts with 23,411 patients. Results: Eleven routinely available variables including age, sex, cardiovascular risk factors, electrocardiography, and hs-cTn were included in the ARTEMIS models. In the validation and generalization cohorts, excellent discriminative performance was confirmed, superior to hs-cTn only. For the serial hs-cTn measurement model, AUC ranged from 0.92 to 0.98. Good calibration was observed. Using a single hs-cTn measurement, the ARTEMIS model allowed direct rule-out of MI with very high and similar safety but up to tripled efficiency compared to the guideline- recommended strategy. Conclusion We developed and validated diagnostic models to accurately estimate the individual probability of MI, which allow for variable hs-cTn use and flexible timing of resampling. Their digital application may provide rapid, safe and efficient personalized patient care.

Aims We aimed to evaluate the impact of age on the performance of the European Society of Cardiology (ESC) 0/1h-algorithms and to derive and externally validate alternative cut-offs specific to older patients. Methods and results We prospectively enrolled patients presenting to the emergency department (ED) with symptoms suggestive of and results acute myocardial infarction in three large diagnostic studies. Final diagnoses were adjudicated by two independent cardiologists. High-sensitivity cardiac troponin (hs-cTn) T and I concentrations were measured at presentation and after 1 h. Patients were stratified according to age [<55 years (young), >= 55 to <70 years (middle-age), >= 70 years (old)]. Rule-out safety of the ESC hs-cTnT 0/1h-algorithm was very high in all age-strata: sensitivity 100% [95% confidence interval (95% CI) 94.9-100] in young, 99.3% (95% CI 96.0-99.9) in middle-age, and 99.3% (95% CI 97.599.8) in old patients. Accuracy of rule-in decreased with age: specificity 97.0% (95% CI 95.8-97.9) in young, 96.1% (95% CI 94.5-97.2) in middle-age, and 92.7% (95% CI 90.7-94.3) in older patients. Triage efficacy decreased with increasing age (young 93%, middle-age 80%, old 55%, P <0.001). Similar results were found for the ESC hs-cTnT 0/1h-algorithm. Alternative, slightly higher cut-off concentrations optimized for older patients maintained very high safety of rule-out, increased specificity of rule-in (P< 0.01), reduced overall efficacy for hs-cTnT (P <0.01), while maintaining efficacy for hs-cTnl. Findings were confirmed in two validation cohorts (n = 2767). Conclusion While safety of the ESC 0/1h-algorithms remained very high, increasing age significantly reduced overall efficacy and the accuracy of rule-in. Alternative slightly higher cut-off concentrations may be considered for older patients, particularly if using hs-cTnl.

AIMS: We aimed to evaluate the impact of age on the performance of the European Society of Cardiology (ESC) 0/1h-algorithms and to derive and externally validate alternative cut-offs specific to older patients. METHODS AND RESULTS: We prospectively enrolled patients presenting to the emergency department (ED) with symptoms suggestive of acute myocardial infarction in three large diagnostic studies. Final diagnoses were adjudicated by two independent cardiologists. High-sensitivity cardiac troponin (hs-cTn) T and I concentrations were measured at presentation and after 1 h. Patients were stratified according to age [<55 years (young), ≥55 to <70 years (middle-age), ≥70 years (old)]. Rule-out safety of the ESC hs-cTnT 0/1h-algorithm was very high in all age-strata: sensitivity 100% [95% confidence interval (95% CI) 94.9-100] in young, 99.3% (95% CI 96.0-99.9) in middle-age, and 99.3% (95% CI 97.5-99.8) in old patients. Accuracy of rule-in decreased with age: specificity 97.0% (95% CI 95.8-97.9) in young, 96.1% (95% CI 94.5-97.2) in middle-age, and 92.7% (95% CI 90.7-94.3) in older patients. Triage efficacy decreased with increasing age (young 93%, middle-age 80%, old 55%, P < 0.001). Similar results were found for the ESC hs-cTnT 0/1h-algorithm. Alternative, slightly higher cut-off concentrations optimized for older patients maintained very high safety of rule-out, increased specificity of rule-in (P < 0.01), reduced overall efficacy for hs-cTnT (P < 0.01), while maintaining efficacy for hs-cTnI. Findings were confirmed in two validation cohorts (n = 2767). CONCLUSION: While safety of the ESC 0/1h-algorithms remained very high, increasing age significantly reduced overall efficacy and the accuracy of rule-in. Alternative slightly higher cut-off concentrations may be considered for older patients, particularly if using hs-cTnI. CLINICAL TRIAL REGISTRATION: https://clinicaltrials.gov/ct2/show/NCT00470587, number NCT00470587 and NCT02355457 (BACC).

AIMS: Serial measurements of high-sensitivity troponin are used to rule out acute myocardial infarction (AMI) with an assay specific cutoff at the 99th percentile. Here, we evaluated the performance of a single admission troponin with a lower cutoff combined with a low risk electrocardiogram (ECG) to rule out AMI. METHODS: Troponin I measured with a high-sensitivity assay (hs-TnI) was determined at admission in 1040 patients presenting with suspected AMI (BACC study). To rule out AMI we calculated the negative predictive value (NPV) utilizing the optimal hs-TnI cutoff combined with a low risk ECG. The results were validated in 3566 patients with suspected AMI [2-h Accelerated Diagnostic Protocol to Assess Patients With Chest Pain Symptoms Using Contemporary Troponins as the Only Biomarker (ADAPT) studies]. Patients were followed for 6 or 12 months. RESULTS: 184 of all patients were diagnosed with AMI. An hs-TnI cutoff of 3 ng/L resulted in a NPV of 99.3% (CI 97.3-100.0), ruling out 35% of all non-AMI patients. Adding the information of a low riskECGresulted in a 100% (CI 97.5-100.0) NPV (28% ruled out). The 2 validation cohorts replicated the high NPV of this approach. The follow-up mortality in the ruled out population was low (0 deaths in BACC and Stenocardia, 1 death in ADAPT). CONCLUSIONS: A single hs-TnI measurement on admission combined with a low risk ECG appears to rule out AMI safely without need for serial troponin testing. Trial Registration: www.clinicaltrials.gov (NCT02355457).

Importance Rapid and accurate diagnosis of acute myocardial infarction (AMI) currently constitutes an unmet need.Objective To test a 1-hour diagnostic algorithm to diagnose AMI using a high-sensitivity troponin I assay with a new cutoff level of 6 ng/L.Design, Setting, and Participants The Biomarkers in Acute Cardiac Care study is a prospective study that investigated the application of the troponin I assay for the diagnosis of AMI in 1040 patients presenting to the emergency department with acute chest pain from July 19, 2013, to December 31, 2014. Results were validated in 2 independent cohorts of 4009 patients. Final follow-up was completed on July 1, 2015, and data were assessed from July 2 to December 15, 2015.Exposure Acute chest pain suggestive of AMI.Main Outcomes and Measures Accurate diagnosis or exclusion of AMI and 12-month mortality in patients with acute chest pain.Results Of the 1040 patients included from the study cohort, 673 (64.7%) were male and had a median age of 65 (interquartile range, 52-75) years. With application of a low troponin I cutoff value of 6 ng/L, the rule-out algorithm showed a high negative predictive value of 99.8% (95% CI, 98.6%-100.0%) after 1 hour for non–ST-segment elevation MI type 1. The 1-hour approach was comparable to a 3-hour approach. Similarly, a rule-in algorithm based on troponin I levels provided a high positive predictive value with 82.8% (95% CI, 73.2%-90.0%). Moreover, application of the cutoff of 6 ng/L resulted in lower follow-up mortality (1.0%) compared with the routinely used 99th percentile (3.7%) for this assay. Two independent cohorts further validated the performance of this algorithm with high negative and positive predictive values.Conclusions and Relevance Patients with possible AMI can be triaged within 1 hour after admission with