A neural computational aid to the diagnosis of acute myocardial infarction

Diagnostic accuracy of signs and symptoms in acute coronary syndrome and acute myocardial infarction

BACKGROUND

Acute coronary syndrome (ACS) and acute myocardial infarction (AMI) account for a large portion of cardiovascular deaths. Signs and symptoms for these syndromes, such as chest pain, are non-specific and can be caused by a variety of non-cardiac conditions, especially in low-prevalence settings such as general practice. The diagnostic value of these signs and symptoms can be assessed using diagnostic meta-analyses, but the last one dates from 2012.

METHODS

We performed a diagnostic meta-analysis in accordance with PRISMA guidelines. We searched PubMed, Embase and CENTRAL from 2006 to 2024. We included studies that assessed the diagnostic accuracy of thirteen different signs and symptoms. We divided patients into two subgroups (AMI and ACS) on which analysis was performed independently.

RESULTS

We selected 24 articles for inclusion. Our analysis indicates that signs and symptoms have a limited role in the diagnosis of AMI or ACS. The most useful (highest diagnostic odds ratios, DOR) in the diagnosis of AMI were pain radiating to both arms (DOR 2.95 (95%CI 1.57-5.06)), absence of chest wall tenderness (DOR 3.51 (95%CI 1.64-6.61)), pain radiating to the right arm (DOR 5.17 (95%CI 1.77-11.9)) and sweating (DOR 5.75 (95%CI 2.51-11.4)). For ACS these were pain radiating to the right arm (DOR 3.9 (95%CI 0.7-12.6)) and absence of chest wall tenderness (DOR 7.73 (95%CI 2.19-19.8)).

CONCLUSION

We report the accuracy of thirteen signs and symptoms in the diagnosis of AMI and ACS. These can be useful to calibrate general practitioners' diagnostic assessment of chest pain in primary care settings.

Diversity in Machine Learning: A Systematic Review of Text-Based Diagnostic Applications

OBJECTIVE

 As the storage of clinical data has transitioned into electronic formats, medical informatics has become increasingly relevant in providing diagnostic aid. The purpose of this review is to evaluate machine learning models that use text data for diagnosis and to assess the diversity of the included study populations.

METHODS

 We conducted a systematic literature review on three public databases. Two authors reviewed every abstract for inclusion. Articles were included if they used or developed machine learning algorithms to aid in diagnosis. Articles focusing on imaging informatics were excluded.

RESULTS

 From 2,260 identified papers, we included 78. Of the machine learning models used, neural networks were relied upon most frequently (44.9%). Studies had a median population of 661.5 patients, and diseases and disorders of 10 different body systems were studied. Of the 35.9% (N = 28) of papers that included race data, 57.1% (N = 16) of study populations were majority White, 14.3% were majority Asian, and 7.1% were majority Black. In 75% (N = 21) of papers, White was the largest racial group represented. Of the papers included, 43.6% (N = 34) included the sex ratio of the patient population.

DISCUSSION

 With the power to build robust algorithms supported by massive quantities of clinical data, machine learning is shaping the future of diagnostics. Limitations of the underlying data create potential biases, especially if patient demographics are unknown or not included in the training.

CONCLUSION

 As the movement toward clinical reliance on machine learning accelerates, both recording demographic information and using diverse training sets should be emphasized. Extrapolating algorithms to demographics beyond the original study population leaves large gaps for potential biases.

Real-time frequency-independent single-Lead and single-beat myocardial infarction detection

This study proposes a novel real-time frequency-independent myocardial infarction detector for Lead II electrocardiograms. The underlying Deep-LSTM network is trained using the PTB-XL database, the largest to date publicly available electrocardiography dataset, and is tested over the same and the older PTB database. By testing the model over distinct datasets, collected under different conditions and from different patients, a more realistic measure of the performance can be gauged from the deployed system. The detector is trained over 3589 myocardial infarction (MI) patients and 7115 healthy controls (HC) while it is evaluated on 1076 MIs and 1840 HCs. The proposed algorithm, achieved an accuracy of 77.12%, recall/sensitivity of 75.85%, and a specificity of 83.02% over the entire PTB database; 85.07%, 81.54%, 87.31% over the PTB-XL validation set (fold 9), and 84.17%, 78.37%, 87.55% over the PTB-XL test set (fold 10). The model also achieves stable performance metrics over the frequency range of 202 Hz to 2.8 kHz. The processing time is dependent on the sampling frequency, ranging from 130 ms at 202 Hz to 1.8 s at 2.8 kHz. Such outcome is within the time required for real-time processing (less than 300 ms for fast heartbeats), between 202 Hz and 500 Hz making the algorithm practically real-time. Therefore, the proposed MI detector could be readily deployed onto existing wearable and/or portable devices and test instruments; potentially having significant societal and clinical impact in the lives of patients at risk for myocardial infarction.

Applications of machine learning to undifferentiated chest pain in the emergency department: A systematic review

BACKGROUND

Chest pain is amongst the most common reason for presentation to the emergency department (ED). There are many causes of chest pain, and it is important for the emergency physician to quickly and accurately diagnose life threatening causes such as acute myocardial infarction (AMI). Multiple clinical decision tools have been developed to assist clinicians in risk stratifying patients with chest. There is growing recognition that machine learning (ML) will have a significant impact on the practice of medicine in the near future and may assist with diagnosis and risk stratification. This systematic review aims to evaluate how ML has been applied to adults presenting to the ED with undifferentiated chest pain and assess if ML models show improved performance when compared to physicians or current risk stratification techniques.

METHODS AND FINDINGS

We conducted a systematic review of journal articles that applied a ML technique to an adult patient presenting to an emergency department with undifferentiated chest pain. Multiple databases were searched from inception through to November 2020. In total, 3361 articles were screened, and 23 articles were included. We did not conduct a metanalysis due to a high level of heterogeneity between studies in both their methods, and reporting. The most common primary outcomes assessed were diagnosis of acute myocardial infarction (AMI) (12 studies), and prognosis of major adverse cardiovascular event (MACE) (6 studies). There were 14 retrospective studies and 5 prospective studies. Four studies reported the development of a machine learning model retrospectively then tested it prospectively. The most common machine learning methods used were artificial neural networks (14 studies), random forest (6 studies), support vector machine (5 studies), and gradient boosting (2 studies). Multiple studies achieved high accuracy in both the diagnosis of AMI in the ED setting, and in predicting mortality and composite outcomes over various timeframes. ML outperformed existing risk stratification scores in all cases, and physicians in three out of four cases. The majority of studies were single centre, retrospective, and without prospective or external validation. There were only 3 studies that were considered low risk of bias and had low applicability concerns. Two studies reported integrating the ML model into clinical practice.

CONCLUSIONS

Research on applications of ML for undifferentiated chest pain in the ED has been ongoing for decades. ML has been reported to outperform emergency physicians and current risk stratification tools to diagnose AMI and predict MACE but has rarely been integrated into practice. Many studies assessing the use of ML in undifferentiated chest pain in the ED have a high risk of bias. It is important that future studies make use of recently developed standardised ML reporting guidelines, register their protocols, and share their datasets and code. Future work is required to assess the impact of ML model implementation on clinical decision making, patient orientated outcomes, and patient and physician acceptability.

TRIAL REGISTRATION

International Prospective Register of Systematic Reviews registration number: CRD42020184977.

Artificial intelligence in disease diagnostics: A critical review and classification on the current state of research guiding future direction

The diagnosis of diseases is decisive for planning proper treatment and ensuring the well-being of patients. Human error hinders accurate diagnostics, as interpreting medical information is a complex and cognitively challenging task. The application of artificial intelligence (AI) can improve the level of diagnostic accuracy and efficiency. While the current literature has examined various approaches to diagnosing various diseases, an overview of fields in which AI has been applied, including their performance aiming to identify emergent digitalized healthcare services, has not yet been adequately realized in extant research. By conducting a critical review, we portray the AI landscape in diagnostics and provide a snapshot to guide future research. This paper extends academia by proposing a research agenda. Practitioners understand the extent to which AI improves diagnostics and how healthcare benefits from it. However, several issues need to be addressed before successful application of AI in disease diagnostics can be achieved.

Predicting women with depressive symptoms postpartum with machine learning methods

Postpartum depression (PPD) is a detrimental health condition that affects 12% of new mothers. Despite negative effects on mothers’ and children’s health, many women do not receive adequate care. Preventive interventions are cost-efficient among high-risk women, but our ability to identify these is poor. We leveraged the power of clinical, demographic, and psychometric data to assess if machine learning methods can make accurate predictions of postpartum depression. Data were obtained from a population-based prospective cohort study in Uppsala, Sweden, collected between 2009 and 2018 (BASIC study, n = 4313). Sub-analyses among women without previous depression were performed. The extremely randomized trees method provided robust performance with highest accuracy and well-balanced sensitivity and specificity (accuracy 73%, sensitivity 72%, specificity 75%, positive predictive value 33%, negative predictive value 94%, area under the curve 81%). Among women without earlier mental health issues, the accuracy was 64%. The variables setting women at most risk for PPD were depression and anxiety during pregnancy, as well as variables related to resilience and personality. Future clinical models that could be implemented directly after delivery might consider including these variables in order to identify women at high risk for postpartum depression to facilitate individualized follow-up and cost-effectiveness.

Patient Phenotypes, Cardiovascular Risk, and Ezetimibe Treatment in Patients After Acute Coronary Syndromes (from IMPROVE-IT)

Risk prediction following acute coronary syndrome (ACS) remains challenging. Data-driven machine-learning algorithms can potentially identify patients at high risk of clinical events. The Improved Reduction of Outcomes: Vytorin Efficacy International Trial randomized 18,144 post-ACS patients to ezetimibe + simvastatin or placebo + simvastatin. We performed hierarchical cluster analysis to identify patients at high risk of adverse events. Associations between clusters and outcomes were assessed using Cox proportional hazards models. The primary outcome was cardiovascular death, nonfatal myocardial infarction, nonfatal stroke, unstable angina hospitalization, or coronary revascularization ≥30 days after randomization. We evaluated ezetimibe's impact on outcomes across clusters and the ability of the cluster analysis to discriminate for outcomes compared with the Global Registry of Acute Coronary Events (GRACE) score. Five clusters were identified. In cluster 1 (n = 13,252), most patients experienced a non-STEMI (54.8%). Cluster 2 patients (n = 2,719) had the highest incidence of unstable angina (n = 83.3%). Cluster 3 patients (n = 782) all identified as Spanish descent, whereas cluster 4 patients (n = 803) were primarily from South America (56.2%). In cluster 5 (n = 587), all patients had ST elevation. Cluster analysis identified patients at high risk of adverse outcomes (log-rank p <0.0001); Cluster 2 (vs 1) patients had the highest risk of outcomes (hazards ratio 1.33, 95% confidence interval 1.24 to 1.43). Compared with GRACE risk, cluster analysis did not provide superior outcome discrimination. A consistent ezetimibe treatment effect was identified across clusters (interaction p = 0.882). In conclusion, cluster analysis identified significant difference in risk of outcomes across cluster groups. Data-driven strategies to identify patients who may differentially benefit from therapies and for risk stratification require further evaluation.

Comparative Study of Back Propagation Artificial Neural Networks and Logistic Regression Model in Predicting Poor Prognosis after Acute Ischemic Stroke

Objective

To investigate the predictive value of clinical variables on the poor prognosis at 90-day follow-up from acute stroke onset, and compare the diagnostic performance between back propagation artificial neural networks (BP ANNs) and Logistic regression (LR) models in predicting the prognosis.

Methods

We studied the association between clinical variables and the functional recovery of 435 acute ischemic stroke patients. The patients were divided into 2 groups according to modified Rankin Scale scores evaluated on the 90th day after stroke onset. Both BP ANNs and LR models were established for predicting the poor outcome and their diagnostic performance were compared by receiver operating curve.

Results

Age, free fatty acid, homocysteine and alkaline phosphatase were closely related with the poor outcome in acute ischemic stroke patients and finally enrolled in models. The accuracy, sensitivity and specificity of BP ANNs were 80.15%, 75.64% and 82.07% respectively. For the LR model, the accuracy, sensitivity and specificity was 70.61%, 88.46% and 63.04% respectively. The area under the ROC curve of the BP ANNs and LR model was 0.881and 0.809.

Conclusions

Both BP ANNs and LR model were promising for the prediction of poor outcome by combining age, free fatty acid, homocysteine and alkaline phosphatase. However, BP ANNs model showed better performance than LR model in predicting the prognosis.

A Review of Automated Methods for Detection of Myocardial Ischemia and Infarction Using Electrocardiogram and Electronic Health Records

There is a growing body of research focusing on automatic detection of ischemia and myocardial infarction (MI) using computer algorithms. In clinical settings, ischemia and MI are diagnosed using electrocardiogram (ECG) recordings as well as medical context including patient symptoms, medical history, and risk factors-information that is often stored in the electronic health records. The ECG signal is inspected to identify changes in the morphology such as ST-segment deviation and T-wave changes. Some of the proposed methods compute similar features automatically while others use nonconventional features such as wavelet coefficients. This review provides an overview of the methods that have been proposed in this area, focusing on their historical evolution, the publicly available datasets that they have used to evaluate their performance, and the details of their algorithms for ECG and EHR analysis. The validation strategies that have been used to evaluate the performance of the proposed methods are also presented. Finally, the paper provides recommendations for future research to address the shortcomings of the currently existing methods and practical considerations to make the proposed technical solutions applicable in clinical practice.

[Intelligent systems tools in the diagnosis of acute coronary syndromes: A systemic review]

BACKGROUND

Acute myocardial infarction is the leading cause of non-communicable deaths worldwide. Its diagnosis is a highly complex task, for which modelling through automated methods has been attempted. A systematic review of the literature was performed on diagnostic tests that applied intelligent systems tools in the diagnosis of acute coronary syndromes.

METHODS

A systematic review of the literature is presented using Medline, Embase, Scopus, IEEE/IET Electronic Library, ISI Web of Science, Latindex and LILACS databases for articles that include the diagnostic evaluation of acute coronary syndromes using intelligent systems. The review process was conducted independently by 2 reviewers, and discrepancies were resolved through the participation of a third person. The operational characteristics of the studied tools were extracted.

RESULTS

A total of 35 references met the inclusion criteria. In 22 (62.8%) cases, neural networks were used. In five studies, the performances of several intelligent systems tools were compared. Thirteen studies sought to perform diagnoses of all acute coronary syndromes, and in 22, only infarctions were studied. In 21 cases, clinical and electrocardiographic aspects were used as input data, and in 10, only electrocardiographic data were used. Most intelligent systems use the clinical context as a reference standard. High rates of diagnostic accuracy were found with better performance using neural networks and support vector machines, compared with statistical tools of pattern recognition and decision trees.

CONCLUSIONS

Extensive evidence was found that shows that using intelligent systems tools achieves a greater degree of accuracy than some clinical algorithms or scales and, thus, should be considered appropriate tools for supporting diagnostic decisions of acute coronary syndromes.

Using neural networks and just nine patient-reportable factors of screen for AMI

The study investigated the effect of different input selections on the performance of artificial neural networks in screening for acute myocardial infarction (AMI) in Malaysian patients complaining of chest pain. We used hospital data to create neural networks with four input selections and used these to diagnose AMI. A 10-fold cross-validation and committee approach was used. All the neural networks using various input selections outperformed a multiple logistic regression model, although the difference was not statistically significant. The neural networks achieved an area under the ROC curve of 0.792 using nine inputs, whereas multiple logistic regression achieved 0.739 using 64 inputs. Sensitivity levels of over 90 per cent were achieved using low output threshold levels. Specificity levels of over 90 per cent were achieved using threshold levels of 0.4-0.5. Thus neural networks can perform as well as multiple logistic regression models even when using far fewer inputs.

An Artificial Neural Network Stratifies the Risks of Reintervention and Mortality after Endovascular Aneurysm Repair; a Retrospective Observational study

Background

Lifelong surveillance after endovascular repair (EVAR) of abdominal aortic aneurysms (AAA) is considered mandatory to detect potentially life-threatening endograft complications. A minority of patients require reintervention but cannot be predictively identified by existing methods. This study aimed to improve the prediction of endograft complications and mortality, through the application of machine-learning techniques.

Methods

Patients undergoing EVAR at 2 centres were studied from 2004-2010. Pre-operative aneurysm morphology was quantified and endograft complications were recorded up to 5 years following surgery. An artificial neural networks (ANN) approach was used to predict whether patients would be at low- or high-risk of endograft complications (aortic/limb) or mortality. Centre 1 data were used for training and centre 2 data for validation. ANN performance was assessed by Kaplan-Meier analysis to compare the incidence of aortic complications, limb complications, and mortality; in patients predicted to be low-risk, versus those predicted to be high-risk.

Results

761 patients aged 75 +/- 7 years underwent EVAR. Mean follow-up was 36+/- 20 months. An ANN was created from morphological features including angulation/length/areas/diameters/volume/tortuosity of the aneurysm neck/sac/iliac segments. ANN models predicted endograft complications and mortality with excellent discrimination between a low-risk and high-risk group. In external validation, the 5-year rates of freedom from aortic complications, limb complications and mortality were 95.9% vs 67.9%; 99.3% vs 92.0%; and 87.9% vs 79.3% respectively (p<0.001)

Conclusion

This study presents ANN models that stratify the 5-year risk of endograft complications or mortality using routinely available pre-operative data.

U.S. Counties’ Vulnerability to Methamphetamine Labs

This national study analyzes county-level risk factors for methamphetamine manufacture. Neural network and probit models are used to test the effectiveness of county-level characteristics in predicting methamphetamine production levels. Data on all 3,143 counties are drawn from the U.S. DEA’s Clandestine Laboratory Surveillance System, the 2000 U.S. Census and health service resources from the 2004 Area Resource File, and the Uniform Crime Reporting Program (UCRP) for the period 2002-2005. The resulting model accurately predicted methamphetamine production levels 85% of the time. The leading variables were existing methamphetamine problems, seizures in contiguous counties, families with “female head of household,” home value, and “percentage of White population.” Several variables that factored heavily in earlier single-community studies had very little impact in this national study. This study’s results suggest a new approach to assessing community vulnerability to drug manufacturer and a need to refocus efforts in fighting the problem.

Discrepancy between clinician and research assistant in TIMI score calculation (TRIAGED CPU)

Introduction

Several studies have attempted to demonstrate that the Thrombolysis in Myocardial Infarction (TIMI) risk score has the ability to risk stratify emergency department (ED) patients with potential acute coronary syndromes (ACS). Most of the studies we reviewed relied on trained research investigators to determine TIMI risk scores rather than ED providers functioning in their normal work capacity. We assessed whether TIMI risk scores obtained by ED providers in the setting of a busy ED differed from those obtained by trained research investigators.

Methods

This was an ED-based prospective observational cohort study comparing TIMI scores obtained by 49 ED providers admitting patients to an ED chest pain unit (CPU) to scores generated by a team of trained research investigators. We examined provider type, patient gender, and TIMI elements for their effects on TIMI risk score discrepancy.

Results

Of the 501 adult patients enrolled in the study, 29.3% of TIMI risk scores determined by ED providers and trained research investigators were generated using identical TIMI risk score variables. In our low-risk population the majority of TIMI risk score differences were small; however, 12% of TIMI risk scores differed by two or more points.

Conclusion

TIMI risk scores determined by ED providers in the setting of a busy ED frequently differ from scores generated by trained research investigators who complete them while not under the same pressure of an ED provider.

Long-term outcome after CT angiography in patients with possible acute coronary syndrome

PURPOSE

To assess the long-term outcome and hospital readmission rate associated with a computed tomographic (CT) angiography-guided strategy used to examine patients who present to the emergency department (ED) with symptoms of possible acute coronary syndrome (ACS).

MATERIALS AND METHODS

The study was approved by the institutional review board, and all patients provided written informed consent. A total of 585 consecutive patients (mean age, 58 years ± 11 [standard deviation]; 58% were male) with ischemic-type chest pain and low to intermediate risk for ACS were evaluated prospectively. Patients underwent coronary CT angiography after single or serial troponin I (TnI) measurement, depending on time of presentation to the ED. Subsequent care was determined with CT angiography findings: Patients without plaque and patients with nonobstructive plaque and at most mild to moderate stenosis (<40% luminal narrowing) were discharged without further investigation. Patients with moderate stenosis (40%-70% narrowing) were discharged and referred for outpatient stress echocardiography. Patients with severe stenosis (>70% narrowing) were admitted. Discharged patients were contacted and their medical records were reviewed to determine rates of death, ACS, revascularization, and hospital admission. By using binomial distribution, Clopper-Pearson confidence intervals (CIs) were calculated for outcome data.

RESULTS

Coronary CT angiography findings were as follows: A total of 196 patients (34%) had no coronary plaque or stenosis, 288 (49%) had nonobstructive plaque, 22 (4%) had moderate stenosis, and 79 (13%) had severe stenosis. At median 47.4-month follow-up (range, 24-57 months) of the 506 discharged patients, five (1%; 95% CI: 0.4%, 2.3%) had been readmitted for chest pain; there were no instances of coronary revascularization, ACS, or death (0% for all; 95% CI: 0%, 0.7%). Follow-up was 100% complete.

CONCLUSION

Use of a CT angiography-guided strategy to investigate patients with low to intermediate risk of ACS who present to the ED with chest pain is safe at long-term follow-up, including patients discharged after single TnI measurement.

Knowledge discovery in clinical decision support systems for pain management: a systematic review

OBJECTIVE

The occurrence of pain accounts for billions of dollars in annual medical expenditures; loss of quality of life and decreased worker productivity contribute to indirect costs. As pain is highly subjective, clinical decision support systems (CDSSs) can be critical for improving the accuracy of pain assessment and offering better support for clinical decision-making. This review is focused on computer technologies for pain management that allow CDSSs to obtain knowledge from the clinical data produced by either patients or health care professionals.

METHODS AND MATERIALS

A comprehensive literature search was conducted in several electronic databases to identify relevant articles focused on computerised systems that constituted CDSSs and include data or results related to pain symptoms from patients with acute or chronic pain, published between 1992 and 2011 in the English language. In total, thirty-nine studies were analysed; thirty-two were selected from 1245 citations, and seven were obtained from reference tracking.

RESULTS

The results highlighted the following clusters of computer technologies: rule-based algorithms, artificial neural networks, nonstandard set theory, and statistical learning algorithms. In addition, several methodologies were found for content processing such as terminologies, questionnaires, and scores. The median accuracy ranged from 53% to 87.5%.

CONCLUSIONS

Computer technologies that have been applied in CDSSs are important but not determinant in improving the systems' accuracy and the clinical practice, as evidenced by the moderate correlation among the studies. However, these systems play an important role in the design of computerised systems oriented to a patient's symptoms as is required for pain management. Several limitations related to CDSSs were observed: the lack of integration with mobile devices, the reduced use of web-based interfaces, and scarce capabilities for data to be inserted by patients.

A clinical decision support system for femoral peripheral arterial disease treatment

One of the major challenges of providing reliable healthcare services is to diagnose and treat diseases in an accurate and timely manner. Recently, many researchers have successfully used artificial neural networks as a diagnostic assessment tool. In this study, the validation of such an assessment tool has been developed for treatment of the femoral peripheral arterial disease using a radial basis function neural network (RBFNN). A data set for training the RBFNN has been prepared by analyzing records of patients who had been treated by the thoracic and cardiovascular surgery clinic of a university hospital. The data set includes 186 patient records having 16 characteristic features associated with a binary treatment decision, namely, being a medical or a surgical one. K-means clustering algorithm has been used to determine the parameters of radial basis functions and the number of hidden nodes of the RBFNN is determined experimentally. For performance evaluation, the proposed RBFNN was compared to three different multilayer perceptron models having Pareto optimal hidden layer combinations using various performance indicators. Results of comparison indicate that the RBFNN can be used as an effective assessment tool for femoral peripheral arterial disease treatment.

Use of health information technology to reduce diagnostic errors

BACKGROUND

Health information technology (HIT) systems have the potential to reduce delayed, missed or incorrect diagnoses. We describe and classify the current state of diagnostic HIT and identify future research directions.

METHODS

A multi-pronged literature search was conducted using PubMed, Web of Science, backwards and forwards reference searches and contributions from domain experts. We included HIT systems evaluated in clinical and experimental settings as well as previous reviews, and excluded radiology computer-aided diagnosis, monitor alerts and alarms, and studies focused on disease staging and prognosis. Articles were organised within a conceptual framework of the diagnostic process and areas requiring further investigation were identified.

RESULTS

HIT approaches, tools and algorithms were identified and organised into 10 categories related to those assisting: (1) information gathering; (2) information organisation and display; (3) differential diagnosis generation; (4) weighing of diagnoses; (5) generation of diagnostic plan; (6) access to diagnostic reference information; (7) facilitating follow-up; (8) screening for early detection in asymptomatic patients; (9) collaborative diagnosis; and (10) facilitating diagnostic feedback to clinicians. We found many studies characterising potential interventions, but relatively few evaluating the interventions in actual clinical settings and even fewer demonstrating clinical impact.

CONCLUSIONS

Diagnostic HIT research is still in its early stages with few demonstrations of measurable clinical impact. Future efforts need to focus on: (1) improving methods and criteria for measurement of the diagnostic process using electronic data; (2) better usability and interfaces in electronic health records; (3) more meaningful incorporation of evidence-based diagnostic protocols within clinical workflows; and (4) systematic feedback of diagnostic performance.

Emergency physician estimates of the probability of acute coronary syndrome in a cohort of patients enrolled in a study of coronary computed tomographic angiography

INTRODUCTION

Little information exists regarding how accurately emergency physicians (EPs) predict the probability of acute coronary syndrome (ACS). Our objective was to determine if EPs can accurately predict ACS in a prospectively identified cohort of emergency department (ED) patients who met enrolment criteria for a study of coronary computed tomographic angiography (CCTA) and were admitted for a "rule out ACS" protocol.

METHODS

A prospective observational pilot study in an academic medical centre was carried out. EPs caring for patients with chest pain provided whole-number estimates of the probability of ACS after clinical review. This substudy was part of the now published Rule Out Myocardial Infarction/Ischemia Using Computer Assisted Tomography (ROMICAT) study, a study of CCTA and admission of patients for a rule out ACS protocol after a nondiagnostic evaluation. Predictions were grouped into probability groups based on the validated Goldman criteria. ACS was determined by an adjudication committee using American Heart Association/American College of Cardiology/European Society of Cardiology guidelines.

RESULTS

A total of 334 predictions were obtained for a study population with a mean age of 54 (SD 12) years, 63% of whom were male. There were 35 ACS events. EPs predicted ACS better than by chance, and increasingly higher estimates were associated with a higher incidence of ACS (p = 0.0004). The percentage of patients with ACS was 0%, 6%, 7%, and 17%, respectively, for very low, low, intermediate, and high probability groups. EPs' estimates had a sensitivity of 63% using a > 20% probability of ACS to define a positive test. Lowering this threshold to > 7% to define a test as positive increased the sensitivity of physician estimates to 89% but lowered specificity from 65% to 24%.

CONCLUSION

Our data suggest that for a selected ED cohort meeting eligibility criteria for a study of CCTA, EPs predict ACS better than by chance, with an increasing proportion of patients proving to have ACS with increasing probability estimates. Lowering the estimate threshold does not result in an overall sensitivity level that is sufficient to send patients home from the ED and is associated with a poor specificity.

CT angiography for safe discharge of patients with possible acute coronary syndromes

BACKGROUND

Admission rates among patients presenting to emergency departments with possible acute coronary syndromes are high, although for most of these patients, the symptoms are ultimately found not to have a cardiac cause. Coronary computed tomographic angiography (CCTA) has a very high negative predictive value for the detection of coronary disease, but its usefulness in determining whether discharge of patients from the emergency department is safe is not well established.

METHODS

We randomly assigned low-to-intermediate-risk patients presenting with possible acute coronary syndromes, in a 2:1 ratio, to undergo CCTA or to receive traditional care. Patients were enrolled at five centers in the United States. Patients older than 30 years of age with a Thrombolysis in Myocardial Infarction risk score of 0 to 2 and signs or symptoms warranting admission or testing were eligible. The primary outcome was safety, assessed in the subgroup of patients with a negative CCTA examination, with safety defined as the absence of myocardial infarction and cardiac death during the first 30 days after presentation.

RESULTS

We enrolled 1370 subjects: 908 in the CCTA group and 462 in the group receiving traditional care. The baseline characteristics were similar in the two groups. Of 640 patients with a negative CCTA examination, none died or had a myocardial infarction within 30 days (0%; 95% confidence interval [CI], 0 to 0.57). As compared with patients receiving traditional care, patients in the CCTA group had a higher rate of discharge from the emergency department (49.6% vs. 22.7%; difference, 26.8 percentage points; 95% CI, 21.4 to 32.2), a shorter length of stay (median, 18.0 hours vs. 24.8 hours; P<0.001), and a higher rate of detection of coronary disease (9.0% vs. 3.5%; difference, 5.6 percentage points; 95% CI, 0 to 11.2). There was one serious adverse event in each group.

CONCLUSIONS

A CCTA-based strategy for low-to-intermediate-risk patients presenting with a possible acute coronary syndrome appears to allow the safe, expedited discharge from the emergency department of many patients who would otherwise be admitted. (Funded by the Commonwealth of Pennsylvania Department of Health and the American College of Radiology Imaging Network Foundation; ClinicalTrials.gov number, NCT00933400.).

Acute chest pain investigation: utility of cardiac CT angiography in guiding troponin measurement

PURPOSE

To assess the impact on length of stay and rate of major adverse cardiovascular events of a cardiac computed tomographic (CT) angiography-guided algorithm to examine patients who present to the emergency department (ED) with low- to intermediate-risk chest pain.

MATERIALS AND METHODS

The study was approved by the institutional review board, and all patients gave written informed consent. Two hundred three consecutive patients (mean age, 55 years ± 11 [standard deviation]; 123 men) with low- to intermediate-risk ischemic-type chest pain were prospectively enrolled. Patients underwent initial cardiac CT angiography with subsequent treatment determined by reference to findings at cardiac CT angiography; patients without overt plaque were immediately discharged from the hospital, patients with nonobstructive plaque and mild-to-moderate stenoses were discharged after a negative 6-hour troponin level, and patients with severe stenoses were admitted to the hospital. Discharged patients were followed up for a mean of 14.2 months. Additionally, length of stay and safety outcomes among these patients were compared with those in 102 consecutive patients with low- to intermediate-risk chest pain who presented to the ED and underwent a standard of care (SOC) work-up without cardiac CT angiography. One-way analysis of variance with Bonferroni correction was used to compare length of stay between groups.

RESULTS

Cardiac CT angiography findings in the 203 patients who underwent cardiac CT angiography were as follows: Sixty-five (32%) patients had no plaque, 107 (53%) had nonobstructive plaque, and 31 (15%) had severe stenoses. At follow-up, there were no deaths or cases of acute coronary syndrome (cardiac CT angiography, 0%, 95% confidence interval [CI]: 0%, 1.85%; SOC, 0%, 95% CI: 0%, 3.63%), and the rate of readmission to the hospital because of chest pain was higher with the SOC approach (9% vs 1%, P = .01). Mean ED length of stay was lower with cardiac CT angiography (6.62 hours ± 0.38 after a single troponin level and 9.15 hours ± 0.30 after serial troponin levels) than with the SOC approach (11.62 hours ± 0.47, P < .001).

CONCLUSION

Tailoring troponin measurement to cardiac CT angiography findings is safe and allows early discharge of patients with low- to intermediate-risk chest pain, resulting in reduced length of stay.

Diagnostic accuracy of the TIMI risk score in patients with chest pain in the emergency department: a meta-analysis

BACKGROUND

The Thrombolysis in Myocardial Infarction (TIMI) risk score uses clinical data to predict the short-term risk of acute myocardial infarction, coronary revascularization or death from any cause. It was originally developed for use in patients with unstable angina or non-ST-elevation myocardial infarction. We sought to expand the clinical application of the TIMI risk score by assessing its prognostic accuracy in patients in the emergency department with potential acute coronary syndromes.

METHODS

We searched five electronic databases, hand-searched reference lists of included studies and contacted content experts to identify articles for review. We included prospective cohort studies that validated the TIMI risk score in emergency department patients. We performed a meta-regression to determine whether a linear relation exists between TIMI risk score and the cumulative incidence of cardiac events.

RESULTS

We included 10 prospective cohort studies (with a total of 17 265 patients) in our systematic review. Data were available for meta-analysis in 8 of the 10 studies. Of patients with a score of zero, 1.8% had a cardiac event within 30 days (sensitivity 97.2%, 95% CI 96.4-97.8; specificity 25.0%, 95% CI 24.3-25.7; positive likelihood ratio 1.30, 95% CI 1.28-1.31; negative likelihood ratio 0.11, 95% CI 0.09-0.15). Meta-regression analysis revealed a strong linear relation between TIMI risk score (p < 0.001) and the cumulative incidence of cardiac events.

INTERPRETATION

Although the TIMI risk score is an effective risk stratification tool for patients in the emergency department with potential acute coronary syndromes, it should not be used as the sole means of determining patient disposition.

Future developments in chest pain diagnosis and management

Much of the focus of research on patients with chest pain is directed at technological advances in the diagnosis and management of acute coronary syndrome (ACS), pulmonary embolism (PE), and acute aortic dissection (AAD), despite there being no significant difference at 4 years as regards mortality, ongoing chest pain, and quality of life between patients presenting to the emergency department with noncardiac chest pain and those with cardiac chest pain. This article examines future developments in the diagnosis and management of patients with suspected ACS, PE, AAD, gastrointestinal disease, and musculoskeletal chest pain.

Prospective validation of a modified thrombolysis in myocardial infarction risk score in emergency department patients with chest pain and possible acute coronary syndrome

OBJECTIVES

This study attempted to prospectively validate a modified Thrombolysis In Myocardial Infarction (TIMI) risk score that classifies patients with either ST-segment deviation or cardiac troponin elevation as high risk. The objectives were to determine the ability of the modified score to risk-stratify emergency department (ED) patients with chest pain and to identify patients safe for early discharge.

METHODS

This was a prospective cohort study in an urban academic ED over a 9-month period. Patients over 24 years of age with a primary complaint of chest pain were enrolled. On-duty physicians completed standardized data collection forms prior to diagnostic testing. Cardiac troponin T-values of >99th percentile (> or =0.01 ng/mL) were considered elevated. The primary outcome was acute myocardial infarction (AMI), revascularization, or death within 30 days. The overall diagnostic accuracy of the risk scores was compared by generating receiver operating characteristic (ROC) curves and comparing the area under the curve. The performance of the risk scores at potential decision thresholds was assessed by calculating the sensitivity and specificity at each potential cut-point.

RESULTS

The study enrolled 1,017 patients with the following characteristics: mean (+/-SD) age 59.3 (+/-13.8) years, 60.6% male, 17.9% with a history of diabetes, and 22.4% with a history of myocardial infarction. A total of 117 (11.5%) experienced a cardiac event within 30 days (6.6% AMI, 8.9% revascularization, 0.2% death of cardiac or unknown cause). The modified TIMI risk score outperformed the original with regard to overall diagnostic accuracy (area under the ROC curve = 0.83 vs. 0.79; p = 0.030; absolute difference 0.037; 95% confidence interval [CI] = 0.004 to 0.071). The specificity of the modified score was lower at all cut-points of >0. Sensitivity and specificity at potential decision thresholds were: >0 = sensitivity 96.6%, specificity 23.7%; >1 = sensitivity 91.5%, specificity 54.2%; and >2 = sensitivity 80.3%, specificity 73.4%. The lowest cut-point (TIMI/modified TIMI >0) was the only cut-point to predict cardiac events with sufficient sensitivity to consider early discharge. The sensitivity and specificity of the modified and original TIMI risk scores at this cut-point were identical.

CONCLUSIONS

The modified TIMI risk score outperformed the original with regard to overall diagnostic accuracy. However, it had lower specificity at all cut-points of >0, suggesting suboptimal risk stratification in high-risk patients. It also lacked sufficient sensitivity and specificity to safely guide patient disposition. Both scores are insufficiently sensitive and specific to recommend as the sole means of determining disposition in ED chest pain patients.

Discrimination of myocardial infarction stages by subjective feature extraction

Lots of studies on myocardial infarction (MI) computer assisted diagnosis are based on certain important ECG components which only account for local information. 12-Lead ECG signals which were regarded as hyper-dimensional time-series data were utilized to extract features from global information in this study. Existing feature extraction techniques for classification attempt to classify all the classes included. However sometimes it is more important to better recognize certain specific classes rather than to discriminate all the classes. A feature extraction method based on subjective-classification was proposed to discriminate the specific classes, which the classification priority was given subjectively, and each of the other classes was separated at the same time. The method includes data reduction by principal component analysis (PCA), data normalization by whitening transformation and derivation of projecting vectors for subjective-classification, etc. The data in the analysis were collected from PTB diagnostic ECG database. The results show that the proposed method can obtain a small number of effective features from 12-lead ECGs to better classify classes with priority, and the other classes can be classified at the same time.

One-year outcomes following coronary computerized tomographic angiography for evaluation of emergency department patients with potential acute coronary syndrome

OBJECTIVES

Coronary computerized tomographic angiography (CTA) has high correlation with cardiac catheterization and has been shown to be safe and cost-effective when used for rapid evaluation of low-risk chest pain patients from the emergency department (ED). The long-term outcome of patients discharged from the ED with negative coronary CTA has not been well studied.

METHODS

The authors prospectively evaluated consecutive low- to intermediate-risk patients who received coronary CTA in the ED for evaluation of a potential acute coronary syndrome (ACS). Patients with cocaine use, known cancer, and significant comorbidity reducing life expectancy and those found to have significant disease (stenosis > or = 50% or ejection fraction < 30%) were excluded. Demographics, medical and cardiac history, labs, and electrocardiogram (ECG) results were collected. Patients were followed by telephone contact and record review for 1 year. The main outcome was 1-year cardiovascular death or nonfatal acute myocardial infarction (AMI).

RESULTS

Of 588 patients who received coronary CTA in the ED, 481 met study criteria. They had a mean (+/-SD) age of 46.1 (+/-8.8) years, 63% were black or African American, and 60% were female. There were 53 patients (11%) rehospitalized and 51 patients (11%) who received further diagnostic testing (stress or catheterization) over the subsequent year. There was one death (0.2%; 95% confidence interval [CI] = 0.01% to 1.15%) with unclear etiology, no AMI (0%; 95% CI = 0 to 0.76%), and no revascularization procedures (0%; 95% CI = 0 to 0.76%) during this time period.

CONCLUSIONS

Low- to intermediate-risk patients with a Thrombosis In Myocardial Infarction (TIMI) score of 0 to 2 who present to the ED with potential ACS and have a negative coronary CTA have a very low likelihood of cardiovascular events over the ensuing year.

Emergency physician high pretest probability for acute coronary syndrome correlates with adverse cardiovascular outcomes

OBJECTIVES

The value of unstructured physician estimate of risk for disease processes, other than acute coronary syndrome (ACS), has been demonstrated. The authors sought to evaluate the predictive value of unstructured physician estimate of risk for ACS in emergency department (ED) patients without obvious initial evidence of a cardiac event.

METHODS

This was a post hoc secondary analysis of the Internet Tracking Registry for Acute Coronary Syndromes (i*trACS), a prospectively collected multicenter data registry of patients over the age of 18 years presenting to the ED with symptoms of ACS between 1999 and 2001. In this registry, following patient history, physical exam, and electrocardiogram (ECG), the unstructured treating physician estimate of risk was recorded. A 30-day follow-up and a medical record review were used to determine rates of adverse cardiac events, death, myocardial infarction (MI), or revascularization procedure. The analysis included all patients with nondiagnostic ECG changes, normal initial biomarkers, and a non-MI initial impression from the registry and excluded those without complete data or who were lost to follow-up. Data were stratified by unstructured physician risk estimate: noncardiac, low risk, high risk, or unstable angina.

RESULTS

Of 15,608 unique patients in the registry, 10,145 met inclusion/exclusion criteria. Patients were defined as having unstable angina in 6.0% of cases; high risk, 23.5% of cases; low risk, 44.2%; and noncardiac, 26.3% of cases. Adverse cardiac event rates had an inverse relationship, decreasing from 22.0% (95% confidence interval [CI] = 18.8% to 25.6%) for unstable angina, 10.2% (95% CI = 9.0% to 11.5%) for those stratified as high risk, 2.2% (95% CI = 1.8% to 2.6%) for low risk, and to 1.8% (95% CI = 1.4% to 2.4%) for noncardiac. The relative risk (RR) of an adverse cardiac event for those with an initial label of unstable angina compared to those with a low-risk designation was 10.2 (95% CI = 8.0 to 13.0). The RR of an event for those with a high-risk initial impression compared to those with a low-risk initial impression was 4.7 (95% CI = 3.8 to 5.9). The risk of an event among those with a low-risk initial impression was the same as for those with a noncardiac initial impression (RR = 0.83, 95% CI = 0.6 to 1.2).

CONCLUSIONS

In ED patients without obvious initial evidence of a cardiac event, unstructured emergency physician (EP) estimate of risk correlates with adverse cardiac outcomes.

Added value of a resting ECG neural network that predicts cardiovascular mortality

BACKGROUND

The resting 12-lead electrocardiogram (ECG) remains the most commonly used test in evaluating patients with suspected cardiovascular disease. Prognostic values of individual findings on the ECG have been reported but may be of limited use.

METHODS

The characteristics of 45,855 ECGs ordered by physician's discretion were first recorded and analyzed using a computerized system. Ninety percent of these ECGs were used to train an artifical neural network (ANN) to predict cardiovascular mortality (CVM) based on 132 ECG and four demographic characteristics. The ANN generated a Resting ECG Neural Network (RENN) score that was then tested in the remaining ECGs. The RENN score was finally assessed in a cohort of 2189 patients who underwent exercise treadmill testing and were followed for CVM.

RESULTS

The RENN score was able to better predict CVM compared to individual ECG markers or a traditional Cox regression model in the testing cohort. Over a mean of 8.6 years, there were 156 cardiovascular deaths in the treadmill cohort. Among the patients who were classified as intermediate risk by Duke Treadmill Scoring (DTS), the third tertile of the RENN score demonstrated an adjusted Cox hazard ratio of 5.4 (95% CI 2.0-15.2) compared to the first RENN tertile. The 10-year CVM was 2.8%, 8.6% and 22% in the first, second and third RENN tertiles, respectively.

CONCLUSIONS

An ANN that uses the resting ECG and demographic variables to predict CVM was created. The RENN score can further risk stratify patients deemed at moderate risk on exercise treadmill testing.

Application of artificial neural networks for decision support in medicine

The emergence of drug resistant pathogens can reduce the efficacy of drugs commonly used to treat infectious diseases. Human immunodeficiency virus (HIV) is particularly sensitive to drug selection pressure, rapidly evolving into drug resistant variants on exposure to anti-HIV drugs. Over 200 mutations within the genetic material of HIV have been shown to be associated with drug resistance to date, and complex mutational patterns have been found in HIV isolates from infected patients exposed to multiple antiretroviral drugs. Genotyping is commonly used in clinical practice as a tool to identify drug resistance mutations in HIV from individual patients. This information is then used to help guide the choice of future therapy for patients whose drug regimen is failing because of the development of drug resistant HIV. Many sets of rules and algorithms are available to predict loss of susceptibility to individual antiretroviral drugs from genotypic data. Although this approach has been helpful, the interpretation of genotypic data remains challenging. We describe here the development and application of ANN models as alternative tools for the interpretation of HIV genotypic drug resistance data. A large amount of clinical and virological data, from around 30,000 patients treated with antiretroviral drugs, has been collected by the HIV Resistance Response Database Initiative (RDI, www.hivrdi.org) in a centralized database. Treatment change episodes (TCEs) have been extracted from these data and used along with HIV drug resistance mutations as the basic input variables to train ANN models. We performed a series of analyses that have helped define the following: (1) the reliability of ANN predictions for HIV patients receiving routine clinical care; (2) the utility of ANN models to identify effective treatments for patients failing therapy; (3) strategies to increase the accuracy of ANN predictions; and (4) performance of ANN models in comparison to the rules-based methods currently in use.

Diagnostic accuracy of clinical prediction rules to exclude acute coronary syndrome in the emergency department setting: a systematic review

OBJECTIVE

We sought to determine the diagnostic accuracy of clinical prediction rules to exclude acute coronary syndrome (ACS) in the emergency department (ED) setting.

METHODS

We searched MEDLINE, EMBASE, Web of Science and the Cochrane Database of Systematic Reviews. We contacted content experts to identify additional articles for review. Reference lists of included studies were hand searched. We selected articles for review based on the following criteria: 1) enrolled consecutive ED patients; 2) incorporated variables from the history or physical examination, electrocardiogram and cardiac biomarkers; 3) did not incorporate cardiac stress testing or coronary angiography into prediction rule; 4) based on original research; 5) prospectively derived or validated; 6) did not require use of a computer; and 7) reported sufficient data to construct a 2 x 2 contingency table. We assessed study quality and extracted data independently and in duplicate using a standardized data extraction form.

RESULTS

Eight studies met inclusion criteria, encompassing 7937 patients. None of the studies verified the prediction rule with a reference standard on all or a random sample of patients. Six studies did not report blinding prediction rule assessors to reference standard results, and vice versa. Three prediction rules were prospectively validated. Sensitivities and specificities ranged from 94% to 100% and 13% to 57%, and positive and negative likelihood ratios from 1.1 to 2.2 and 0.01 to 0.17, respectively.

CONCLUSION

Current prediction rules for ACS have substantial methodological limitations and have not been successfully implemented in the clinical setting. Future methodologically sound studies are needed to guide clinical practice.

Analysis of meal patterns with the use of supervised data mining techniques--artificial neural networks and decision trees

BACKGROUND

At present, the analysis of dietary patterns is based on the intake of individual foods. This article demonstrates how a coding system at the meal level might be analyzed by using data mining techniques.

OBJECTIVE

The objective was to evaluate the usability of supervised data mining methods to predict an aspect of dietary quality based on dietary intake with a food-based coding system and a novel meal-based coding system.

DESIGN

Food consumption databases from the North-South Ireland Food Consumption Survey 1997-1999 were used. This was a randomized cross-sectional study of 7-d recorded food and nutrient intakes of a representative sample of 1379 Irish adults. Meal definitions were recorded by the respondent. A healthy eating index (HEI) score was developed. Artificial neural networks (ANNs) and decision trees were used to predict quintiles of the HEI based on combinations of foods consumed at breakfast and main meals.

RESULTS

This study applied both data mining techniques to the food and meal-based coding systems. The ANN had a slightly higher accuracy than did the decision tree in relation to its ability to predict HEI quintiles 1 and 5 based on the food coding system (78.7% compared with 76.9% and 71.9% compared with 70.1%, respectively). However, the decision tree had higher accuracies than did the ANN on the basis of the meal coding system (67.5% compared with 54.6% and 75.1% compared with 72.4%, respectively).

CONCLUSIONS

ANNs and decision trees were successfully used to predict an aspect of dietary quality. However, further exploration of the use of ANNs and decision trees in dietary pattern analysis is warranted.

Coronary computed tomographic angiography for rapid discharge of low-risk patients with potential acute coronary syndromes

STUDY OBJECTIVE

Coronary computed tomographic (CT) angiography has excellent performance characteristics relative to coronary angiography and exercise or pharmacologic stress testing. We hypothesize that coronary CT angiography can identify a cohort of emergency department (ED) patients with a potential acute coronary syndrome who can be safely discharged with a less than 1% risk of 30-day cardiovascular death or nonfatal myocardial infarction.

METHODS

We conducted a prospective cohort study at an urban university hospital ED that enrolled consecutive patients with potential acute coronary syndromes and a low TIMI risk score who presented to the ED with symptoms suggestive of a potential acute coronary syndrome and received a coronary CT angiography. Our intervention was either immediate coronary CT angiography in the ED or after a 9- to 12-hour observation period that included cardiac marker determinations, depending on time of day. The main clinical outcome was 30-day cardiovascular death or nonfatal myocardial infarction.

RESULTS

Five hundred sixty-eight patients with potential acute coronary syndrome were evaluated: 285 of these received coronary CT angiography immediately in the ED and 283 received coronary CT angiography after a brief observation period. Four hundred seventy-six (84%) were discharged home after coronary CT angiography. During the 30-day follow-up period, no patients died of a cardiovascular event (0%; 95% confidence interval [CI] 0% to 0.8%) or sustained a nonfatal myocardial infarction (0%; 95% CI 0 to 0.8%).

CONCLUSION

ED patients with symptoms concerning for a potential acute coronary syndrome with a low TIMI risk score and a nonischemic initial ECG result can be safely discharged home after a negative coronary CT angiography test result.

Signs and symptoms in diagnosing acute myocardial infarction and acute coronary syndrome: a diagnostic meta-analysis

BACKGROUND

Prompt diagnosis of acute myocardial infarction or acute coronary syndrome is very important.

AIM

A systematic review was conducted to determine the accuracy of 10 important signs and symptoms in selected and non-selected patients.

DESIGN OF STUDY

Diagnostic meta-analysis.

METHOD

Using MEDLINE, CINAHL, EMBASE, tracing references, and by contacting experts, studies were sought out that described one of the 10 signs and symptoms on one or both conditions. Studies were excluded if they were not based on original data. Validity was assessed using QUADAS and all data were pooled using a random effects model.

RESULTS

Sixteen of the 28 included studies were about patients who were non-selected. In this group, absence of chest-wall tenderness on palpation had a pooled sensitivity of 92% (95% confidence interval [CI] = 86 to 96) for acute myocardial infarction and 94% (95% CI = 91 to 96) for acute coronary syndrome. Oppressive pain followed with a pooled sensitivity of 60% (95% CI = 55 to 66) for acute myocardial infarction. Sweating had the highest pooled positive likelihood ratio (LR+), namely 2.92 (95% CI = 1.97 to 4.23) for acute myocardial infarction. The other pooled LR+ fluctuated between 1.05 and 1.49. Negative LRs (LR-) varied between 0.98 and 0.23. Absence of chest-wall tenderness on palpation had a LR- of 0.23 (95% CI = 0.18 to 0.29).

CONCLUSIONS

Based on this meta-analysis it was not possible to define an important role for signs and symptoms in the diagnosis of acute myocardial infarction or acute coronary syndrome. Only chest-wall tenderness on palpation largely ruled out acute myocardial infarction or acute coronary syndrome in low-prevalence settings.

Automatic covariate selection in logistic models for chest pain diagnosis: a new approach

A newly established method for optimizing logistic models via a minorization-majorization procedure is applied to the problem of diagnosing acute coronary syndromes (ACS). The method provides a principled approach to the selection of covariates which would otherwise require the use of a suboptimal method owing to the size of the covariate set. A strategy for building models is proposed and two models optimized for performance and for simplicity are derived via 10-fold cross-validation. These models confirm that a relatively small set of covariates including clinical and electrocardiographic features can be used successfully in this task. The performance of the models is comparable with previously published models using less principled selection methods. The models prove to be portable when tested on data gathered from three other sites. Whilst diagnostic accuracy and calibration diminishes slightly for these new settings, it remains satisfactory overall. The prospect of building predictive models that are as simple as possible for a required level of performance is valuable if data-driven decision aids are to gain wide acceptance in the clinical situation owing to the need to minimize the time taken to gather and enter data at the bedside.

A study to derive a clinical decision rule for triage of emergency department patients with chest pain: design and methodology

BACKGROUND

Chest pain is the second most common chief complaint in North American emergency departments. Data from the U.S. suggest that 2.1% of patients with acute myocardial infarction and 2.3% of patients with unstable angina are misdiagnosed, with slightly higher rates reported in a recent Canadian study (4.6% and 6.4%, respectively). Information obtained from the history, 12-lead ECG, and a single set of cardiac enzymes is unable to identify patients who are safe for early discharge with sufficient sensitivity. The 2007 ACC/AHA guidelines for UA/NSTEMI do not identify patients at low risk for adverse cardiac events who can be safely discharged without provocative testing. As a result large numbers of low risk patients are triaged to chest pain observation units and undergo provocative testing, at significant cost to the healthcare system. Clinical decision rules use clinical findings (history, physical exam, test results) to suggest a diagnostic or therapeutic course of action. Currently no methodologically robust clinical decision rule identifies patients safe for early discharge.

METHODS/DESIGN

The goal of this study is to derive a clinical decision rule which will allow emergency physicians to accurately identify patients with chest pain who are safe for early discharge. The study will utilize a prospective cohort design. Standardized clinical variables will be collected on all patients at least 25 years of age complaining of chest pain prior to provocative testing. Variables strongly associated with the composite outcome acute myocardial infarction, revascularization, or death will be further analyzed with multivariable analysis to derive the clinical rule. Specific aims are to: i) apply standardized clinical assessments to patients with chest pain, incorporating results of early cardiac testing; ii) determine the inter-observer reliability of the clinical information; iii) determine the statistical association between the clinical findings and the composite outcome; and iv) use multivariable analysis to derive a highly sensitive clinical decision rule to guide triage decisions.

DISCUSSION

The study will derive a highly sensitive clinical decision rule to identify low risk patients safe for early discharge. This will improve patient care, lower healthcare costs, and enhance flow in our busy and overcrowded emergency departments.

Comparison of the levels of accuracy of an artificial neural network model and a logistic regression model for the diagnosis of acute appendicitis

An accurate diagnosis of acute appendicitis in the early stage is often difficult, and decision support tools to improve such a diagnosis might be required. This study compared the levels of accuracy of artificial neural network models and logistic regression models for the diagnosis of acute appendicitis. Data from 169 patients presenting with acute abdomen were used for the analyses. Nine variables were used for the evaluation of the accuracy of the two models. The constructed models were validated by the ".632+ bootstrap method". The levels of accuracy of the two models for diagnosis were compared by error rate and areas under receiver operating characteristic curves. The artificial neural network models provided more accurate results than did the logistic regression models for both indices, especially when categorical variables or normalized variables were used. The most accurate diagnosis was obtained by the artificial neural network model using normalized variables.

Early detection of acute coronary syndromes and risk stratification by multimarker analysis

Cardiac troponin is the standard biomarker for the diagnosis of acute myocardial infarction (AMI) and risk stratification for short-term adverse cardiac events (death, AMI or need for revascularization). Unfortunately, the concentration of troponin in blood is normal in AMI patients who present early after the onset of symptoms. As such, there is active research being conducted in finding early markers of AMI and risk stratification. Despite years of testing dozens of candidates, no single test has had the necessary clinical sensitivity and specificity for this indication. Therefore, many researchers have advocated multimarker testing. There are two approaches that have been taken for discovering new markers. The proteomic approach involves focusing on the differences in the biochemical signatures between the tissues or biological fluids of normal compared with diseased individuals. Specific biochemical targets are not preselected. The pathophysiologic approach involves combining biomarkers that indicate a particular pathway or event known to be involved in the disease process. In both approaches, some bioinformatic algorithm will be necessary in order to combine the information provided by the individual tests. Representative approaches include the Multimarker Index™, classification and regression tree analysis and neural networks.