scholarly journals Machine Learning prediction of cardiac resynchronisation therapy response from combination of clinical and model-driven data

2021 ◽  
Author(s):  
Svyatoslav Khamzin ◽  
Arsenii Dokuchaev ◽  
Anastasia Bazhutina ◽  
Tatiana Chumarnaya ◽  
Stepan Zubarev ◽  
...  
Keyword(s):  
Machine Learning ◽  
Predictive Model ◽  
Clinical Data ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Left Ventricular Ejection ◽  
Model Driven ◽  
Device Implantation ◽  
Ventricular Activation ◽  
Crt Response

AbstractBackgroundUp to 30%-50% of chronic heart failure patients who underwent cardiac resynchronization therapy (CRT) do not respond to the treatment. Therefore, patient stratification for CRT and optimization of CRT device settings remain a challenge.ObjectiveThe main goal of our study is to develop a predictive model of CRT outcome using a combination of clinical data recorded in patients before CRT and simulations of the response to biventricular (BiV) pacing in personalized computational models of the cardiac electrophysiology.Materials and MethodsRetrospective data from 57 patients who underwent CRT device implantation was utilized. Positive response to CRT was defined by a 10% increase in the left ventricular ejection fraction in a year after implantation. For each patient, an anatomical model of the heart and torso was reconstructed from MRI and CT images and tailored to ECG recorded in the participant. The models were used to compute ventricular activation time, ECG duration and electrical dyssynchrony indices during intrinsic rhythm and BiV pacing from active poles of leads. For building a predictive model of CRT response, we used clinical data recorded before CRT device implantation together with model-derived biomarkers of ventricular excitation in the left bundle branch block mode of activation and under BiV stimulation. Several Machine Learning (ML) classifiers and feature selection algorithms were tested on the hybrid dataset, and the quality of predictors was assessed using the area under receiver operating curve (ROC AUC). The classifiers on the hybrid data were compared with ML models built on clinical data only.ResultsThe best ML classifier utilizing a hybrid set of clinical and model-driven data demonstrated ROC AUC of 0.82, an accuracy of 0.82, sensitivity of 0.85, and specificity of 0.78, improving quality over that of ML predictors built on clinical data from much larger datasets. Distance from the LV pacing site to the post-infarction zone and ventricular activation characteristics under BiV pacing were shown as the most relevant model-driven features for CRT response classification.ConclusionOur results suggest that combination of clinical and model-driven data increases the accuracy of classification models for CRT outcomes.

scholarly journals Machine Learning Prediction of Cardiac Resynchronisation Therapy Response From Combination of Clinical and Model-Driven Data

2021 ◽  
Vol 12 ◽  
Author(s):  
Svyatoslav Khamzin ◽  
Arsenii Dokuchaev ◽  
Anastasia Bazhutina ◽  
Tatiana Chumarnaya ◽  
Stepan Zubarev ◽  
...  
Keyword(s):  
Machine Learning ◽  
Predictive Model ◽  
Clinical Data ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Left Ventricular Ejection ◽  
Model Driven ◽  
Device Implantation ◽  
Ventricular Activation ◽  
Crt Response

Background: Up to 30–50% of chronic heart failure patients who underwent cardiac resynchronization therapy (CRT) do not respond to the treatment. Therefore, patient stratification for CRT and optimization of CRT device settings remain a challenge.Objective: The main goal of our study is to develop a predictive model of CRT outcome using a combination of clinical data recorded in patients before CRT and simulations of the response to biventricular (BiV) pacing in personalized computational models of the cardiac electrophysiology.Materials and Methods: Retrospective data from 57 patients who underwent CRT device implantation was utilized. Positive response to CRT was defined by a 10% increase in the left ventricular ejection fraction in a year after implantation. For each patient, an anatomical model of the heart and torso was reconstructed from MRI and CT images and tailored to ECG recorded in the participant. The models were used to compute ventricular activation time, ECG duration and electrical dyssynchrony indices during intrinsic rhythm and BiV pacing from the sites of implanted leads. For building a predictive model of CRT response, we used clinical data recorded before CRT device implantation together with model-derived biomarkers of ventricular excitation in the left bundle branch block mode of activation and under BiV stimulation. Several Machine Learning (ML) classifiers and feature selection algorithms were tested on the hybrid dataset, and the quality of predictors was assessed using the area under receiver operating curve (ROC AUC). The classifiers on the hybrid data were compared with ML models built on clinical data only.Results: The best ML classifier utilizing a hybrid set of clinical and model-driven data demonstrated ROC AUC of 0.82, an accuracy of 0.82, sensitivity of 0.85, and specificity of 0.78, improving quality over that of ML predictors built on clinical data from much larger datasets by more than 0.1. Distance from the LV pacing site to the post-infarction zone and ventricular activation characteristics under BiV pacing were shown as the most relevant model-driven features for CRT response classification.Conclusion: Our results suggest that combination of clinical and model-driven data increases the accuracy of classification models for CRT outcomes.

scholarly journals Importance of systematic right ventricular assessment in cardiac resynchronization therapy candidates: a machine-learning approach

2021 ◽  
Vol 22 (Supplement_1) ◽  
Author(s):  
E Galli ◽  
V Le Rolle ◽  
OA Smiseth ◽  
J Duchenne ◽  
JM Aalen ◽  
...  
Keyword(s):  
Machine Learning ◽  
Heart Failure ◽  
Cardiac Resynchronization Therapy ◽  
Right Ventricular ◽  
Systolic Heart Failure ◽  
Left Ventricular ◽  
Supervised Machine Learning ◽  
Cardiac Resynchronization ◽  
Resynchronization Therapy ◽  
Crt Response

Abstract Funding Acknowledgements Type of funding sources: None. Background Despite having all a systolic heart failure and broad QRS, patients proposed for cardiac resynchronization therapy (CRT) are highly heterogeneous and it remains extremely complicated to predict the impact of the device on left ventricular (LV) function and outcomes. Objectives We sought to evaluate the relative impact of clinical, electrocardiographic, and echocardiographic data on the left ventricular (LV) remodeling and prognosis of CRT-candidates by the application of machine learning (ML) approaches. Methods 193 patients with systolic heart failure undergoing CRT according to current recommendations were prospectively included in this multicentre study. We used a combination of the Boruta algorithm and random forest methods to identify features predicting both CRT volumetric response and prognosis (Figure 1). The model performance was tested by the area under the receiver operating curve (AUC). We also applied the K-medoid method to identify clusters of phenotypically-similar patients. Results From 28 clinical, electrocardiographic, and echocardiographic-derived variables, 16 features were predictive of CRT-response; 11 features were predictive of prognosis. Among the predictors of CRT-response, 7 variables (44%) pertained to right ventricular (RV) size or function. Tricuspid annular plane systolic excursion was the main feature associated with prognosis. The selected features were associated with a very good prediction of both CRT response (AUC 0.81, 95% CI: 0.74-0.87) and outcomes (AUC 0.84, 95% CI: 0.75-0.93) (Figure 1, Supervised Machine Learning Panel). An unsupervised ML approach allowed the identifications of two phenogroups of patients who differed significantly in clinical and parameters, biventricular size and RV function. The two phenogroups had significant different prognosis (HR 4.70, 95% CI: 2.1-10.0, p < 0.0001; log –rank p < 0.0001; Figure 1, Unsupervised Machine Learning Panel). Conclusions Machine learning can reliably identify clinical and echocardiographic features associated with CRT-response and prognosis. The evaluation of both RV-size and function parameters has pivotal importance for the risk stratification of CRT-candidates and should be systematically assessed in patients undergoing CRT. Abstract Figure 1

Left bundle branch pacing improved heart function in a 10-year-old child after a 3-month follow-up

EP Europace ◽  
2020 ◽  
Vol 22 (8) ◽  
pp. 1234-1239
Author(s):  
Wei Ji ◽  
Xueying Chen ◽  
Jie Shen ◽  
Diqi Zhu ◽  
Yiwei Chen ◽  
...  
Keyword(s):  
Heart Function ◽  
Interventricular Septum ◽  
Left Ventricular ◽  
Left Ventricular Ejection ◽  
Bundle Branch Block ◽  
Ventricular Ejection Fraction ◽  
Intracardiac Electrogram ◽  
Ventricular Activation ◽  
Pacing Lead

Abstract Aims As a physiological pacing strategy, left bundle branch pacing (LBBP) were used to correct left bundle branch block (LBBB), however, there is no relevant report in children. We aimed to evaluate the feasibility of LBBP in children. Methods and results Left bundle branch pacing was performed in a 10-year-old girl with a second-degree atrioventricular and LBBB. Under the guide of fluoroscopy, the pacing lead was deeply screwed into the interventricular septum to gain right bundle branch block (RBBB) pattern of paced QRS. Selective LBBP was achieved with a typical RBBB pattern of paced morphology and a discrete component between stimulus and ventricular activation in intracardiac electrogram and reached the standard of the stimulus to left ventricular activation time of 56 ms. At a 3-month follow-up, the LBBP acquired the reduction of left ventricular size and enhancement of left ventricular ejection fraction. Conclusion The application of LBBP in a child was first achieved with inspiring preliminary results. The LBBP can be carried out in children by cautiousness under the premise of strict grasp of indications.

scholarly journals 1185 Three-dimensional speckle tracking echocardiography for the global and regional assessment of myocardial deformation in breast cancer patients submitted to anthracyclines

2020 ◽  
Vol 21 (Supplement_1) ◽  
Author(s):  
M Coutinho Cruz ◽  
L Moura-Branco ◽  
G Portugal ◽  
A Galrinho ◽  
M Mota-Carmo ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Cancer Patients ◽  
Speckle Tracking ◽  
Speckle Tracking Echocardiography ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Left Ventricular Ejection ◽  
Breast Cancer Patients ◽  
Timely Initiation ◽  
Anthracycline Therapy

Abstract Introduction Serial echocardiographic assessment of 2D/3D left ventricular ejection fraction (LVEF) and 2D global longitudinal strain (GLS) is the gold standard for screening for cancer therapeutics-related cardiac dysfunction (CTRCD). Although 3D speckle tracking echocardiography (STE) has several technical advantages, is more reproducible, and has a better correlation to magnetic resonance than 2D STE, it is still not currently used in this setting. We aimed to investigate the usefulness of 3D STE in evaluating left ventricle mechanics and its relation to CTRCD. Methods Prospective study of female breast cancer patients submitted to anthracycline chemotherapy who underwent one transthoracic echocardiography (ETT) before and at least one ETT during/after chemotherapy. Standard ETT parameters and 3D volumetric measurements were assessed. STE was used to estimate 2D GLS – average and 18 segments – and 3D GLS, global circumferential strain (GCS), global radial strain (GRS) and global area strain (GAS) – average and 17 segments. CTRCD was defined as an absolute decrease in 2D or 3D LVEF >10% to a value <54% or a relative decrease in 2D GLS >15%. Results 105 patients (mean age 53.8 ± 12.5 years, 52.4% immunotherapy, 77.2% radiotherapy, 2.8 echocardiograms/patient) were included. During a mean follow-up of 12.1 months, 24 patients (22.9%) developed CTRCD. During anthracycline therapy, there was a significant worsening of 2D LVEF (65.6 vs. 57.8), 3D LVEF (61.5 vs. 54.4), 2D GLS (-21.1 vs. -18.0), 3D GLS (-15.6 vs. -10.9), 3D GCS (-14.0 vs. -11.0), 3D GRS (42.0 vs. 28.5) and 3D GAS (-27.0 vs. -20.0) [all p <0.001]. More than 73% of patients presented 3D global strain values below the limits of normal during chemotherapy. On 3D strain regional analysis, impaired contractility was observed in the anterior, inferior and septal walls. Logistic regression analysis showed that 3D GRS and 3D GCS were associated with a higher incidence of CTRCD. In the multivariate model, 3D GRS remained the only independent predictor of CTRCD. The receiver operating curve analysis showed a good calibration and discrimination of 3D GCS and 3D GRS in predicting CTRCD with areas under de curve of 0.748 and 0.719, with the optimal cut-off values being 0.342 for GCS and 0.344 for GRS. These variations were observed a median of 45 days and 22.5 days before the diagnosis of CTRCD, respectively. Conclusion 3D strain parameters worsened during anthracycline therapy, with predominant involvement of septal, anterior and inferior walls. Variations of 3D GCS and GRS were predictive of subsequent CTRCD, and thus can be considered an earlier sign of CTRCD, with added value over the currently recommended 2D/3D LVEF and 2D GLS. Routine application of this technique should be considered in order to offer targeted monitoring and timely initiation of cardioprotective treatment.

scholarly journals Machine Learning–Based Risk Assessment for Cancer Therapy–Related Cardiac Dysfunction in 4300 Longitudinal Oncology Patients

2020 ◽  
Vol 9 (23) ◽  
Author(s):  
Yadi Zhou ◽  
Yuan Hou ◽  
Muzna Hussain ◽  
Sherry‐Ann Brown ◽  
Thomas Budd ◽  
...  
Keyword(s):  
Machine Learning ◽  
Cancer Therapy ◽  
Cancer Patients ◽  
Cardiac Dysfunction ◽  
De Novo ◽  
Left Ventricular ◽  
Left Ventricular Ejection ◽  
Oncology Patients ◽  
Relevant Variables ◽  
After Cancer

Background The growing awareness of cardiovascular toxicity from cancer therapies has led to the emerging field of cardio‐oncology, which centers on preventing, detecting, and treating patients with cardiac dysfunction before, during, or after cancer treatment. Early detection and prevention of cancer therapy–related cardiac dysfunction (CTRCD) play important roles in precision cardio‐oncology. Methods and Results This retrospective study included 4309 cancer patients between 1997 and 2018 whose laboratory tests and cardiovascular echocardiographic variables were collected from the Cleveland Clinic institutional electronic medical record database (Epic Systems). Among these patients, 1560 (36%) were diagnosed with at least 1 type of CTRCD, and 838 (19%) developed CTRCD after cancer therapy (de novo). We posited that machine learning algorithms can be implemented to predict CTRCDs in cancer patients according to clinically relevant variables. Classification models were trained and evaluated for 6 types of cardiovascular outcomes, including coronary artery disease (area under the receiver operating characteristic curve [AUROC], 0.821; 95% CI, 0.815–0.826), atrial fibrillation (AUROC, 0.787; 95% CI, 0.782–0.792), heart failure (AUROC, 0.882; 95% CI, 0.878–0.887), stroke (AUROC, 0.660; 95% CI, 0.650–0.670), myocardial infarction (AUROC, 0.807; 95% CI, 0.799–0.816), and de novo CTRCD (AUROC, 0.802; 95% CI, 0.797–0.807). Model generalizability was further confirmed using time‐split data. Model inspection revealed several clinically relevant variables significantly associated with CTRCDs, including age, hypertension, glucose levels, left ventricular ejection fraction, creatinine, and aspartate aminotransferase levels. Conclusions This study suggests that machine learning approaches offer powerful tools for cardiac risk stratification in oncology patients by utilizing large‐scale, longitudinal patient data from healthcare systems.

Ventricular arrhythmia risk is associated with myocardial scar but not with response to cardiac resynchronization therapy

EP Europace ◽  
2020 ◽  
Vol 22 (9) ◽  
pp. 1391-1400
Author(s):  
Markus Linhart ◽  
Adelina Doltra ◽  
Juan Acosta ◽  
Roger Borràs ◽  
Beatriz Jáuregui ◽  
...  
Keyword(s):  
Ventricular Arrhythmia ◽  
Cardiac Resynchronization Therapy ◽  
Primary Outcome ◽  
Left Ventricular ◽  
Myocardial Scar ◽  
Cardiac Resynchronization ◽  
Left Ventricular Ejection ◽  
Resynchronization Therapy ◽  
Ischaemic Cardiomyopathy ◽  
Crt Response

Abstract Aims Sudden cardiac death (SCD) risk estimation in patients referred for cardiac resynchronization therapy (CRT) remains a challenge. By CRT-mediated improvement of left ventricular ejection fraction (LVEF), many patients loose indication for primary prevention implantable cardioverter-defibrillator (ICD). Increasing evidence shows the importance of myocardial scar for risk prediction. The aim of this study was to investigate the prognostic impact of myocardial scar depending on the echocardiographic response in patients undergoing CRT. Methods and results Patients with indication for CRT were prospectively enrolled. Decision about ICD or pacemaker implantation was based on clinical criteria. All patients underwent delayed-enhancement cardiac magnetic resonance imaging. Median follow-up duration was 45 (24–75) months. Primary outcome was a composite of sustained ventricular arrhythmia, appropriate ICD therapy, or SCD. A total of 218 patients with LVEF 25.5 ± 6.6% were analysed [158 (73%) male, 64.9 ± 10.7 years]. Myocardial scar was observed in 73 patients with ischaemic cardiomyopathy (ICM) (95% of ICM patients); in 62 with non-ischaemic cardiomyopathy (45% of these patients); and in all but 1 of 36 (17%) patients who reached the primary outcome. Myocardial scar was the only significant predictor of primary outcome [odds ratio 27.7 (3.8–202.7)], independent of echocardiographic CRT response. A total of 55 (25%) patients died from any cause or received heart transplant. For overall survival, only a combination of the absence of myocardial scar with CRT response was associated with favourable outcome. Conclusion Malignant arrhythmic events and SCD depend on the presence of myocardial scar but not on CRT response. All-cause mortality improved only with the combined absence of myocardial scar and CRT response.

scholarly journals Septal contraction predicts acute haemodynamic improvement and paced QRS width reduction in cardiac resynchronization therapy

2020 ◽  
Vol 21 (8) ◽  
pp. 845-852 ◽  
Author(s):  
Stian Ross ◽  
Eirik Nestaas ◽  
Erik Kongsgaard ◽  
Hans H Odland ◽  
Trine F Haland ◽  
...  
Keyword(s):  
Cardiac Resynchronization Therapy ◽  
Diastolic Pressure ◽  
Electrical Response ◽  
Electrical Parameter ◽  
Biventricular Pacing ◽  
Left Ventricular ◽  
Cardiac Resynchronization ◽  
Resynchronization Therapy ◽  
Device Implantation ◽  
Crt Response

Abstract Aims  Three distinct septal contraction patterns typical for left bundle branch block may be assessed using echocardiography in heart failure patients scheduled for cardiac resynchronization therapy (CRT). The aim of this study was to explore the association between these septal contraction patterns and the acute haemodynamic and electrical response to biventricular pacing (BIVP) in patients undergoing CRT implantation. Methods and results  Thirty-eight CRT candidates underwent speckle tracking echocardiography prior to device implantation. The patients were divided into two groups based on whether their septal contraction pattern was indicative of dyssynchrony (premature septal contraction followed by various amount of stretch) or not (normally timed septal contraction with minimal stretch). CRT implantation was performed under invasive left ventricular (LV) pressure monitoring and we defined acute CRT response as ≥10% increase in LV dP/dtmax. End-diastolic pressure (EDP) and QRS width served as a diastolic and electrical parameter, respectively. LV dP/dtmax improved under BIVP (737 ± 177 mmHg/s vs. 838 ± 199 mmHg/s, P < 0.001) and 26 patients (68%) were defined as acute CRT responders. Patients with premature septal contraction (n = 27) experienced acute improvement in systolic (ΔdP/dtmax: 18.3 ± 8.9%, P < 0.001), diastolic (ΔEDP: −30.6 ± 29.9%, P < 0.001) and electrical (ΔQRS width: −23.3 ± 13.2%, P < 0.001) parameters. No improvement under BIVP was observed in patients (n = 11) with normally timed septal contraction (ΔdP/dtmax: 4.0 ± 7.8%, P = 0.12; ΔEDP: −8.8 ± 38.4%, P = 0.47 and ΔQRS width: −0.9 ± 11.4%, P = 0.79). Conclusion  Septal contraction patterns are an excellent predictor of acute CRT response. Only patients with premature septal contraction experienced acute systolic, diastolic, and electrical improvement under BIVP.

scholarly journals Padang Skoring Elektrokardiografi untuk Memprediksi Fraksi Ejeksi Ventrikel Kiri pada Gagal Jantung Kronik

2019 ◽  
Vol 39 (4) ◽  
Author(s):  
Wiza Erlanda ◽  
Hauda El Rasyid ◽  
Masrul Syafri ◽  
Ricvan Dana Nindrea
Keyword(s):  
Heart Failure ◽  
Chronic Heart Failure ◽  
Ejection Fraction ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Left Ventricular Ejection ◽  
Qtc Interval ◽  
Chi Square ◽  
Ventricular Ejection Fraction ◽  
Receiver Operating

ABSTRAK Latar Belakang:  Gagal Jantung dibagi menjadi 3 kelompok yaitu gagal jantung fraksi ejeksi menurun (HfrEF <40%), rentang tengah (HFmrEF 40-49%), dan terpelihara (HfpEF ≥50%). Ekokardiografi masih menjadi pemeriksaan standar saat ini, sayangnya pemeriksaan tersebut masih terbatas dibeberapa pusat kesehatan. Perlu pemeriksaan awal yang lebih sederhana salah satunya elektrokardiografi (EKG) yang mudah digunakan. Penentuan skoring dari EKG diharapkan dapat memudahkan memprediksi fraksi ejeksi dan memberikan terapi yang tepat.  Metode Penelitian: Pendekatan observasional dengan desain potong lintang. Diambil data rekam medis pasien gagal jantung kronik (GJK) di poliklinik Jantung RSUP Dr. M.Djamil Padang bulan Januari-Agustus 2017. Dilakukan analisis bivariat pada varibel EKG terhadap fraksi ejeksi dengan metode chi-square. Analisis multivariate dengan uji regresi binari logistik untuk mendapatkan variabel pada kalkulasi skor dengan uji Hosmer-Lameshow (p<0,25). Skoring dilakukan uji sensitivitas, spesifisitas dan analisis receiver operating curve (ROC).    Hasil Penelitian : 283 subjek GJK dibagi menjadi tiga kelompok. Variabel yang memenuhi persyaratan untuk dilakukan kalkulasi skor adalah pembesaran atrium kiri (LAE) (OR=6,36; p= 0.000) dengan skor 2, QRS lebar (OR=13,06; p= 0.000) dengan skor 3, interval QTc memanjang (OR=2,18; p= 0.065) dengan skor 1 dan perubahan gelombang ST-T (OR=5,05; p= 0.000) dengan skor 2. Subjek dengan HFpEF mempunyai skor <3, HFmrEF mempunyai skor 3-4, dan HFrEF mempunyai skor >4. Sistem skoring EKG memiliki sensitivitas 71,4% dan spesifisitas 88,6% dengan AUC 87,9%  Kesimpulan : Sistem skoring EKG pada penelitian ini dapat digunakan sebagai pedoman awal dalam menentukan fraksi ejeksi ventrikel kiri pada pasien GJK  Kata kunci : elektrokardiografi, gagal jantung kronik, fraksi ejeksi ventrikel kiri Background: Heart failure (HF) are divided into HF reduced ejection fraction (HFrEF<40%), mid range (HFmrEF 40-49%), and preserved (HFpEF ≥50%). Nowadays echocardiography is used as gold standard examination, but it is limited only in several health centers. For this reason, a preliminary examination tools is needed. Electrocardiographic (ECG) examinations tool that available almost at every health center and easy to be used. Calculating the scores from ECG variables to determine the EF will make clinician’s earlier to give initial terapi Method: An observational approach with cross sectional study design. The data was taken from patient’s medical record with chronic heart failure (CHF) who went to the Tepartement of Cardiology at Dr. M. Djamil Padang Hospital in January-August 2017. Bivariate analysis was performed on each ECG variable then correlated with LVEF by chi-square method. Multivariate analysis with logistic binary regression test was conducted to obtain variables that would go into the score calculation stage with the Hosmer-Lameshow test (p <0.25). The sensitivity, specificity test and receiver operating curve (ROC) analysis were performed. Result: 283 subjects of CHF who had been divided into three groups. Obtained variables that met the requirements for calculating scores were left atrial enlargement (LAE) (OR = 6.36; p = 0.000) score was 2, wide QRS (OR = 13.06; p = 0,000) score was 3 , prolonged QTc interval (OR = 2,18; p = 0,065) score was 1 and ST-T change (OR = 5.05; p = 0.000) score was 2. Subjects with HFpEF if the scored were <3, HFmrEF if the scored were 3-4, and HFrEF if the scored were >4. It has sensitivity 71,4%, specificity 88,6% with AUC 87,9%. Conclusion: Electrocardiography scoring system in this study can be used as an initial tools to determining LVEF in patients with CHF. Keywords: electrocardiography, chronic heart failure, left ventricular ejection fraction

scholarly journals A machine learning predictive model of in-hospital mortality in patients with sepsis complicated by anemia: a retrospective study based on the MIMIC-III database

2021 ◽  
Author(s):  
Xiaobin Liu ◽  
Yu Zhao ◽  
Yingyi Qin ◽  
Dan Wang ◽  
Xi Yin ◽  
...  
Keyword(s):  
Machine Learning ◽  
Hospital Mortality ◽  
Predictive Model ◽  
Clinical Data ◽  
Predictive Models ◽  
Goodness Of Fit ◽  
Advantages And Disadvantages ◽  
Extreme Gradient Boosting ◽  
Comorbidity Index ◽  
Mimic Iii

Abstract BackgroudPatients with sepsis complicated by anemia have a higher risk of mortality. It is clinically important to study the risk factors associated with the prognosis of this disease. The aim of this study was to establish a predictive model of mortality during hospitalization by extracting clinical data from the Medical Information Mart for Intensive Care III (MIMIC-III) database. MethodsThe clinical data of patients with sepsis complicated by anemia in the MIMIC-III database were retrospectively analyzed. Indexes were screened by stepwise logistic regression (LR), and machine learning predictive models such as Decision Tree (DT), Random Forests (RF), and eXtreme Gradient Boosting (XGBoost) were developed and compared, identifying advantages and disadvantages of each model. ResultsA total of 13,547 patients with sepsis complicated by anemia were included in the study, among which 1,827 died during hospitalization and 11,720 were still alive at discharge. The preliminary stepwise regression model selected 20 clinical indexes, including Elixhauser comorbidity index, maximum blood urea nitrogen (BUN), and maximum hemoglobin reduction. The predictive models showed good discriminative ability (area under the receiver operating characteristic curve [AUROC]:LR, 0.777; DT, 0.726; RF, 0.788; XGBoost, 0.815) and goodness of fit (area under the precision-recall curve [AUPRC]: LR, 0.350; DT, 0.290; RF, 0.400; XGBoost, 0.428). The Shapley Additive exPlanation (SHAP) values in the XGBoost model showed that Elixhauser comorbidity index, maximum BUN, maximum hemoglobin reduction, ventilator use within 24 hours of admission, and age were significant features for predicting in-hospital mortality in patients with sepsis complicated by anemia. ConclusionsThe XGBoost model had better discrimination ability and goodness of fit when compared with other models. Machine learning algorithms have significant practical value in the development of an early warning system for patients with sepsis complicated by anemia.

External validation of an electrocardiography artificial intelligence-generated algorithm to detect left ventricular systolic function in a general cardiac clinic in Uganda

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
C.K Mondo ◽  
Z.I Attia ◽  
E.D Benavente ◽  
P Friedman ◽  
P Noseworthy ◽  
...  
Keyword(s):  
Artificial Intelligence ◽  
Heart Failure ◽  
Negative Predictive Value ◽  
Predictive Value ◽  
Left Ventricular Systolic Dysfunction ◽  
External Validation ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Left Ventricular Ejection ◽  
Ventricular Ejection Fraction

Abstract Background Left ventricular systolic dysfunction (LVSD) is associated with increased morbidity and mortality. Although there are effective treatments for patients with LVSD to prevent mortality, heart failure and to improve symptoms, many patients remain undetected and untreated. We have recently derived a deep learning algorithm to detect LVSD using the electrocardiogram (ECG) which could have an important screening role, particularly in limited resources settings. We evaluated the accuracy of this algorithm for the first time in Africa in a sample of subjects attending a cardiology clinic. Methods We conducted a retrospective study in a general cardiac clinic in Uganda. Consecutive patients ≥18 years who had a digital ECG and echocardiogram done within two days of each other were included. We excluded patients with pacemakers or missing information regarding left ventricular ejection fraction (LVEF). Routine 10-second, twelve-lead surface rest ECG were performed using an Edan PC ECG Model SE-1515, Hamburg, Germany. The probability of LVSD was estimated with the Mayo Clinic artificial intelligence (AI) ECG algorithm. LVEF was calculated by the MMode (Teichholz method) using a Philips Ultrasound system, HD7XE, Bothel, Washington, USA. LVSD was defined as a LVEF≤35%. We assessed the overall diagnostic performance of the algorithm to identify LVSD in this population with the area under the receiver operating curve (AUC), and estimated sensitivity, specificity and accuracy using a pre-specified cut-off based on the probability for LVSD generated by the algorithm. We conducted secondary analyses using different LVEF cutoff values. Results We included 634 subjects, 32% (200) of whom had hypertension and 12% (77) clinical heart failure. Mean age was 57±18.8 years, 58% were women and the overall prevalence of LVSD was 4%. The AI-ECG had an AUC of 0.866 (see figure below), sensitivity 73.08%, specificity 91.10%, negative predictive value 98.75%, positive predictive value 26.03% and an accuracy of 90.96% using the original threshold. Using the optimal cutoff based on the AUCs, the sensitivity was 80.77% and specificity was 81.05% with a negative predictive value of 98.99%. The ROC for the detection of LVEF of 40% or below was 0.821. Conclusion The Mayo AI-ECG algorithm demonstrated good accuracy, sensitivity and specificity to detect LVSD in patients seen in a clinical setting in Uganda. This tool may facilitate the identification of people at a high risk for LVSD in settings with low resources. ROC Funding Acknowledgement Type of funding source: None

Sign in / Sign up

Export Citation Format

Share Document