Classification of chest X-ray images by incorporation of medical domain knowledge into operation branch networks.

BACKGROUND: This study was conducted to alleviate a common difficulty in chest X-ray image diagnosis: The attention region in a convolutional neural network (CNN) does not often match the doctor's point of focus. The method presented herein, which guides the area of attention in CNN to a medically plausible region, can thereby improve diagnostic capabilities. METHODS: The model is based on an attention branch network, which has excellent interpretability of the classification model. This model has an additional new operation branch that guides the attention region to the lung field and heart in chest X-ray images. We also used three chest X-ray image datasets (Teikyo, Tokushima, and ChestX-ray14) to evaluate the CNN attention area of interest in these fields. Additionally, after devising a quantitative method of evaluating improvement of a CNN's region of interest, we applied it to evaluation of the proposed model. RESULTS: Operation branch networks maintain or improve the area under the curve to a greater degree than conventional CNNs do. Furthermore, the network better emphasizes reasonable anatomical parts in chest X-ray images. CONCLUSIONS: The proposed network better emphasizes the reasonable anatomical parts in chest X-ray images. This method can enhance capabilities for image interpretation based on judgment.

Clinical Utility of Overlap Time for Incomplete Relaxation to Predict Cardiac Events in Heart Failure.

BACKGROUND: The overlap time of transmitral flow can be a novel marker of subclinical left ventricular dysfunction for predicting adverse events in heart failure (HF). We aimed to (1) investigate the role of overlap time of the E-A wave in association with clinical parameters and (2) evaluate whether the overlap time could add prognostic information with respect to other conventional clinical prognosticators in HF. METHODS: We prospectively evaluated 153 patients hospitalized with HF (mean age 68 ± 15 years; 63% male). The primary endpoint was readmission following HF or cardiac death. RESULTS: During a median period of 25 months, 43 patients were readmitted or died. Overlap time appeared to be associated with worse outcomes. After adjustment for readmission scores and ratios of diastolic filling period and cardiac cycle length in a Cox proportional-hazards model, overlap time was associated with event-free survival, independent of elevated left atrial pressure based on guidelines. When overlap time was added to the model based on clinical variables and elevated left atrial pressure, the C-statistic significantly improved from 0.70 (95% CI: 0.63-0.77) to 0.77 (95% CI: 0.69-0.83, compared) (P = 0.035). CONCLUSION: This preliminary study suggested that prolonged overlap time may have potential for predicting readmission and cardiac mortality risk assessment in patients with HF.

Association Between Right Ventricular Contractile Function and Cardiac Events in Isolated Postcapillary and Combined Pre- and Postcapillary Pulmonary Hypertension.

BACKGROUND: Recent studies have shown that patients with combined pre- and postcapillary pulmonary hypertension (CpcPH) had worse outcomes than those with isolated postcapillary pulmonary hypertension (IpcPH). However, the prognostic factors including right ventricular (RV) function have not been well documented. The aim of this study was to assess the differentiation of PH phenotypes, using echocardiography, and the association between RV longitudinal strain and cardiac events. METHODS AND RESULTS: We prospectively recruited consecutive patients who had undergone right heart catheterization. The primary endpoint was cardiovascular death or readmission due to heart failure. We included 137 patients with Group 2 PH. A RV longitudinal strain of 17% was sensitive (85%) and specific (70%) to determine the CpcPH. During a median period of 31 months, 43 patients experienced the primary endpoint during follow-up. In a multivariate analysis, RV longitudinal strain was associated with the primary endpoint in both CpcPH and IpcPH (HR: 0.84, P = 0.003; HR: 0.86, P = 0.001). CONCLUSIONS: Lower RV longitudinal strain was independently associated with worse outcomes in CpcPH and IpcPH. RV longitudinal strain may play a prognostic role in PH phenotypes.

Association of Local Epicardial Adipose Tissue Depots and Left Ventricular Diastolic Performance in Patients With Preserved Left Ventricular Ejection Fraction.

BACKGROUND: Although full-volume quantification of epicardial adipose tissue (EAT) is a predictor of LV diastolic dysfunction (LVDD), how localized EAT depots are linked to LVDD remains unclear. We evaluated the effect of local EAT depots on LV diastolic function parameters in patients with preserved LV ejection fraction (LVEF).Methods and Results:From 423 consecutive patients who underwent cardiac CT angiography, we recruited 252 with sinus rhythm and normal LVEF. The EAT volume index (EATV/body surface area) and the localized EAT thickness around the right coronary artery (EATRCA), left anterior descending artery (EATLAD), left circumflex artery (EATLCX), right ventricle (EATRV), left ventricle (EATLV), right atrium (EATRA), and left atrium (EATLA) were measured using cardiac CT. In the LVDD group (n=71), the EATV index (75±30 vs. 64±28 mL/m2, P=0.010), EATLCX(10.7±3.8 vs. 9.4±3.4 mm, P=0.008), and EATLV(2.6±1.6 vs. 2.1±1.4 mm, P=0.024) were greater than in the non-LVDD group (n=181). In contrast, EATLCXand EATLVwere markedly associated with decreased lateral e' and increased lateral E/e'. Multiple regression analysis indicated that EATLCXand EATLVwere strongly associated with LV diastolic function parameters. CONCLUSIONS: Localized EAT depots are linked to altered mitral annular motion. Further study is warranted to clarify whether localized EAT depots are functionally linked to the clinical manifestations of LVDD.

Association between Sarcopenia/Lower Muscle Mass and Short-Term Regression of Deep Vein Thrombosis Using Direct Oral Anticoagulants.

Advanced age, obesity, and muscle weakness are independent factors in the onset of deep vein thrombosis (DVT). Recently, an association between sarcopenia and DVT has been reported. We hypothesized that sarcopenia related factors, observed by ultrasonography, are associated with the regression effect on the thrombus following anticoagulation therapy. The present study focused on gastrocnemius muscle (GCM) thickness and the GCM's internal echogenic brightness. We examined the association with DVT regression following direct oral anticoagulants (DOACs) treatment.The prospective cohort study period was between October 2017 and August 2018. We enrolled 46 patients diagnosed with DVT by ultrasonography, who were aged >60 years old and treated with DOACs. Sarcopenia was evaluated using the Asian Working Group for Sarcopenia flowchart. The average DOACs treatment period was 94 days, and 29 patients exhibited thrombus regression. On univariate logistic regression analysis, sarcopenia, average GCM diameter index, and gastrocnemius integrated backscatter index were significantly associated with thrombus regression. In a multivariate model, only the average GCM diameter index correlated with thrombus regression.The average GCM diameter index is associated with DVT regression treated with DOACs. Considering the GCM diameter during DVT treatment can be a marker to make a decision for the treatment of DVT.

Association of Echocardiography Before Major Elective Non-Cardiac Surgery With Improved Postoperative Outcomes　- Possible Implications for Patient Care.

BACKGROUND: Whether preoperative echocardiography improves postoperative outcomes is not well established, so we examined the value of echocardiographic assessment on the onset of postoperative heart failure (HF), and determining which patients benefitted most from undergoing echocardiography prior to major elective non-cardiac surgery.Methods and Results:We identified all patients aged 50 years and older who had major elective non-cardiac surgery, and excluded patients with previously identified severe cardiovascular disease. The primary endpoint was the onset of HF during hospitalization. A total of 806 patients were included in the analysis. During hospitalization, 49 patients (6%) reached the primary endpoint. Within the matched cohort, preoperative echocardiography was associated with a statistically significant decrease in postoperative HF (hazard ratio: 0.46, P=0.01). In subgroup analyses, age, sex, body surface area, hypertension, diabetes mellitus, prior HF, surgical type, chronic kidney disease, pulmonary disease, and malignancy influenced the association of echocardiography with postoperative HF. CONCLUSIONS: The use of echocardiography in elderly patients with certain risk factors was associated with improved postoperative outcomes. The basis for this finding remains to be determined; particularly whether echocardiography is simply a marker of a population with better outcomes or whether it leads to better management that improves outcomes.

Prognostic Value of Frailty and Diastolic Dysfunction in Elderly Patients.

BACKGROUND: High prevalence of frailty and of diastolic dysfunction (DD) in heart failure and high mortality in frail adults have been noted. We characterized frailty by quantifying differences on echocardiography, and assessed the added prognostic utility of frailty and DD grade in an elderly population. METHODS AND RESULTS: One hundred and ninety-one patients ≥65 years who had at least 1 cardiovascular risk factor were prospectively recruited for clinically indicated echocardiography at the present institute. Weight loss, exhaustion, and deficits in physical activity, gait speed, and handgrip strength were used to categorize patients as frail (≥3 features), intermediately frail (1 or 2 features), or non-frail (0 features). DD grade ≥2 was defined as severe. Frailty was associated with larger left atrial volume, smaller stroke volume, and worse DD grade after adjustment for age. In a period of 14 months, 29 patients (15%) had cardiovascular events. The addition of frailty score and severe DD significantly improved the prognostic power of a model containing male gender (model 1, male gender, χ2=6.4; model 2, model 1 plus frailty score, χ2=16.7, P=0.004; model 3, model 2 plus severe DD, χ2=25.5, P=0.015). CONCLUSIONS: Both frailty and DD grade were significantly associated with future cardiovascular events in an elderly population with preserved ejection fraction and ≥1 risk factor of cardiovascular disease.

Local Thickness of Epicardial Adipose Tissue Surrounding the Left Anterior Descending Artery Is a Simple Predictor of Coronary Artery Disease　- New Prediction Model in Combination With Framingham Risk Score.

BACKGROUND: Compared with global cardiac adiposity, the local accumulation of fat surrounding coronary arteries might have a more direct impact on coronary artery disease (CAD). Here, we compared the local epicardial adipose tissue (EAT) thickness and global cardiac adiposity volumes for predicting CAD.Methods and Results:A total of 197 consecutive subjects underwent 320-slice multi-detector computed tomography coronary angiography and were segregated into CAD (≥1 coronary artery branch stenosis ≥50%) and non-CAD groups. EAT thickness was measured at the right coronary artery (EATRCA), the left anterior descending artery (EATLAD), and the left circumflex artery (EATLCX). Although EATRCAand EATLCXwere similar between the 2 groups, EATLADwas larger in the CAD group than in the non-CAD group (5.45±2.16 mm vs. 6.86±2.19 mm, P<0.001). EATLAD, after correcting for confounding factors, was strongly associated with CAD (r=0.276, P<0.001) and Gensini score (r=0.239, P<0.001). On multiple regression analysis, Framingham risk score combined with EATLADwas a strong predictor of CAD (adjusted R2=0.121; P<0.001). CONCLUSIONS: The local fat thickness surrounding the LAD is a simple and useful surrogate marker for estimating the presence, severity, and extent of CAD, independent of classical cardiovascular risk factors.

Effect of the Epicardial Adipose Tissue Volume on the Prevalence of Paroxysmal and Persistent Atrial Fibrillation.

BACKGROUND: Although increasing evidence suggests that epicardial adipose tissue volume (EATV) is associated with atrial fibrillation (AF), it is controversial whether there is a dose-response relationship of increasing EATV along the continuum of AF. We evaluated the effect of the EATV on the prevalence of paroxysmal AF (PAF) and persistent AF (PeAF) and the relationships with cardiac structure and functional remodeling.Methods and Results:Subjects who underwent multidetector computed tomography (MDCT) coronary angiography because of symptoms suggestive of coronary artery disease were divided into sinus rhythm (SR) (n=112), PAF (n=133), and PeAF (n=71) groups. The EATV index (EATV/body surface area, mL/m2) was strongly associated with the prevalence of PAF and PeAF on the model adjusted for known AF risk factors. The effect of the EATV index on the prevalence of PeAF, but not on that of PAF, was modified by the left atrial (LA) dimension, suggesting that extension of the LA dimension is related to EATV expansion in PeAF. The cutoff value of the EATV index for the prevalence was higher in PeAF than in PAF (64 vs. 55 mL/m2, P<0.01). CONCLUSIONS: The EATV index is associated with the prevalence of PAF and PeAF, and its cutoff values are predictive for PAF and PeAF development independently of other AF risk factors.

A clinical application of preload stress echocardiography for predicting future hemodynamic worsening in patients with early-stage heart failure.

AIM: To improve the prognosis of patients with heart failure, risk stratification in their early stage is important. We assessed whether the change in transmitral flow (TMF) velocity pattern during preload augmentation can predict future hemodynamic worsening in early-stage heart failure patients with impaired relaxation TMF pattern. METHODS: We designed a prospective cohort study that included 155 consecutive patients with impaired relaxation (IR) pattern at rest. Preload stress echocardiography was achieved using leg-positive pressure (LPP), and changes in TMF pattern during the LPP was observed during baseline echocardiographic examination. The patients whose TMF pattern developed to pseudonormal (PN) pattern throughout the study period were classified into the change to PN group, and patients whose TMF pattern stayed in IR pattern were classified into the stay in IR group. RESULTS: The median follow-up period was 17 months. The average age was 68 ± 11 years old, and 97 patients (63%) were male. Among 155 patients, 27 were classified into the change to PN group. A Cox proportional hazard analysis confirmed that the change in the peak atrial systolic TMF velocity during the LPP (ΔA, hazard ratio = 0.58 per 1SD; 95% CI = 0.39-0.88, P = 0.010) was the powerful independent predictor of change into PN pattern. Kaplan-Meier analysis revealed that the patients with ΔA ≤ -7 cm/s had more likely to develop into PN pattern than patients with ΔA > -7 cm/s (P = 0.001). CONCLUSIONS: Evaluation of a response in TMF during the LPP might provide an incremental diagnostic value to detect future overt heart failure in patients with early-stage heart failure.

Prognostic Implications of Non-Invasive Vascular Function Tests in High-Risk Atherosclerosis Patients.

BACKGROUND: The aim of this study was to assess the role of clinically available vascular function tests as predictors of cardiovascular events and decline in kidney function. METHODS AND RESULTS: One hundred and fourteen patients who had at least 2 cardiovascular risk factors were recruited for vascular function assessment including ankle-brachial blood pressure index (ABI), brachial-ankle pulse wave velocity (baPWV), cardio-ankle vascular index (CAVI) and flow-mediated vasodilatation (%FMD). During a median period of 51 months, 35 patients reached the primary endpoint (29 cardiovascular events and 6 cardiac deaths), and 30 patients reached the secondary endpoint (decline in kidney function: defined as a 5% per year decline of estimated glomerular filtration rate). In sequential Cox models, a model on the basis of the Framingham risk score, hemoglobin, and high-sensitivity C-reactive protein (chi-squared, 16.6) was improved by the ABI (chi-squared: 21.5; P=0.047). The baPWV (hazard ratio: 1.42 per 1 SD increase; P=0.025) and the CAVI (hazard ratio: 1.52 per 1 SD increase; P=0.040) were associated with the secondary endpoint. The %FMD was only slightly associated with the primary and secondary endpoints. CONCLUSIONS: Both ABI and baPWV are significantly associated with future cardiovascular events in high-risk patients with cardiovascular disease. The predictive capabilities of these parameters are greater than that of other parameters in this cohort.

Comparison of carotid plaque tissue characteristics in patients with acute coronary syndrome or stable angina pectoris: assessment by iPlaque, transcutaneous carotid ultrasonography with integrated backscatter analysis.

BACKGROUND: The association of the tissue characteristics of carotid plaques with coronary artery disease has attracted interest. The present study compared the tissue characteristics of carotid plaques in patients with acute coronary syndrome (ACS) with those in patients with stable angina pectoris (SAP) using the iPlaque system, which is based on ultrasound integrated backscatter. METHODS AND RESULTS: Carotid ultrasound examinations were performed in 26 patients with ACS, and 38 age- and gender-matched patients with SAP. Neither plaque area nor maximal intima-media thickness differed significantly between the two groups. However, the average integrated backscatter value within the plaque was greater in the ACS patients than in the SAP patients. iPlaque analysis revealed that the percentage blue area (lipid pool) was greater in the ACS patients than in the SAP patients (43.4 ± 11.2 vs 18.3 ± 10.3%, p < 0.0001), and that the percentage green area (fibrosis) was lower in the ACS than in the SAP patients (7.5 ± 7.5% vs 20.7 ± 11.7%, p < 0.0001). CONCLUSIONS: The lipid component of carotid plaques is greater in ACS patients than in SAP patients. Our iPlaque system provides a useful and feasible method for the tissue characterization of carotid plaques in the clinical setting.

Optimal Analysis of Left Atrial Strain by Speckle Tracking Echocardiography: P-wave versus R-wave Trigger.

BACKGROUND: Left atrial (LA) strain analysis using speckle tracking echocardiography is useful for assessing LA function. However, there is no established procedure for this method. Most investigators have determined the electrocardiographic R-wave peak as the starting point for LA strain analysis. To test our hypothesis that P-wave onset should be used as the starting point, we measured LA strain using 2 different starting points and compared the strain values with the corresponding LA volume indices obtained by three-dimensional (3D) echocardiography. METHODS: We enrolled 78 subjects (61 ± 17 years, 25 males) with and without various cardiac diseases in this study and assessed global longitudinal LA strain by two-dimensional speckle tracking strain echocardiography using EchoPac software. We used either R-wave peak or P-wave onset as the starting point for determining LA strains during the reservoir (Rres, Pres), conduit (Rcon, Pcon), and booster pump (Rpump, Ppump) phases. We determined the maximum, minimum, and preatrial contraction LA volumes, and calculated the LA total, passive, and active emptying fractions using 3D echocardiography. RESULTS: The correlation between Pres and LA total emptying fraction was better than the correlation between Rres and LA total emptying fraction (r = 0.458 vs. 0.308, P = 0.026). Pcon and Ppump exhibited better correlation with the corresponding 3D echocardiographic parameters than Rcon (r = 0.560 vs. 0.479, P = 0.133) and Rpump (r = 0.577 vs. 0.345, P = 0.003), respectively. CONCLUSIONS: LA strain in any phase should be analyzed using P-wave onset as the starting point rather than R-wave peak.

Reducing echocardiographic examination time through routine use of fully automated software: a comparative study of measurement and report creation time.

BACKGROUND: Manual interpretation of echocardiographic data is time-consuming and operator-dependent. With the advent of artificial intelligence (AI), there is a growing interest in its potential to streamline echocardiographic interpretation and reduce variability. This study aimed to compare the time taken for measurements by AI to that by human experts after converting the acquired dynamic images into DICOM data. METHODS: Twenty-three consecutive patients were examined by a single operator, with varying image quality and different medical conditions. Echocardiographic parameters were independently evaluated by human expert using the manual method and the fully automated US2.ai software. The automated processes facilitated by the US2.ai software encompass real-time processing of 2D and Doppler data, measurement of clinically important variables (such as LV function and geometry), automated parameter assessment, and report generation with findings and comments aligned with guidelines. We assessed the duration required for echocardiographic measurements and report creation. RESULTS: The AI significantly reduced the measurement time compared to the manual method (159 ± 66 vs. 325 ± 94 s, p < 0.01). In the report creation step, AI was also significantly faster compared to the manual method (71 ± 39 vs. 429 ± 128 s, p < 0.01). The incorporation of AI into echocardiographic analysis led to a 70% reduction in measurement and report creation time compared to manual methods. In cases with fair or poor image quality, AI required more corrections and extended measurement time than in cases of good image quality. Report creation time was longer in cases with increased report complexity due to human confirmation of AI-generated findings. CONCLUSIONS: This fully automated software has the potential to serve as an efficient tool for echocardiographic analysis, offering results that enhance clinical workflow by providing rapid, zero-click reports, thereby adding significant value.

Echocardiographic artificial intelligence for pulmonary hypertension classification.

OBJECTIVE: The classification of pulmonary hypertension (PH) is crucial for determining the appropriate therapeutic strategy. We investigated whether machine learning (ML) algorithms may assist in echocardiographic PH prediction, where current guidelines recommend integrating several different parameters. METHODS: We obtained physical and echocardiographic data from 885 patients who underwent right heart catheterisation (RHC). Patients were classified into three groups: non-PH, precapillary PH and postcapillary PH, based on values obtained from RHC. Using 24 parameters, we created predictive models employing four different classifiers and selected the one with the highest area under the curve. We then calculated the macro-average classification accuracy for PH on the derivation cohort (n=720) and prospective validation data set (n=165), comparing the results with guideline-based echocardiographic assessment obtained from each cohort. RESULTS: Logistic regression with elastic net regularisation had the highest classification accuracy, with areas under the curves of 0.789, 0.766 and 0.742 for normal, precapillary PH and postcapillary PH, respectively. The ML model demonstrated significantly better predictive accuracy than the guideline-based echocardiographic assessment in the derivation cohort (59.4% vs 51.6%, p<0.01). In the independent validation data set, the ML model's accuracy was comparable to the guideline-based PH classification (59.4% vs 57.8%, p=0.638). CONCLUSIONS: This preliminary study suggests promising potential for our ML model in predicting echocardiographic PH. Further research and validation are needed to fully assess its clinical utility in PH diagnosis and treatment decision-making.

Correlation between energy loss index and B-type natriuretic peptide: a vector flow mapping study.

BACKGROUND: Vector Flow Mapping (VFM) and Energy Loss (EL) evaluation are emerging echocardiographic techniques that offer detailed insights into cardiac function. This study aimed to explore the relationship between EL parameters and B-type natriuretic peptide (BNP) levels, a well-established marker of heart failure severity. METHODS: Our study prospectively enrolled 62 patients experiencing shortness of breath and suspected heart failure, who underwent echocardiography and had BNP levels measured between January 2018 and August 2020. Patients were stratified based on BNP levels, and their clinical and echocardiographic characteristics were evaluated. Univariate and multivariate regression analyses were performed to assess the correlation between BNP levels and various echocardiographic variables, including VFM parameters. RESULTS: Patients were stratified into two groups based on their BNP levels: BNP < 200 pg/ml (n = 53) and BNP ≥ 200 pg/ml (n = 9). Patients with BNP ≥ 200 pg/ml presented significantly different clinical and echocardiographic characteristics, such as older age, larger left ventricular mass and volume indices, higher pulmonary artery systolic pressure, higher E/e' ratio, and larger EL parameters. Multivariate regression analysis demonstrated the E/e' ratio and ELA (EL during Atrial contraction phase/A wave ratio as significant determinants of logBNP. Receiver operating characteristic curve analysis showed ELA/A > 36.0 J/m2 as a significant predictor of high BNP with 89% sensitivity and 85% specificity. ELA/A demonstrated an incremental diagnostic value over elevated left atrial pressure for predicting high BNP (C statistic = 0.98 vs 0.74, P = 0.006). CONCLUSION: This study provides novel insights into the potential utility of EL parameters as auxiliary indicators of cardiac load, thereby enhancing our understanding of heart failure.

Unsupervised cluster analysis reveals different phenotypes in patients after transcatheter aortic valve replacement.

Aims: The aim of this study was to identify phenotypes with potential prognostic significance in aortic stenosis (AS) patients after transcatheter aortic valve replacement (TAVR) through a clustering approach. Methods and results: This multi-centre retrospective study included 1365 patients with severe AS who underwent TAVR between January 2015 and March 2019. Among demographics, laboratory, and echocardiography parameters, 20 variables were selected through dimension reduction and used for unsupervised clustering. Phenotypes and outcomes were compared between clusters. Patients were randomly divided into a derivation cohort (n = 1092: 80%) and a validation cohort (n = 273: 20%). Three clusters with markedly different features were identified. Cluster 1 was associated predominantly with elderly age, a high aortic valve gradient, and left ventricular (LV) hypertrophy; Cluster 2 consisted of preserved LV ejection fraction, larger aortic valve area, and high blood pressure; and Cluster 3 demonstrated tachycardia and low flow/low gradient AS. Adverse outcomes differed significantly among clusters during a median of 2.2 years of follow-up (P < 0.001). After adjustment for clinical and echocardiographic data in a Cox proportional hazards model, Cluster 3 (hazard ratio, 4.18; 95% confidence interval, 1.76-9.94; P = 0.001) was associated with increased risk of adverse outcomes. In sequential Cox models, a model based on clinical data and echocardiographic variables (χ2 = 18.4) was improved by Cluster 3 (χ2 = 31.5; P = 0.001) in the validation cohort. Conclusion: Unsupervised cluster analysis of patients after TAVR revealed three different groups for assessment of prognosis. This provides a new perspective in the categorization of patients after TAVR that considers comorbidities and extravalvular cardiac dysfunction.

Pulmonary pressure-flow responses to exercise in heart failure treated with angiotensin receptor neprilysin inhibitor.

BACKGROUND: The role of the angiotensin receptor neprilysin inhibitor (ARNI) in cardiac function, particularly its impact on pulmonary circulation, remains underexplored. Recent studies have described abnormal mean pulmonary artery pressure (mPAP)-cardiac output (CO) responses as having the potential to assess the disease state. The aim of this study was to assess the effects of ARNI on pulmonary circulation in heart failure. We measured echocardiographic parameters post 6-min walk (6 MW) and compared the changes with baseline and follow-up. Our hypothesis was that pulmonary pressure-flow relationship of the pulmonary circulation obtained by 6 MW stress echocardiography would be improved with treatment. METHODS: We prospectively enrolled 39 heart failure patients and conducted the 6 MW test indoors. Post-6 MW echocardiography measured echocardiographic variables, and CO was derived from electric cardiometry. Individualized ARNI doses were optimized, with follow-up echocardiographic evaluations after 1 year. RESULTS: Left ventricular (LV) volume were significantly reduced (160.7 ± 49.6 mL vs 136.0 ± 54.3 mL, P < 0.001), and LV ejection fraction was significantly improved (37.6 ± 11.3% vs 44.9 ± 11.5%, P < 0.001). Among the 31 patients who underwent 6 MW stress echocardiographic study at baseline and 1 year later, 6 MW distance increased after treatment (380 m vs 430 m, P = 0.003). The ΔmPAP/ΔCO by 6 MW stress decreased with treatment (6.9 mmHg/L/min vs 2.8 mmHg/L/min, P = 0.002). The left atrial volume index was associated with the response group receiving ARNI treatment for pulmonary circulation. CONCLUSIONS: Initiation of ARNI was associated with improvement of left ventricular size and LVEF. Additionally, the 6 MW distance increased and the ΔmPAP/ΔCO was improved to within normal range with treatment.

Deep learning approach for analyzing chest x-rays to predict cardiac events in heart failure.

Background: A deep learning (DL) model based on a chest x-ray was reported to predict elevated pulmonary artery wedge pressure (PAWP) as heart failure (HF). Objectives: The aim of this study was to (1) investigate the role of probability of elevated PAWP for the prediction of clinical outcomes in association with other parameters, and (2) to evaluate whether probability of elevated PAWP based on DL added prognostic information to other conventional clinical prognostic factors in HF. Methods: We evaluated 192 patients hospitalized with HF. We used a previously developed AI model to predict HF and calculated probability of elevated PAWP. Readmission following HF and cardiac mortality were the primary endpoints. Results: Probability of elevated PAWP was associated with diastolic function by echocardiography. During a median follow-up period of 58 months, 57 individuals either died or were readmitted. Probability of elevated PAWP appeared to be associated with worse clinical outcomes. After adjustment for readmission score and laboratory data in a Cox proportional-hazards model, probability of elevated PAWP at pre-discharge was associated with event free survival, independent of elevated left atrial pressure (LAP) based on echocardiographic guidelines (p < 0.001). In sequential Cox models, a model based on clinical data was improved by elevated LAP (p = 0.005), and increased further by probability of elevated PAWP (p < 0.001). In contrast, the addition of pulmonary congestion interpreted by a doctor did not statistically improve the ability of a model containing clinical variables (compared p = 0.086). Conclusions: This study showed the potential of using a DL model on a chest x-ray to predict PAWP and its ability to add prognostic information to other conventional clinical prognostic factors in HF. The results may help to enhance the accuracy of prediction models used to evaluate the risk of clinical outcomes in HF, potentially resulting in more informed clinical decision-making and better care for patients.