Acute myocardial infarction prognosis prediction with reliable and interpretable artificial intelligence system.

OBJECTIVE: Predicting mortality after acute myocardial infarction (AMI) is crucial for timely prescription and treatment of AMI patients, but there are no appropriate AI systems for clinicians. Our primary goal is to develop a reliable and interpretable AI system and provide some valuable insights regarding short, and long-term mortality. MATERIALS AND METHODS: We propose the RIAS framework, an end-to-end framework that is designed with reliability and interpretability at its core and automatically optimizes the given model. Using RIAS, clinicians get accurate and reliable predictions which can be used as likelihood, with global and local explanations, and "what if" scenarios to achieve desired outcomes as well. RESULTS: We apply RIAS to AMI prognosis prediction data which comes from the Korean Acute Myocardial Infarction Registry. We compared FT-Transformer with XGBoost and MLP and found that FT-Transformer has superiority in sensitivity and comparable performance in AUROC and F1 score to XGBoost. Furthermore, RIAS reveals the significance of statin-based medications, beta-blockers, and age on mortality regardless of time period. Lastly, we showcase reliable and interpretable results of RIAS with local explanations and counterfactual examples for several realistic scenarios. DISCUSSION: RIAS addresses the "black-box" issue in AI by providing both global and local explanations based on SHAP values and reliable predictions, interpretable as actual likelihoods. The system's "what if" counterfactual explanations enable clinicians to simulate patient-specific scenarios under various conditions, enhancing its practical utility. CONCLUSION: The proposed framework provides reliable and interpretable predictions along with counterfactual examples.

Perioperative adverse cardiac events and mortality after non-cardiac surgery: a multicenter study.

BACKGROUND: Perioperative adverse cardiac events (PACE), a composite of myocardial infarction, coronary revascularization, congestive heart failure, arrhythmic attack, acute pulmonary embolism, cardiac arrest, and stroke during 30-day postoperative period, is associated with long-term mortality, but with limited clinical evidence. We compared long-term mortality with PACE using data from nationwide multicenter electronic health records. METHODS: Data from 7 hospitals, converted to Observational Medical Outcomes Partnership Common Data Model, were used. We extracted records of 277,787 adult patients over 18 years old undergoing non-cardiac surgery for the first time at the hospital and had medical records for more than 180 days before surgery. We performed propensity score matching and then an aggregated meta­analysis. RESULTS: After 1:4 propensity score matching, 7,970 patients with PACE and 28,807 patients without PACE were matched. The meta­analysis showed that PACE was associated with higher one-year mortality risk (hazard ratio [HR]: 1.33, 95% CI [1.10, 1.60], P = 0.005) and higher three-year mortality (HR: 1.18, 95% CI [1.01, 1.38], P = 0.038). In subgroup analysis, the risk of one-year mortality by PACE became greater with higher-risk surgical procedures (HR: 1.20, 95% CI [1.04, 1.39], P = 0.020 for low-risk surgery; HR: 1.69, 95% CI [1.45, 1.96], P < 0.001 for intermediate-risk; and HR: 2.38, 95% CI [1.47, 3.86], P = 0.034 for high-risk). CONCLUSIONS: A nationwide multicenter study showed that PACE was significantly associated with increased one-year mortality. This association was stronger in high-risk surgery, older, male, and chronic kidney disease subgroups. Further studies to improve mortality associated with PACE are needed.

Effect of High-Intensity Rosuvastatin vs. Combination of Low-Intensity Rosuvastatin and Ezetimibe on HbA1c Levels in Patients without Diabetes: A Randomized IDEAL Trial.

There is a dearth of studies investigating whether the combination of low-intensity statins with ezetimibe can reduce the risk of diabetes in patients requiring statin therapy. Therefore, we aimed to evaluate the effects of combination therapy on the prevention of glycated hemoglobin (HbA1c) elevation in patients without diabetes. Sixty-eight patients were randomly assigned in a 1:1 ratio to receive a combination of low-intensity rosuvastatin (5 mg/day) and ezetimibe (10 mg/day) or high-intensity rosuvastatin (20 mg/day). The primary endpoint was the absolute difference in the HbA1c levels at 12 weeks. The HbA1c level showed an overall elevation of 0.11% at 12 weeks compared to that at baseline (mean ± standard deviation: 5.78 ± 0.3%, 95% confidence interval [CI]: 5.86-6.07, p = 0.044). The HbA1c levels did not differ between the groups at 12 weeks (least square mean difference: 0.001, 95% CI: 0.164-0.16, p = 0.999). Our study found that the combination of low-intensity rosuvastatin and ezetimibe did not yield significant differences in HbA1c levels compared to high-intensity rosuvastatin alone after 12 weeks in patients without diabetes. This suggests that the combination of low-intensity rosuvastatin and ezetimibe may not be an effective strategy for preventing HbA1c elevation in patients without diabetes requiring statins.

Adjusted mortality of extracorporeal membrane oxygenation for acute myocardial infarction patients in cardiogenic shock.

Cardiogenic shock (CS) is a common cause of death following acute myocardial infarction (MI). This study aimed to evaluate the adjusted mortality of venoarterial extracorporeal membrane oxygenation (VA-ECMO) with intra-aortic balloon counterpulsation (IABP) for patients with MI-CS. We included 300 MI patients selected from a multinational registry and categorized into VA-ECMO + IABP (N = 39) and no VA-ECMO (medical management ± IABP) (N = 261) groups. Both groups' 30-day and 1-year mortality were compared using the weighted Kaplan-Meier, propensity score, and inverse probability of treatment weighting methods. Adjusted incidences of 30-day (VA-ECMO + IABP vs No VA-ECMO, 77.7% vs 50.7; P = .083) and 1-year mortality (92.3% vs 84.8%; P = .223) along with propensity-adjusted and inverse probability of treatment weighting models in 30-day (hazard ratio [HR], 1.57; 95% confidence interval [CI], 0.92-2.77; P = .346 and HR, 1.44; 95% CI, 0.42-3.17; P = .452, respectively) and 1-year mortality (HR, 1.56; 95% CI, 0.95-2.56; P = .076 and HR, 1.33; 95% CI, 0.57-3.06; P = .51, respectively) did not differ between the groups. However, better survival benefit 30 days post-ECMO could be supposed (31.6% vs 83.4%; P = .022). Therefore, patients with MI-CS treated with IABP with additional VA-ECMO and those not supported with ECMO have comparable overall 30-day and 1-year mortality risks. However, VA-ECMO-supported survivors might have better long-term clinical outcomes.

False-positive Xpert® Xpress SARS-CoV-2 assay in an emergency room and trauma center: A retrospective chart review study.

OBJECTIVES: To review reports false-positive Xpert results in an emergency room and trauma center. METHODS: Patients' data with false-positive Xpert results from November 2020 to February 2022 at Pusan National University Hospital, Busan, Republic of Korea, were extracted from the electronic medical records. RESULTS: The positive predictive value of Xpert was 40%. Of the 12 patients with false-positive results, 5 (41.7%) were re-positives (such as, patients recovered from coronavirus disease-19 [COVID-19]), and 4 (33.3%) had head or facial trauma. Two out of 4 head or facial trauma cases had documented sample contamination with blood. CONCLUSION: We found a high incidence of false-positive Xpert results among patients who recovered from COVID-19 and those with head or facial injury. Careful history taking for COVID-19 and physical examination of the sample collection site is essential before Xpert analysis.

Incremental Prognostic Value of Left Ventricular Global Longitudinal Strain in Patients with Preserved Ejection Fraction Undergoing Transcatheter Aortic Valve Implantation.

BACKGROUND: Evaluation of left ventricular global longitudinal strain (LVGLS) has allowed better characterization than left ventricular ejection fraction (LVEF) of subtle differences in left ventricular performance. The aim of this study was to determine whether LVGLS has prognostic value in patients with severe aortic stenosis and preserved LVEF undergoing transcatheter aortic valve implantation (TAVI). METHODS: Among 412 consecutive patients who underwent TAVI, 344 patients (mean age, 78.9 ± 5.0 years; 161 men) with preserved LVEF (≥50%) at baseline were analyzed. Patients with low LVEF (<50%) were used as a comparison group in the survival analysis. The primary and secondary end points were all-cause death and a composite of cardiovascular death and heart failure hospitalization, respectively. RESULTS: The mean LVGLS and LVEF were -17.1 ± 2.7% and 63 ± 5%, respectively. Impaired LVGLS was defined as >-16%, which was the first LVGLS quartile (cutoff value, -16%). The estimated actuarial 5-year survival rate was 81.7 ± 4.2% in the normal LVGLS group and 66.8 ± 7.5% in the impaired LVGLS group (P = .005). In the multivariable analysis, impaired LVGLS was an independent predictor of all-cause death (adjusted hazard ratio, 2.26; 95% CI, 1.11-4.60) and the composite outcome (adjusted hazard ratio, 3.03; 95% CI, 1.45-6.33). Moreover, the impaired LVGLS group had a poor prognosis, similar to the impaired LVEF group (<50%). The addition of the absolute value of LVGLS to the clinical parameters and LVEF led to significant improvement in the prediction of all-cause mortality. CONCLUSIONS: In patients with severe aortic stenosis and preserved LVEF undergoing TAVI, subclinical LV dysfunction defined by impaired LVGLS is independently associated with poor clinical outcomes. LVGLS measurement provides incremental prognostic value above the established clinical and echocardiographic parameters.

Role of early short-term cardiac rehabilitation in patients undergoing pacemaker implantation.

Exercise-based cardiac rehabilitation (CR) improves the clinical outcomes in patients with cardiovascular diseases. However, few data exist regarding the role of early short-term CR in patients undergoing pacemaker (PM) implantation. We assessed whether short-term CR following PM implantation was sufficient to improve both physical function and quality of life (QOL). A total of 27 patients with a 6-minute walking distance (6MWD) of less than 85% of the predicted value on the day following PM implantation were randomly assigned to either the CR group (n = 12, 44.4%) or the non-CR group (n = 15, 55.6%). The CR group involved individualized exercise-based training with moderate intensity for 4 weeks after PM implantation. Cardiopulmonary exercise test (CPET), 6MWD, muscle strength, and Short Form (SF)-36 were assessed at baseline and at the 4-week follow-up. After a mean follow-up period of 38.3 days, both groups showed significantly improved 6MWD. Peak oxygen uptake improved in both groups on CPET, but the difference was not statistically significant. Knee extension power and handgrip strength were similar in both groups. Regarding QOL, only the CR group showed improved SF-36 scores in the items of vitality and mental health. There was no difference in any subscale in the non-CR group. Neither lead dislodgement nor significant changes in PM parameters were observed in any patient. Early short-term CR following PM implantation was associated with improved psychological subscales and can be safely performed without increasing the risk of procedure-related complications.

Prognostic impact of left ventricular mass regression after transcatheter aortic valve replacement in patients with left ventricular hypertrophy.

BACKGROUND: Paravalvular regurgitation (PVR) has been known to be the primary determinant of poor left ventricular (LV) mass regression after transcatheter aortic valve replacement (TAVR). However, the incidence of significant PVR has been reduced considerably as TAVR technology evolved rapidly. This study aimed to investigate the time course and impact of LV mass index (LVMi) regression on long-term clinical outcomes in severe aortic stenosis (AS) patients without significant PVR after TAVR. METHODS: Of 412 patients who underwent TAVR, 146 who had LV hypertrophy (LVMi ≥115 g/m2 for men and ≥ 95 g/m2 for women) at baseline and were alive at one year after TAVR were enrolled. The primary outcome was cardiovascular deaths and the impact of LVMi regression on clinical outcomes were examined. The patients with significant PVR were excluded. RESULTS: During a median follow-up of 40 months (interquartile range, 26-58 months), 9 (6.2%) cardiovascular deaths, 21 (14.4%) all-cause deaths, and 9 (6.2%) hospitalizations occurred. In the multivariable analysis, the percentage change of LVMi was an independent predictor of cardiovascular deaths (adjusted hazard ratio [HR], 1.03; 95% confidential interval [CI], 1.01-1.05; P = 0.010), and composite outcome of cardiovascular deaths and rehospitalization for heart failure (adjusted HR, 1.02; 95% CI, 1.00-1.04; P = 0.022). Baseline LVMi, eccentric hypertrophy, and TAVR-induced left bundle branch block were independently associated with LVMi regression. CONCLUSIONS: In patients with severe AS who received successful TAVR without significant PVR, the degree of LVMi regression is an independent predictor of postoperative outcomes after TAVR.

Clinical Situations Associated with Inappropriately Large Regurgitant Volumes in the Assessment of Mitral Regurgitation Severity Using the Proximal Flow Convergence Method in Patients with Chordae Rupture.

BACKGROUND: Regurgitant volume (RVol) calculated using the proximal flow convergence method (proximal isovelocity surface area [PISA]) has been accepted as a key quantitative parameter for the diagnosis of and clinical decision-making with regard to severe mitral regurgitation (MR). However, a recent prospective study showed a significant overestimation of RVol by the echocardiographic PISA method compared with the MR volume measured using magnetic resonance imaging. We aimed to evaluate the frequency of overestimation of RVol by the PISA method and the clinical conditions that require a different quantitative method to correct the overestimation. METHODS: We retrospectively enrolled 166 consecutive patients with degenerative MR and chordae rupture, in whom RVol was measured using both the PISA and two-dimensional Doppler volumetric methods. The volumetric method was used to measure total stroke volume using the two-dimensional Simpson biplane method, and forward stroke volume was measured using pulsed Doppler tracing at the left ventricular (LV) outflow tract. RVol by the volumetric method was calculated using total stroke volume - forward stroke volume. Severe MR was defined as an RVol >60 mL. RESULTS: All patients had severe MR based on RVol by the PISA method, but 68 (41.1%) showed RVol by the volumetric method values of <60 mL, resulting in discordant results. The patients with discordant results were characterized by a higher prevalence of female sex, lower body surface area, smaller LV diastolic and systolic dimensions and volumes, smaller left atrial volume, smaller PISA angle, and lower frequency of flail leaflets (39.7% vs 62.2%, P = .004). Multivariate analysis revealed that LV end-diastolic volume (LVEDV) and PISA angle were independent factors, with the best cutoff LVEDV and PISA angle being 173 mL and 103°, respectively. During follow-up (median, 3.4 years; interquartile range, 2.0-4.8 years), mitral valve repair and replacement were performed in 103 and six patients, respectively. The 2-year mitral valve surgery-free survival rate was higher in the discordant group (51.8% ± 0.06% vs 31.2% ± 0.05%, P < .001). CONCLUSIONS: Even in the patients with documented chordae rupture, the PISA method alone resulted in inappropriate overestimation of MR severity in a significant proportion of patients. Thus, an additive quantitative method is absolutely necessary in patients with a small LVEDV or narrow PISA angle.

Early decongestive therapy versus high-flow nasal cannula for the prevention of adverse clinical events in patients with acute cardiogenic pulmonary edema.

BACKGROUND: Few studies have investigated the role of decongestive therapy and high-flow nasal cannula (HFNC) in preventing reintubation and in-hospital mortality in patients with acute cardiogenic pulmonary edema (ACPE). METHODS: Data from patients with ACPE who were weaned from mechanical ventilation in the cardiac intensive care unit between January 2013 and December 2017 were retrospectively evaluated. All patients were treated with HFNC or conventional oxygen therapy (COT), such as a nasal cannula or venturi mask, immediately after extubation. Decongestive therapy (intravenous furosemide infusion) was administered at the discretion of the attending physician. RESULTS: Of 212 patients treated during the study period, 47 were excluded due to recent open-heart surgery and two, due to insufficient clinical data. The remaining 163 patients had a mean age of 67.4±14.3 years, and 92 (56.4%) were male; 44 patients received HFNC, and 119 COT. Mean weight loss within 72 hours of extubation was -0.86±2.03 kg. A total of 38 patients (23.3%) required reintubation, 21 of whom (12.9%) required reintubation within 72 hours of extubation. In-hospital mortality occurred in 16 patients (9.8%). Multivariate analysis showed that weight increase within 72 hours of extubation was independent determinants of reintubation (OR =1.7; 95% CI: 1.2-2.2; P<0.001) and in-hospital mortality (OR =1.5; 95% CI: 1.1-2.1; P=0.005). The use of HFNC was not associated with reintubation or in-hospital mortality. CONCLUSIONS: Our findings indicate that early weight loss resulted in reduced reintubation and in-hospital mortality in patients with ACPE. However, HFNC and COT did not differ in the prevention of reintubation and in-hospital mortality. Therefore, aggressive decongestive therapy, rather than HFNC, should be considered early after extubation.

Erratum: Correction of Figure in the Article: Normal Reference Plots for the Bioelectrical Impedance Vector in Healthy Korean Adults.

This corrects the article on e198 in vol. 34, PMID: 31373183.

Normal Reference Plots for the Bioelectrical Impedance Vector in Healthy Korean Adults.

BACKGROUND: Accurate volume measurement is important in the management of patients with congestive heart failure or renal insufficiency. A bioimpedance analyser can estimate total body water in litres and has been widely used in clinical practice due to its non-invasiveness and ease of results interpretation. To change impedance data to volumetric data, bioimpedance analysers use equations derived from data from healthy subjects, which may not apply to patients with other conditions. Bioelectrical impedance vector analysis (BIVA) was developed to overcome the dependence on those equations by constructing vector plots using raw impedance data. BIVA requires normal reference plots for the proper interpretation of individual vectors. The aim of this study was to construct normal reference vector plots of bioelectrical impedance for Koreans. METHODS: Bioelectrical impedance measurements were collected from apparently healthy subjects screened according to a comprehensive physical examination and medical history performed by trained physicians. Reference vector contours were plotted on the RXc graph using the probability density function of the bivariate normal distribution. We further compared them with those of other ethnic groups. RESULTS: A total of 242 healthy subjects aged 22 to 83 were recruited (137 men and 105 women) between December 2015 and November 2016. The centers of the tolerance ellipses were 306.3 Ω/m and 34.9 Ω/m for men and 425.6 Ω/m and 39.7 Ω/m for women. The ellipses were wider for women than for men. The confidence ellipses for Koreans were located between those for Americans and Spaniards without overlap for both genders. CONCLUSION: This study presented gender-specific normal reference BIVA plots and corresponding tolerance and confidence ellipses on the RXc graph, which is important for the interpretation of BIA-reported volume status in patients with congestive heart failure or renal insufficiency. There were noticeable differences in reference ellipses with regard to gender and ethnic groups.

Mid-term Clinical Outcomes in a Cohort of Asymptomatic or Mildly Symptomatic Korean Patients with Bicuspid Aortic Valve in a Tertiary Referral Hospital.

BACKGROUND: Although bicuspid aortic valve (BAV) is one of the most common congenital heart diseases, clinical outcome data regarding BAV are still limited. We evaluated clinical characteristics and mid-term clinical outcomes of asymptomatic Korean patients with bicuspid aortic valve. METHODS: We initiated a prospective registry in 2014 at a tertiary referral hospital. To develop a cohort of asymptomatic patients, we excluded patients who previously underwent open heart surgery (OHS) or who had OHS within 6 months of referral. RESULTS: A total of 170 patients (117 male [69%], age 50 ± 13 years) were enrolled. More than 70% (n = 124, 73%) were incidentally diagnosed with BAV during routine health examinations or preoperative screening for non-cardiac surgery. At the time of enrollment, moderate to severe aortic stenosis (AS) or regurgitation (AR) was present in 77 patients (45%) and 98 (58%) showed aortic dilation: 42 (25%) had non-significant valvular dysfunction without aortic dilation. During a median follow-up of 4 years, AS severity increased significantly (p < 0.001), while there was no significant change in AR severity (p = 0.361). A total of 27 patients (16%) underwent OHS, including isolated aortic valve (AV) surgery (n = 11, 41%), AV with combined aortic surgery (n = 12, 44%), and isolated aortic surgery (n = 4, 15%): no patient developed aortic dissection. Moderate to severe AS (hazard ratio [HR] 4.61; 95% confidence interval [CI] 1.83-11.62; p = 0.001), NYHA class ≥ 2 (HR 2.53; 95% CI 1.01-6.35; p = 0.048) and aortic dilation (HR 2.13; 95% CI 0.87-5.21; p = 0.099) were associated with surgical events. CONCLUSIONS: Progression patterns of valvular dysfunction and impacts of BAV phenotype on OHS should be explored in future studies with longer follow-up durations.