Using artificial intelligence to predict post-operative outcomes in congenital heart surgeries: a systematic review

INTRODUCTION

Congenital heart disease (CHD) represents the most common group of congenital anomalies, constitutes a significant contributor to the burden of non-communicable diseases, highlighting the critical need for improved risk assessment tools. Artificial intelligence (AI) holds promise in enhancing outcome predictions for congenital cardiac surgery. This study aims to systematically review the utilization of AI in predicting post-operative outcomes in this population.

METHODS

Following PRISMA guidelines, a comprehensive search of Pubmed, Scopus, and Web of Science databases was conducted. Two independent reviewers screened articles based on predefined criteria. Included studies focused on AI models predicting various post-operative outcomes in congenital heart surgery.

RESULTS

The review included 35 articles, primarily published within the last four years, indicating growing interest in AI applications. Models predominantly targeted mortality and survival (n = 16), prolonged length of hospital or ICU stay (n = 7), postoperative complications (n = 6), prolonged mechanical ventilatory support time (n = 4), with additional focus on specific outcomes such as peri-ventricular leucomalacia (n = 2) and malnutrition (n = 1). Performance metrics, such as area under the curve (AUC), ranged from 0.52 to 0.997. Notably, these AI models consistently outperformed traditional risk stratification categories. For instance, in assessing the risk of morbidity and mortality, the AI models demonstrated superior performance compared to conventional methods.

CONCLUSION

AI-driven prediction models show significant promise in improving outcome predictions for congenital heart surgery. They surpass traditional risk prediction tools not only in immediate postoperative risks but also in long-term outcomes such as 1-year survival and malnutrition. Further studies with robust external validation are necessary to assess the practical applicability of these models in clinical settings. The protocol of this review was prospectively registered on PROSPERO (CRD42024550942).

Clinical Characteristics and Outcomes in Adults With Moderate-to-Severe Complexity Congenital Heart Disease Undergoing Palliation or Surgical Repair

Background

Congenital heart disease (CHD) survival rate has improved dramatically due to advances in diagnostic and therapeutic techniques. However, concerning the unrepaired CHD population of moderate and severe complexity, the data regarding risk predictors and surgical outcomes are scarce. Our aim was to describe the surgical results and predictors of in-hospital outcomes in adult patients with moderate-to-severe complexity CHD that were not repaired in childhood.

Methods

We conducted a retrospective cohort study that included 49 adult patients with moderate-to-complex CHD who were treated in a single medical centre. Clinical and echocardiographic variables were obtained on admission, after surgical procedures and during follow-up.

Results

Most of the patients were female (66%). Left ventricular ejection fraction and right ventricular outflow tract fractional shortening were within the normal range. The median pulmonary artery systolic pressure was 37 (27-55) mm Hg. The median time was 118 (80-181) minutes for extracorporeal circulation and 76 (49-121) minutes for aortic cross-clamping. The most frequent complication was postoperative complete atrioventricular block (12.2%). In-hospital survival rate was 87.7%. The development of low cardiac output syndrome with predominant right ventricle failure in the postoperative period was the most important predictor of in-hospital death (P = 0.03).

Conclusions

Deciding to treat adults with CHD is challenging in moderate and severe unrepaired cases. Adequate clinical, functional, and imaging evaluation is essential to determine each patient's suitability for surgical management and to achieve the best clinical outcome for this population.

Special Considerations in Critical Care of the Congenital Heart Disease Patient

Outcomes of congenital heart disease have improved markedly over the past 20 years, with survival to adulthood now close to 90%. The mean age of admission to an intensive care unit (ICU) is 40 years. The incidence of hospital and critical care admissions have increased significantly as a consequence of this improved survival. Intensivists are now confronted with the management not only of complex adult congenital heart disease (ACHD) lesions from a cardiac perspective, but also of extracardiac organ consequences of years of abnormal circulation after surgical or palliative correction. Kidney and liver dysfunction and respiratory and hematologic abnormalities are very common in this population. ACHD patients can present to the ICU for a vast number of reasons, classified in this review as medical noncardiac, medical cardiac, and surgical. Community/hospital-acquired infections, cerebrovascular accidents, and respiratory failure, alongside arrhythmias and heart failure, are responsible for medical admissions. Surgical admissions include postoperative management after correction or palliation, but also medical optimisation and work-up for advanced therapies. ICU management of this large heterogeneous group requires a thorough understanding of the pathophysiology in order to apply conventional adult critical care modalities; left ventricular or right ventricular dysfunction, pulmonary hypertension, intracardiac, extracardiac, and palliative surgical shunts can be present and require additional consideration. This review focuses on the pathophysiology, long-term sequelae, and different treatment modalities to supply a framework for the ICU physician caring for these patients. Successful outcome, especially in complex lesions, depends on early involvement of specialised ACHD centres.

Enhanced Assessment of Perioperative Mortality Risk in Adults With Congenital Heart Disease

BACKGROUND

In-hospital mortality is a rare, yet feared complication following cardiac surgery in adult congenital heart disease (ACHD). A risk score, developed and validated in ACHD, can be helpful to optimize risk assessment.

OBJECTIVES

The purpose of this study was to assess the performance of EuroSCORE II components and procedure-related Adult Congenital Heart Surgery (ACHS) score, identify additional risk factors, and develop a novel risk score for predicting in-hospital mortality after ACHD surgery.

METHODS

We assessed perioperative survival in patients aged >16 years undergoing congenital heart surgery in a large tertiary center between 2003 and 2019. A risk variable-derived PEACH (PErioperative ACHd) score was calculated for each patient. Internal and external validation of the model was undertaken, including testing in a validation cohort of patients operated in a second European ACHD center.

RESULTS

The development cohort comprised 1,782 procedures performed during the study period. Re-sternotomy was undertaken in 897 (50.3%). There were 31 (1.7%) in-hospital deaths. The PEACH score showed excellent discrimination ability (area under the curve [AUC]: 0.88; 95% CI: 0.83-0.94), and performed better than the ACHS score in our population (ACHS AUC: 0.69; 95% CI: 0.6-0.78; P = 0.0003). A simple 3-tiered risk stratification was formed: PEACH score 0 (in-hospital mortality 0.2%), 1-2 (3.6%), and ≥3 (17.2%). In a validation cohort of 975 procedures, the PEACH score retained its discriminative ability (AUC: 0.75; 95% CI: 0.72-0.77) and was well calibrated (Hosmer-Lemeshow chi-square goodness-of-fit P = 0.55). There was agreement in expected and observed perioperative mortality between cohorts.

CONCLUSIONS

The PEACH score is a simple, novel perioperative risk score developed and validated specifically for ACHD patients undergoing cardiac surgery.

Prediction Models and Scores in Adult Congenital Heart Disease

Nowadays, most patients with congenital heart disease survive to adulthood due to advances in pediatric cardiac surgery but often present with various comorbidities and long-term complications, posing challenges in their management. The development and clinical use of risk scores for the prediction of morbidity and/or mortality in adults with congenital heart disease (ACHD) is fundamental in achieving optimal management for these patients, including appropriate follow-up frequency, treatment escalation, and timely referral for invasive procedures or heart transplantation. In comparison with other fields of cardiovascular medicine, there are relatively few studies that report prediction models developed in the ACHD population, given the small sample size, heterogeneity of the population, and relatively low event rate. Some studies report risk scores originally developed in pediatric congenital or non-congenital population, externally validated in ACHD with variable success. Available risk scores are designed to predict heart failure or arrhythmic events, all-cause mortality, post-intervention outcomes, infective endocarditis, or atherosclerosis-related cardiovascular disease in ACHD. A substantial number of these scores are derived from retrospective studies and are not internally or externally validated. Adequately validated risk scores can be invaluable in clinical practice and an important step towards personalized medicine. Multicenter collaboration, adequate study design, and the potential use of artificial intelligence are important elements in the effort to develop reliable risk scores for the ACHD population.

Risk evaluation in adult congenital heart surgery: analysis of the Society of Thoracic Surgeons Congenital Heart Surgery Database risk models on data from the European Congenital Heart Surgeons Association Congenital Database

 

OBJECTIVES

We sought to evaluate the predictive power of the Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery (STAT) mortality score and the adult congenital heart surgery (ACHS) mortality score for the adults undergoing congenital heart operations entered into the European Congenital Heart Surgeons Association (ECHSA) database.

METHODS

The data set comprised 17 662 major operations performed between 1997 and 2019, on patients 18 years of age or older, in European centres participating in the ECHSA database. Each operation was assigned a STAT mortality score and category and an ACHS mortality score. Operative mortality was based on the 30-day status and on the status at hospital discharge. The discriminatory power of the STAT and ACHS scores was assessed using the area under the receiver operating characteristic curve (c-index).

RESULTS

A total of 17 214 (97.46%) operations were assigned ACHS scores. The 3 most frequent primary procedures were closure of the atrial septal defect (19.0%), aortic valve replacement (8.8%) and non-valve-sparing aortic root replacement (6.1%). Operative mortality for ACHS-coded operations was 2.07%. The procedures with the highest mortality were atrial septal defect creation/enlargement (19.0%), lung transplantation (18.8%) and heart transplantation (18.2%). A total of 17 638 (99.86%) operations were assigned a STAT score and category. The operative mortality for STAT-coded operations was 2.27%. The c-index for mortality was 0.720 for the STAT mortality score and 0.701 for the ACHS score.

CONCLUSIONS

The ACHS mortality score and the STAT mortality score reached similar, moderate predictive power in adult patients undergoing congenital heart surgery in ECHSA database.

Improving preoperative risk-of-death prediction in surgery congenital heart defects using artificial intelligence model: A pilot study

BACKGROUND

Congenital heart disease accounts for almost a third of all major congenital anomalies. Congenital heart defects have a significant impact on morbidity, mortality and health costs for children and adults. Research regarding the risk of pre-surgical mortality is scarce.

OBJECTIVES

Our goal is to generate a predictive model calculator adapted to the regional reality focused on individual mortality prediction among patients with congenital heart disease undergoing cardiac surgery.

METHODS

Two thousand two hundred forty CHD consecutive patients' data from InCor's heart surgery program was used to develop and validate the preoperative risk-of-death prediction model of congenital patients undergoing heart surgery. There were six artificial intelligence models most cited in medical references used in this study: Multilayer Perceptron (MLP), Random Forest (RF), Extra Trees (ET), Stochastic Gradient Boosting (SGB), Ada Boost Classification (ABC) and Bag Decision Trees (BDT).

RESULTS

The top performing areas under the curve were achieved using Random Forest (0.902). Most influential predictors included previous admission to ICU, diagnostic group, patient's height, hypoplastic left heart syndrome, body mass, arterial oxygen saturation, and pulmonary atresia. These combined predictor variables represent 67.8% of importance for the risk of mortality in the Random Forest algorithm.

CONCLUSIONS

The representativeness of "hospital death" is greater in patients up to 66 cm in height and body mass index below 13.0 for InCor's patients. The proportion of "hospital death" declines with the increased arterial oxygen saturation index. Patients with prior hospitalization before surgery had higher "hospital death" rates than who did not required such intervention. The diagnoses groups having the higher fatal outcomes probability are aligned with the international literature. A web application is presented where researchers and providers can calculate predicted mortality based on the CgntSCORE on any web browser or smartphone.

External Validation of 3 Risk Scores in Adults with Congenital Heart Disease

Background and Objectives

Adults with congenital heart disease (CHD) are an increasing group of patients thanks to the survival of over 85% of children with CHD. 20% of these patients shall warrant a surgical procedure during their life span. However, currently there is no one risk score that assess correctly the mortality of these procedures. Thus, we analyse the risk scores used at our institution.

Methods

From May 1991 till June 2017, 608 procedures in adults with CHD were performed. The 3 risk scores (risk adjustment for congenital heart surgery [RACHS-1], Aristotle, and Euroscore I) of each procedure were analysed. We used area under the receiver operating characteristic curve (c-index) to measure model discrimination, and Hosmer-Lemeshow (H-L) statistic along with calibration plots to measure calibration.

Results

There was no statistical difference between the area under the curve for the 3 scores (χ²=0.58 with 2 df, p=0.750). There was no evidence of lack of fit for RACHS-1 (H-L, χ²=2.61; p=0.271) and Aristotle score (H-L, χ²=5.69; p=0.459). However, there was evidence in lack of calibration in the Euroscore I scoring system (H-L, χ²=33.69; p<0.001). The calibration slope for RACHS-1 was 0.912, for Aristotle (stratified in risk groups) was −0.14 and for Euroscore 1 (stratified in risk groups) was 0.46.

Conclusions

RACHS-1 seems to be best risk scoring system for calculating mortality applied to surgery in adults with CHD.

Validation of the grown-ups with congenital heart disease score

Objectives

Adults with congenital heart disease in need of heart surgery frequently present with significant comorbidity. Furthermore, additional technical difficulties often related to redo operations increase the risk for postoperative mortality and morbidity. Hence, next to the type of the procedure, additional procedure-dependent and procedure-independent factors have to be considered for risk evaluation. The recently proposed grown-ups with congenital heart disease (GUCH) mortality and morbidity scores account for these additional risk factors. We sought to validate their predictive power in a large population operated in a single centre.

Methods

Data of all consecutive patients aged 18 years or more, who underwent surgery for congenital heart disease between 2005 and 2016, were collected. Mortality was defined as hospital mortality or mortality within 30 days following surgery. Morbidity was defined as occurrence of one or more of the following complications: renal failure requiring dialysis, neurologic deficit persisting at discharge, atrioventricular block requiring permanent pacemaker implantation, mechanical circulatory support, phrenic nerve injury and unplanned reoperation. The discriminatory power of the GUCH scores was assessed using the area under the receiver operating characteristics curve (c-index, including 95% CI).

Results

Eight hundred and twenty-four operations were evaluated. Additional procedure-dependent and procedure-independent factors, as defined in the GUCH scores, were present in 165 patients (20.0%) and 544 patients (66.0%), respectively. Hospital mortality and morbidity was 3.4% and 10.0%, respectively. C-index for GUCH mortality score was 0.809 (0.742–0.877). C-index for GUCH morbidity score was 0.676 (0.619–0.734).

Conclusions

We could confirm the good predictive power of the GUCH mortality score for postoperative mortality in a large population of adults with congenital heart disease.

Management of the Critically Ill Adult With Congenital Heart Disease

Survival of adults with congenital heart disease (CHD) has improved significantly over the last 2 decades, leading to an increase in hospital and intensive care unit (ICU) admissions of these patients. Whereas most of the ICU admissions in the past were related to perioperative management, the incidence of medical emergencies from long-term sequelae of palliative or corrective surgical treatment of these patients is rising. Intensivists now are confronted with patients who not only have complex anatomy after congenital cardiac surgery, but also complex pathophysiology due to decades of living with abnormal cardiac anatomy and diseases of advanced age. Comorbidities affect all organ systems, including cognitive function, pulmonary and cardiovascular systems, liver, and kidneys. Critical care management requires an in-depth understanding of underlying anatomy and pathophysiology in order to apply contemporary concepts of adult ICU care to this population and optimize patient outcomes. In this review, the main CHD lesions and their common surgical management approaches are described, and the sequelae of CHD physiology are discussed. In addition, the effects of chronic comorbidities on the management of critically ill adults are explored, and the adjustments of current ICU management modalities and pharmacology to optimize care are discussed.

Contemporary cardiac surgery for adults with congenital heart disease

OBJECTIVE

Advances in early management of congenital heart disease (CHD) have led to an exponential growth in adults with CHD (ACHD). Many of these patients require cardiac surgery. This study sought to examine outcome and its predictors for ACHD cardiac surgery.

METHODS

This is an observational cohort study of prospectively collected data on 1090 consecutive adult patients with CHD, undergoing 1130 cardiac operations for CHD at the Royal Brompton Hospital between 2002 and 2011. Early mortality was the primary outcome measure. Midterm to longer-term survival, cumulative incidence of reoperation, other interventions and/or new-onset arrhythmia were secondary outcome measures. Predictors of early/total mortality were identified.

RESULTS

Age at surgery was 35±15 years, 53% male, 52.3% were in New York Heart Association (NYHA) class I, 37.2% in class II and 10.4% in class III/IV. Early mortality was 1.77% with independent predictors NYHA class ≥ III, tricuspid annular plane systolic excursion (TAPSE) <15 mm and female gender. Over a mean follow-up of 2.8±2.6 years, 46 patients died. Baseline predictors of total mortality were NYHA class ≥ III, TAPSE <15 mm and non-elective surgery. The number of sternotomies was not independently associated with neither early nor total mortality. At 10 years, probability of survival was 94%. NYHA class among survivors was significantly improved, compared with baseline.

CONCLUSIONS

Contemporary cardiac surgery for ACHD performed at a single, tertiary reference centre with a multidisciplinary approach is associated with low mortality and improved functional status. Also, our findings emphasise the point that surgery should not be delayed because of reluctance to reoperate only.