Application of machine learning algorithms for clinical predictive modeling: a data-mining approach in SCT

Novel approach exploring the correlation between presepsin and routine laboratory parameters using explainable artificial intelligence

Although presepsin, a crucial biomarker for the diagnosis and management of sepsis, has gained prominence in contemporary medical research, its relationship with routine laboratory parameters, including demographic data and hospital blood test data, remains underexplored. This study integrates machine learning with explainable artificial intelligence (XAI) to provide insights into the relationship between presepsin and these parameters. Advanced machine learning classifiers provide a multilateral view of data and play an important role in highlighting the interrelationships between presepsin and other parameters. XAI enhances analysis by ensuring transparency in the model's decisions, especially in selecting key parameters that significantly enhance classification accuracy. Utilizing XAI, this study successfully identified critical parameters that increased the predictive accuracy for sepsis patients, achieving a remarkable ROC AUC of 0.97 and an accuracy of 0.94. This breakthrough is possibly attributed to the comprehensive utilization of XAI in refining parameter selection, thus leading to these significant predictive metrics. The presence of missing data in datasets is another concern; this study addresses it by employing Extreme Gradient Boosting (XGBoost) to manage missing data, effectively mitigating potential biases while preserving both the accuracy and relevance of the results. The perspective of examining data from higher dimensions using machine learning transcends traditional observation and analysis. The findings of this study hold the potential to enhance patient diagnoses and treatment, underscoring the value of merging traditional research methods with advanced analytical tools.

Development and evaluation of machine learning models for predicting large-for-gestational-age newborns in women exposed to radiation prior to pregnancy

INTRODUCTION

The correlation between radiation exposure before pregnancy and abnormal birth weight has been previously proven. However, for large-for-gestational-age (LGA) babies in women exposed to radiation before becoming pregnant, there is no prediction model yet.

MATERIAL AND METHODS

The data were collected from the National Free Preconception Health Examination Project in China. A sum of 455 neonates (42 SGA births and 423 non-LGA births) were included. A training set (n = 319) and a test set (n = 136) were created from the dataset at random. To develop prediction models for LGA neonates, conventional logistic regression (LR) method and six machine learning methods were used in this study. Recursive feature elimination approach was performed by choosing 10 features which made a big contribution to the prediction models. And the Shapley Additive Explanation model was applied to interpret the most important characteristics that affected forecast outputs.

RESULTS

The random forest (RF) model had the highest average area under the receiver-operating-characteristic curve (AUC) for predicting LGA in the test set (0.843, 95% confidence interval [CI]: 0.714-0.974). Except for the logistic regression model (AUC: 0.603, 95%CI: 0.440-0.767), other models' AUCs displayed well. Thereinto, the RF algorithm's final prediction model using 10 characteristics achieved an average AUC of 0.821 (95% CI: 0.693-0.949).

CONCLUSION

The prediction model based on machine learning might be a promising tool for the prenatal prediction of LGA births in women with radiation exposure before pregnancy.

Prediction of longitudinal clinical outcomes after acute myocardial infarction using a dynamic machine learning algorithm

Several regression-based models for predicting outcomes after acute myocardial infarction (AMI) have been developed. However, prediction models that encompass diverse patient-related factors over time are limited. This study aimed to develop a machine learning-based model to predict longitudinal outcomes after AMI. This study was based on a nationwide prospective registry of AMI in Korea (n = 13,104). Seventy-seven predictor candidates from prehospitalization to 1 year of follow-up were included, and six machine learning approaches were analyzed. Primary outcome was defined as 1-year all-cause death. Secondary outcomes included all-cause deaths, cardiovascular deaths, and major adverse cardiovascular event (MACE) at the 1-year and 3-year follow-ups. Random forest resulted best performance in predicting the primary outcome, exhibiting a 99.6% accuracy along with an area under the receiver-operating characteristic curve of 0.874. Top 10 predictors for the primary outcome included peak troponin-I (variable importance value = 0.048), in-hospital duration (0.047), total cholesterol (0.047), maintenance of antiplatelet at 1 year (0.045), coronary lesion classification (0.043), N-terminal pro-brain natriuretic peptide levels (0.039), body mass index (BMI) (0.037), door-to-balloon time (0.035), vascular approach (0.033), and use of glycoprotein IIb/IIIa inhibitor (0.032). Notably, BMI was identified as one of the most important predictors of major outcomes after AMI. BMI revealed distinct effects on each outcome, highlighting a U-shaped influence on 1-year and 3-year MACE and 3-year all-cause death. Diverse time-dependent variables from prehospitalization to the postdischarge period influenced the major outcomes after AMI. Understanding the complexity and dynamic associations of risk factors may facilitate clinical interventions in patients with AMI.

Analyzing predictors of in-hospital mortality in patients with acute ST-segment elevation myocardial infarction using an evolved machine learning approach

Acute ST-segment elevation myocardial infarction (STEMI) is a severe cardiac ailment characterized by the sudden complete blockage of a portion of the coronary artery, leading to the interruption of blood supply to the myocardium. This study examines the medical records of 3205 STEMI patients admitted to the coronary care unit of the First Affiliated Hospital of Wenzhou Medical University from January 2014 to December 2021. In this research, a novel predictive framework for STEMI is proposed, incorporating evolutionary computational methods and machine learning techniques. A variant algorithm, AGCOSCA, is introduced by integrating crossover operation and observation bee strategy into the original Sine Cosine Algorithm (SCA). The effectiveness of AGCOSCA is initially validated using IEEE CEC 2017 benchmark functions, demonstrating its ability to mitigate the deficiency in local mining after SCA random perturbation. Building upon this foundation, the AGCOSCA approach has been paired with Support Vector Machine (SVM) to forge the predictive framework referred to as AGCOSCA-SVM. Specifically, AGCOSCA is employed to refine the selection of predictors from a substantial feature set before SVM is utilized to forecast the occurrence of STEMI. In our analysis, we observed that SVM excels at managing nonlinear data relationships, a strength that becomes particularly prominent in smaller datasets of STEMI patients. To assess the effectiveness of AGCOSCA-SVM, diagnostic experiments were conducted based on the STEMI sample data. Results indicate that AGCOSCA-SVM outperforms traditional machine learning methods, achieving superior Accuracy, Sensitivity, and Specificity values of 97.83 %, 93.75 %, and 96.67 %, respectively. The selected features, such as acute kidney injury (AKI) stage, fibrinogen, mean platelet volume (MPV), free triiodothyronine (FT3), diuretics, and Killip class during hospitalization, are identified as crucial for predicting STEMI. In conclusion, AGCOSCA-SVM emerges as a promising model framework for supporting the diagnostic process of STEMI, showcasing potential applications in clinical settings.

An artificial intelligence-driven predictive model for pediatric allogeneic hematopoietic stem cell transplantation using clinical variables

BACKGROUND

Hematopoietic stem cell transplantation (HSCT) is a procedure with high morbidity and mortality. Identifying patients for maximum benefit and risk assessment is crucial in the decision-making process. This has led to the development of predictive risk models for HSCT in adults, which have limitations when applied to pediatric population. Our goal was to develop an automatic learning algorithm to predict survival in children with malignant disorders undergoing HSCT.

METHODS

We studied allogenic HSCTs performed on children with malignant disorders at a third-level hospital between 1991 and 2021. Survival was analyzed using the Kaplan-Meier method, log-rank test for the univariate analysis, and Cox regression for the multivariate analysis. A prognostic index was constructed based on these findings. Lastly, we constructed a predictive model using a random forest algorithm to forecast 1-year survival after HSCT.

RESULTS

We analyzed 229 HSCTs in 201 patients with a median follow-up of 1.64 years. Variables that impacted on the multivariate analysis were older age (hazard ratio [HR] 1.40, 95% confidence interval [CI] 1.12-1.76, p = .003), oldest period of HSCT (HR 0.46, 95% CI 0.29-0.73, p < .001), and mismatched donor (HR 2.65, 95% CI 1.51-4.65, p = .001). Our prognostic index was associated with 3-year overall survival (OS; p < .001). A random forest was developed using as variables: diagnosis, age, year of HSCT, time from diagnosis to HSCT, disease stage, donor type, and conditioning. This achieved 72% accuracy in predicting 1-year OS.

CONCLUSIONS

Our index and random forest was effective in predicting 1-year survival. However, further validation in diverse populations is necessary to establish their generalizability.

Predicting pressure gradient using artificial intelligence for transcatheter aortic valve replacement

Objective

After transcatheter aortic valve replacement, the mean transvalvular pressure gradient indicates the effectiveness of the therapy. The objective is to develop artificial intelligence to predict the post-transcatheter aortic valve replacement aortic valve pressure gradient and aortic valve area from preprocedural echocardiography and computed tomography data.

Methods

A retrospective study was conducted on patients who underwent transcatheter aortic valve replacement due to aortic valve stenosis. A total of 1091 patients were analyzed for pressure gradient predictions (mean age 76.8 ± 9.2 years, 57.8% male), and 1063 patients were analyzed for aortic valve area predictions (mean age 76.7 ± 9.3 years, 57.2% male). An artificial intelligence learning model was trained (training: n = 663 patients, validation: n = 206 patients) and tested (testing: n = 222 patients) to predict pressure gradient, and a separate artificial intelligence learning model was trained (training: n = 640 patients, validation: n = 218 patients) and tested (testing: n = 205 patients) for predicting aortic valve area.

Results

The mean absolute error for pressure gradient and aortic valve area predictions was 3.0 mm Hg and 0.45 cm2, respectively. Valve sheath size, body surface area, and age were determined to be the top 3 predictors for pressure gradient, and valve sheath size, left ventricular ejection fraction, and aortic annulus mean diameter were identified to be the top 3 predictors of post-transcatheter aortic valve replacement aortic valve area. A training dataset size of more than 500 patients demonstrated good robustness of the artificial intelligence models for pressure gradient and aortic valve area.

Conclusions

The artificial intelligence-based algorithm has demonstrated potential in predicting post-transcatheter aortic valve replacement transvalvular pressure gradient predictions for patients with aortic valve stenosis. Further studies are necessary to differentiate pressure gradient between valve types.

The predictive value of machine learning for mortality risk in patients with acute coronary syndromes: a systematic review and meta-analysis

BACKGROUND

Acute coronary syndromes (ACS) are the leading cause of global death. Optimizing mortality risk prediction and early identification of high-risk patients is essential for developing targeted prevention strategies. Many researchers have built machine learning (ML) models to predict the mortality risk in ACS patients. Our meta-analysis aimed to evaluate the predictive value of various ML models in predicting death in ACS patients at different times.

METHODS

PubMed, Embase, Web of Science, and Cochrane Library were searched systematically from database establishment to March 12, 2022 for studies developing or validating at least one ML predictive model for death in ACS patients. We used PROBAST to assess the risk of bias in the reported predictive models and a random-effects model to assess the pooled C-index and accuracy of these models.

RESULTS

Fifty papers were included, involving 216 ML prediction models, 119 of which were externally validated. The combined C-index of the ML models in the validation cohort predicting the in-hospital mortality, 30-day mortality, 3- or 6-month mortality, and 1 year or above mortality in ACS patients were 0.8633 (95% CI 0.8467-0.8802), 0.8296 (95% CI 0.8134-0.8462), 0.8205 (95% CI 0.7881-0.8541), and 0.8197 (95% CI 0.8042-0.8354), respectively, with the corresponding combined accuracy of 0.8569 (95% CI 0.8411-0.8715), 0.8282 (95% CI 0.7922-0.8591), 0.7303 (95% CI 0.7184-0.7418), and 0.7837 (95% CI 0.7455-0.8175), indicating that the ML models were relatively excellent in predicting ACS mortality at different times. Furthermore, common predictors of death in ML models included age, sex, systolic blood pressure, serum creatinine, Killip class, heart rate, diastolic blood pressure, blood glucose, and hemoglobin.

CONCLUSIONS

The ML models had excellent predictive power for mortality in ACS, and the methodologies may need to be addressed before they can be used in clinical practice.

Machine Learning for the Prediction of Survival Post-Allogeneic Hematopoietic Cell Transplantation: A Single-Center Experience

INTRODUCTION

Prediction of outcomes following allogeneic hematopoietic cell transplantation (HCT) remains a major challenge. Machine learning (ML) is a computational procedure that may facilitate the generation of HCT prediction models. We sought to investigate the prognostic potential of multiple ML algorithms when applied to a large single-center allogeneic HCT database.

METHODS

Our registry included 2,697 patients that underwent allogeneic HCT from January 1976 to December 2017. 45 pretransplant baseline variables were included in the predictive assessment of each ML algorithm on overall survival (OS) as determined by area under the curve (AUC). Pretransplant variables used in the EBMT ML study (Shouval et al., 2015) were used as a benchmark for comparison.

RESULTS

On the entire dataset, the random forest (RF) algorithm performed best (AUC 0.71 ± 0.04) compared to the second-best model, logistic regression (LR) (AUC = 0.69 ± 0.04) (p < 0.001). Both algorithms demonstrated improved AUC scores using all 45 variables compared to the limited variables examined by the EBMT study. Survival at 100 days post-HCT using RF on the full dataset discriminated patients into different prognostic groups with different 2-year OS (p < 0.0001). We then examined the ML methods that allow for significant individual variable identification, including LR and RF, and identified matched related donors (HR = 0.49, p < 0.0001), increasing TBI dose (HR = 1.60, p = 0.006), increasing recipient age (HR = 1.92, p < 0.0001), higher baseline Hb (HR = 0.59, p = 0.0002), and increased baseline FEV1 (HR = 0.73, p = 0.02), among others.

CONCLUSION

The application of multiple ML techniques on single-center allogeneic HCT databases warrants further investigation and may provide a useful tool to identify variables with prognostic potential.

Unsupervised meta-clustering identifies risk clusters in acute myeloid leukemia based on clinical and genetic profiles

BACKGROUND

Increasingly large and complex biomedical data sets challenge conventional hypothesis-driven analytical approaches, however, data-driven unsupervised learning can detect inherent patterns in such data sets.

METHODS

While unsupervised analysis in the medical literature commonly only utilizes a single clustering algorithm for a given data set, we developed a large-scale model with 605 different combinations of target dimensionalities as well as transformation and clustering algorithms and subsequent meta-clustering of individual results. With this model, we investigated a large cohort of 1383 patients from 59 centers in Germany with newly diagnosed acute myeloid leukemia for whom 212 clinical, laboratory, cytogenetic and molecular genetic parameters were available.

RESULTS

Unsupervised learning identifies four distinct patient clusters, and statistical analysis shows significant differences in rate of complete remissions, event-free, relapse-free and overall survival between the four clusters. In comparison to the standard-of-care hypothesis-driven European Leukemia Net (ELN2017) risk stratification model, we find all three ELN2017 risk categories being represented in all four clusters in varying proportions indicating unappreciated complexity of AML biology in current established risk stratification models. Further, by using assigned clusters as labels we subsequently train a supervised model to validate cluster assignments on a large external multicenter cohort of 664 intensively treated AML patients.

CONCLUSIONS

Dynamic data-driven models are likely more suitable for risk stratification in the context of increasingly complex medical data than rigid hypothesis-driven models to allow for a more personalized treatment allocation and gain novel insights into disease biology.

Application of supervised machine learning algorithms for classification and prediction of type-2 diabetes disease status in Afar regional state, Northeastern Ethiopia 2021

Ethiopia has been challenged by the growing magnitude of diabetes in general and type-2 diabetes in particular. Knowledge extraction from stored dataset can be an important base for better decision on diabetes rapid diagnosis, suggestive on prediction for early intervention. Thus, this study was addressed these problem by application of supervised machine learning algorithms for classification and prediction of type 2 diabetes disease status and might provide context-specific information to program planners and policy makers so that, priority will be given to the more affected groups. To apply supervised machine learning algorithms; compare these algorithms and select the best algorithm based on their performance for classification and prediction of type-2 diabetic disease status (positive or negative) in public hospitals of Afar regional state, Northeastern Ethiopia. This study was conducted at Afar regional state from February to June of 2021. Decision tree; pruned J 48, Artificial neural network, K-nearest neighbor, Support vector machine, Binary logistic regression, Random forest, and Naïve Bayes supervised machine learning algorithms were applied using secondary data from the medical database record review. A total of 2239 sample Dataset diagnosed for diabetes from 2012 to April 22/2020 (1523 with type-2 diabetes and 716 without type-2 diabetes) was checked for its completeness prior to analysis. For all algorithms, WEKA3.7 tool was used for analysis purposes. Moreover, all algorithms were compared based on their correctly classification rate, kappa statistics, confusion matrix, area under the curve, sensitivity, and specificity. From the seven major supervised machine learning algorithms, the best classification and prediction results were obtained from random forest [correctly classified rate (93.8%), kappa statistics (0.85), sensitivity (0.98), area under the curve (0.97) and confusion matrix (out of 454 actual positive prediction for 446)] which was followed by decision tree pruned J 48 [correctly classified rate (91.8%), kappa statistics (0.80), sensitivity (0.96), area under the curve (0.91) and confusion matrices (out of 454 actual positive prediction for 438)] and k-nearest neighbor [correctly classified rate (89.8%), kappa statistics (0.76), sensitivity (0.92), area under the curve (0.88) and confusion matrices (out of 454 actual positive prediction for 421)]. Random forest, Decision tree pruned J48 and k-nearest neighbor algorithms have better classification and prediction performance for classifying and predicting type-2 diabetes disease status. Therefore, based on this performance, random forest algorithm can be judged as suggestive and supportive for clinicians at the time of type-2 diabetes diagnosis.

Prediction of permanent pacemaker implantation after transcatheter aortic valve replacement: The role of machine learning

BACKGROUND

Atrioventricular block requiring permanent pacemaker (PPM) implantation is an important complication of transcatheter aortic valve replacement (TAVR). Application of machine learning could potentially be used to predict pre-procedural risk for PPM.

AIM

To apply machine learning to be used to predict pre-procedural risk for PPM.

METHODS

A retrospective study of 1200 patients who underwent TAVR (January 2014-December 2017) was performed. 964 patients without prior PPM were included for a 30-d analysis and 657 patients without PPM requirement through 30 d were included for a 1-year analysis. After the exclusion of variables with near-zero variance or ≥ 50% missing data, 167 variables were included in the random forest gradient boosting algorithm (GBM) optimized using 5-fold cross-validations repeated 10 times. The receiver operator curve (ROC) for the GBM model and PPM risk score models were calculated to predict the risk of PPM at 30 d and 1 year.

RESULTS

Of 964 patients included in the 30-d analysis without prior PPM, 19.6% required PPM post-TAVR. The mean age of patients was 80.9 ± 8.7 years. 42.1 % were female. Of 657 patients included in the 1-year analysis, the mean age of the patients was 80.7 ± 8.2. Of those, 42.6% of patients were female and 26.7% required PPM at 1-year post-TAVR. The area under ROC to predict 30-d and 1-year risk of PPM for the GBM model (0.66 and 0.72) was superior to that of the PPM risk score (0.55 and 0.54) with a P value < 0.001.

CONCLUSION

The GBM model has good discrimination and calibration in identifying patients at high risk of PPM post-TAVR.

Predictive models in extracorporeal membrane oxygenation (ECMO): a systematic review

PURPOSE

Extracorporeal membrane oxygenation (ECMO) has been increasingly used in the last years to provide hemodynamic and respiratory support in critically ill patients. In this scenario, prognostic scores remain essential to choose which patients should initiate ECMO. This systematic review aims to assess the current landscape and inform subsequent efforts in the development of risk prediction tools for ECMO.

METHODS

PubMed, CINAHL, Embase, MEDLINE and Scopus were consulted. Articles between Jan 2011 and Feb 2022, including adults undergoing ECMO reporting a newly developed and validated predictive model for mortality, were included. Studies based on animal models, systematic reviews, case reports and conference abstracts were excluded. Data extraction aimed to capture study characteristics, risk model characteristics and model performance. The risk of bias was evaluated through the prediction model risk-of-bias assessment tool (PROBAST). The protocol has been registered in Open Science Framework ( https://osf.io/fevw5 ).

RESULTS

Twenty-six prognostic scores for in-hospital mortality were identified, with a study size ranging from 60 to 4557 patients. The most common candidate variables were age, lactate concentration, creatinine concentration, bilirubin concentration and days in mechanical ventilation prior to ECMO. Five out of 16 venous-arterial (VA)-ECMO scores and 3 out of 9 veno-venous (VV)-ECMO scores had been validated externally. Additionally, one score was developed for both VA and VV populations. No score was judged at low risk of bias.

CONCLUSION

Most models have not been validated externally and apply after ECMO initiation; thus, some uncertainty whether ECMO should be initiated still remains. It has yet to be determined whether and to what extent a new methodological perspective may enhance the performance of predictive models for ECMO, with the ultimate goal to implement a model that positively influences patient outcomes.

Machine learning-based predictive risk models for 30-day and 1-year mortality in severe aortic stenosis patients undergoing transcatheter aortic valve implantation

BACKGROUND

Predictive risk score for mortality plays an important role in the decision-making in patient selection and risk stratification for TAVI. Existing established predictive risk scores had poor discrimination performance in the prediction of mortality after the TAVI.

OBJECTIVES

The present study aimed to develop machine learning-based predictive models for 30-day and 1-year mortality in severe aortic stenosis patients undergoing TAVI.

METHODS

A total of 186 patients in a retrospective cohort study were analyzed. The models were fitted by a decision tree. Each model was tested in 100 iterations of 80:20 stratified random splitting into training/testing samples and 10-fold cross-validation.

RESULTS

Variables that predict 30-day mortality are a set of factors driven mainly by height, chronic lung disease, STS score, preoperative LVEF, age, and preoperative LVOT VTI. Variables that predict 1-year mortality are a set of factors consisting of preoperative LVEF, STS score, heart rate, systolic blood pressure, home oxygen use, serum creatinine level, and preoperative LVOT Vmax. This decision tree-generated predictive models for 30-day and 1- year mortality provided the most precise accuracy of 0.97 and 0.90 with the AUC-ROC curves of 0.83 and 0.71 on 30-day and 1-year mortality on testing data and had better discrimination performance compared to the existing established TAVI predictive risk scores.

CONCLUSIONS

These machine learning models show excellent accuracy and have a better prediction for 30-day and 1-year mortality than the existing established TAVI predictive risk scores. A customized predictive model deems to be properly developed for better risk discrimination among cohorts.

Big Data in Gastroenterology Research

Studying individual data types in isolation provides only limited and incomplete answers to complex biological questions and particularly falls short in revealing sufficient mechanistic and kinetic details. In contrast, multi-omics approaches to studying health and disease permit the generation and integration of multiple data types on a much larger scale, offering a comprehensive picture of biological and disease processes. Gastroenterology and hepatobiliary research are particularly well-suited to such analyses, given the unique position of the luminal gastrointestinal (GI) tract at the nexus between the gut (mucosa and luminal contents), brain, immune and endocrine systems, and GI microbiome. The generation of 'big data' from multi-omic, multi-site studies can enhance investigations into the connections between these organ systems and organisms and more broadly and accurately appraise the effects of dietary, pharmacological, and other therapeutic interventions. In this review, we describe a variety of useful omics approaches and how they can be integrated to provide a holistic depiction of the human and microbial genetic and proteomic changes underlying physiological and pathophysiological phenomena. We highlight the potential pitfalls and alternatives to help avoid the common errors in study design, execution, and analysis. We focus on the application, integration, and analysis of big data in gastroenterology and hepatobiliary research.

Use of Machine Learning to Develop Prediction Models for Mortality and Stroke in Patients Undergoing Balloon Aortic Valvuloplasty

OBJECTIVE

To develop an artificial intelligence, machine learning prediction model for estimating in-hospital mortality and stroke in patients undergoing balloon aortic valvuloplasty (BAV).

METHODS

The National Inpatient Sample (NIS) database was used to identify patients who underwent BAV from 2005 to 2017. Outcomes analyzed were in-hospital all-cause mortality and stroke after BAV. Predictors of mortality and stroke were selected using LASSO regularization. A conventional logistic regression and a random forest machine learning algorithm were used to train the models for predicting outcomes. The performance of all the modeling algorithms for predicting in-hospital mortality and stroke was compared between models using c-statistic, F1 score, brier score loss, diagnostic accuracy, and Kolmogorov-Smirnov plots.

RESULTS

A total of 6962 patients with severe aortic stenosis who underwent BAV were identified. The performance of random forest classifier was comparable with logistic regression for predicting in-hospital mortality for all measures of performance (F1 score 0.422 vs 0.409, ROC-AUC 0.822 [95 % CI 0.787-0.855] vs 0.815 [95 % CI 0.779-0.849], diagnostic accuracy 70.42 % vs 70.93 %, KS-statistic 0.513 vs 0.494 and brier score loss 0.295 vs 0.291). The random forest algorithm significantly outperformed logistic regression in predicting in-hospital stroke with respect to all performance metrics: F1 score 0.225 vs 0.095, AUC 0.767 [0.662-0.858] vs 0.637 [0.499-0.754], brier score loss [0.399 vs 0.407], and KS-statistic [0.465 vs 0.254].

CONCLUSIONS

The good discrimination of machine learning models reveal the potential of artificial intelligence to improve patient risk stratification for BAV.

Survival Prediction of Children Undergoing Hematopoietic Stem Cell Transplantation Using Different Machine Learning Classifiers by Performing Chi-Square Test and Hyperparameter Optimization: A Retrospective Analysis

Bone marrow transplant (BMT) is an effective surgical treatment for bone marrow-related disorders. However, several associated risk factors can impair long-term survival after BMT. Machine learning (ML) technologies have been proven useful in survival prediction of BMT receivers along with the influences that limit their resilience. In this study, an efficient classification model predicting the survival of children undergoing BMT is presented using a public dataset. Several supervised ML methods were investigated in this regard with an 80-20 train-test split ratio. To ensure prediction with minimal time and resources, only the top 11 out of the 59 dataset features were considered using Chi-square feature selection method. Furthermore, hyperparameter optimization (HPO) using the grid search cross-validation (GSCV) technique was adopted to increase the accuracy of prediction. Four experiments were conducted utilizing a combination of default and optimized hyperparameters on the original and reduced datasets. Our investigation revealed that the top 11 features of HPO had the same prediction accuracy (94.73%) as the entire dataset with default parameters, however, requiring minimal time and resources. Hence, the proposed approach may aid in the development of a computer-aided diagnostic system with satisfactory accuracy and minimal computation time by utilizing medical data records.

Diagnosis and management of AML in adults: 2022 recommendations from an international expert panel on behalf of the ELN

The 2010 and 2017 editions of the European LeukemiaNet (ELN) recommendations for diagnosis and management of acute myeloid leukemia (AML) in adults are widely recognized among physicians and investigators. There have been major advances in our understanding of AML, including new knowledge about the molecular pathogenesis of AML, leading to an update of the disease classification, technological progress in genomic diagnostics and assessment of measurable residual disease, and the successful development of new therapeutic agents, such as FLT3, IDH1, IDH2, and BCL2 inhibitors. These advances have prompted this update that includes a revised ELN genetic risk classification, revised response criteria, and treatment recommendations.

A SuperLearner Approach to Predict Run-In Selection in Clinical Trials

A critical early step in a clinical trial is defining the study sample that appropriately represents the target population from which the sample will be drawn. Envisaging a “run-in” process in study design may accomplish this task; however, the traditional run-in requires additional patients, increasing times, and costs. The possible use of the available a-priori data could skip the run-in period. In this regard, ML (machine learning) techniques, which have recently shown considerable promising usage in clinical research, can be used to construct individual predictions of therapy response probability conditional on patient characteristics. An ensemble model of ML techniques was trained and validated on twin randomized clinical trials to mimic a run-in process within this framework. An ensemble ML model composed of 26 algorithms was trained on the twin clinical trials. SuperLearner (SL) performance for the Verum (Treatment) arm is above 70% sensitivity. The Positive Predictive Value (PPP) achieves a value of 80%. Results show good performance in the direction of being useful in the simulation of the run-in period; the trials conducted in similar settings can train an optimal patient selection algorithm minimizing the run-in time and costs of conduction.

Predictive models for small-for-gestational-age births in women exposed to pesticides before pregnancy based on multiple machine learning algorithms

Background

The association between prenatal pesticide exposures and a higher incidence of small-for-gestational-age (SGA) births has been reported. No prediction model has been developed for SGA neonates in pregnant women exposed to pesticides prior to pregnancy.

Methods

A retrospective cohort study was conducted using information from the National Free Preconception Health Examination Project between 2010 and 2012. A development set (n = 606) and a validation set (n = 151) of the dataset were split at random. Traditional logistic regression (LR) method and six machine learning classifiers were used to develop prediction models for SGA neonates. The Shapley Additive Explanation (SHAP) model was applied to determine the most influential variables that contributed to the outcome of the prediction.

Results

757 neonates in total were analyzed. SGA occurred in 12.9% (n = 98) of cases overall. With an area under the receiver-operating-characteristic curve (AUC) of 0.855 [95% confidence interval (CI): 0.752–0.959], the model based on category boosting (CatBoost) algorithm obtained the best performance in the validation set. With the exception of the LR model (AUC: 0.691, 95% CI: 0.554–0.828), all models had good AUCs. Using recursive feature elimination (RFE) approach to perform the feature selection, we included 15 variables in the final model based on CatBoost classifier, achieving the AUC of 0.811 (95% CI: 0.675–0.947).

Conclusions

Machine learning algorithms can develop satisfactory tools for SGA prediction in mothers exposed to pesticides prior to pregnancy, which might become a tool to predict SGA neonates in the high-risk population.

Predicting perinatal health outcomes using smartphone-based digital phenotyping and machine learning in a prospective Swedish cohort (Mom2B): study protocol

INTRODUCTION

Perinatal complications, such as perinatal depression and preterm birth, are major causes of morbidity and mortality for the mother and the child. Prediction of high risk can allow for early delivery of existing interventions for prevention. This ongoing study aims to use digital phenotyping data from the Mom2B smartphone application to develop models to predict women at high risk for mental and somatic complications.

METHODS AND ANALYSIS

All Swedish-speaking women over 18 years, who are either pregnant or within 3 months postpartum are eligible to participate by downloading the Mom2B smartphone app. We aim to recruit at least 5000 participants with completed outcome measures. Throughout the pregnancy and within the first year postpartum, both active and passive data are collected via the app in an effort to establish a participant's digital phenotype. Active data collection consists of surveys related to participant background information, mental and physical health, lifestyle, and social circumstances, as well as voice recordings. Participants' general smartphone activity, geographical movement patterns, social media activity and cognitive patterns can be estimated through passive data collection from smartphone sensors and activity logs. The outcomes will be measured using surveys, such as the Edinburgh Postnatal Depression Scale, and through linkage to national registers, from where information on registered clinical diagnoses and received care, including prescribed medication, can be obtained. Advanced machine learning and deep learning techniques will be applied to these multimodal data in order to develop accurate algorithms for the prediction of perinatal depression and preterm birth. In this way, earlier intervention may be possible.

ETHICS AND DISSEMINATION

Ethical approval has been obtained from the Swedish Ethical Review Authority (dnr: 2019/01170, with amendments), and the project fully fulfils the General Data Protection Regulation (GDPR) requirements. All participants provide consent to participate and can withdraw their participation at any time. Results from this project will be disseminated in international peer-reviewed journals and presented in relevant conferences.

Development and Evaluation of a Machine Learning Prediction Model for Small-for-Gestational-Age Births in Women Exposed to Radiation before Pregnancy

Exposure to radiation has been associated with increased risk of delivering small-for-gestational-age (SGA) newborns. There are no tools to predict SGA newborns in pregnant women exposed to radiation before pregnancy. Here, we aimed to develop an array of machine learning (ML) models to predict SGA newborns in women exposed to radiation before pregnancy. Patients’ data was obtained from the National Free Preconception Health Examination Project from 2010 to 2012. The data were randomly divided into a training dataset (n = 364) and a testing dataset (n = 91). Eight various ML models were compared for solving the binary classification of SGA prediction, followed by a post hoc explainability based on the SHAP model to identify and interpret the most important features that contribute to the prediction outcome. A total of 455 newborns were included, with the occurrence of 60 SGA births (13.2%). Overall, the model obtained by extreme gradient boosting (XGBoost) achieved the highest area under the receiver-operating-characteristic curve (AUC) in the testing set (0.844, 95% confidence interval (CI): 0.713–0.974). All models showed satisfied AUCs, except for the logistic regression model (AUC: 0.561, 95% CI: 0.355–0.768). After feature selection by recursive feature elimination (RFE), 15 features were included in the final prediction model using the XGBoost algorithm, with an AUC of 0.821 (95% CI: 0.650–0.993). ML algorithms can generate robust models to predict SGA newborns in pregnant women exposed to radiation before pregnancy, which may thus be used as a prediction tool for SGA newborns in high-risk pregnant women.

Predicting Long-term Survival After Allogeneic Hematopoietic Cell Transplantation in Patients With Hematologic Malignancies: Machine Learning-Based Model Development and Validation

Background

Scoring systems developed for predicting survival after allogeneic hematopoietic cell transplantation (HCT) show suboptimal prediction power, and various factors affect posttransplantation outcomes.

Objective

A prediction model using a machine learning–based algorithm can be an alternative for concurrently applying multiple variables and can reduce potential biases. In this regard, the aim of this study is to establish and validate a machine learning–based predictive model for survival after allogeneic HCT in patients with hematologic malignancies.

Methods

Data from 1470 patients with hematologic malignancies who underwent allogeneic HCT between December 1993 and June 2020 at Asan Medical Center, Seoul, South Korea, were retrospectively analyzed. Using the gradient boosting machine algorithm, we evaluated a model predicting the 5-year posttransplantation survival through 10-fold cross-validation.

Results

The prediction model showed good performance with a mean area under the receiver operating characteristic curve of 0.788 (SD 0.03). Furthermore, we developed a risk score predicting probabilities of posttransplantation survival in 294 randomly selected patients, and an agreement between the estimated predicted and observed risks of overall death, nonrelapse mortality, and relapse incidence was observed according to the risk score. Additionally, the calculated score demonstrated the possibility of predicting survival according to the different transplantation-related factors, with the visualization of the importance of each variable.

Conclusions

We developed a machine learning–based model for predicting long-term survival after allogeneic HCT in patients with hematologic malignancies. Our model provides a method for making decisions regarding patient and donor candidates or selecting transplantation-related resources, such as conditioning regimens.

A systematic review of the predicted outcomes related to hematopoietic stem cell transplantation: focus on applied machine learning methods' performance

INTRODUCTION

: Hematopoietic stem cell transplantation (HSCT) is a critical therapeutic procedure in blood diseases, and the investigation of HSCT data can provide valuable information. Machine learning (ML) techniques are novel and useful data analysis tools that have been applied in many studies to predict HSCT survival and estimate the risk of transplantation.

AREAS COVERED

: A systematic review was performed with a search of PubMed, Science Direct, Embase, Scopus, and the European Society for Blood and Marrow Transplantation, the Center for International Blood and Marrow Transplant Research, and the American Society for Transplantation and Cellular Therapy publications for articles published by September 2020.

EXPERT OPINION

: After investigating the results, 24 papers that met eligibility criteria were included in this study. The applied ML algorithms with the highest performance were Random Survival Forests (AUC=0.72) for survival-related, Random Survival Forests and Logistic Regression (AUC=0.77) for mortality-related, Deep Learning (AUC=0.8) for relapse, L2-Regularized Logistic Regression (AUC=0.66) for Acute-Graft Versus Host Disease, Random Survival Forests (AUC=0.88) for sepsis, Elastic-Net Regression (AUC=0.89) for cognitive impairment, and Bayesian Network (AUC=0.997) for oral mucositis outcome. This review reveals the potential of ML techniques to predict HSCT outcomes and apply them to developing clinical decision support systems.

A Clinical Prognostic Model Based on Machine Learning from the Fondazione Italiana Linfomi (FIL) MCL0208 Phase III Trial

BACKGROUND

Multicenter clinical trials are producing growing amounts of clinical data. Machine Learning (ML) might facilitate the discovery of novel tools for prognostication and disease-stratification. Taking advantage of a systematic collection of multiple variables, we developed a model derived from data collected on 300 patients with mantle cell lymphoma (MCL) from the Fondazione Italiana Linfomi-MCL0208 phase III trial (NCT02354313).

METHODS

We developed a score with a clustering algorithm applied to clinical variables. The candidate score was correlated to overall survival (OS) and validated in two independent data series from the European MCL Network (NCT00209222, NCT00209209); Results: Three groups of patients were significantly discriminated: Low, Intermediate (Int), and High risk (High). Seven discriminants were identified by a feature reduction approach: albumin, Ki-67, lactate dehydrogenase, lymphocytes, platelets, bone marrow infiltration, and B-symptoms. Accordingly, patients in the Int and High groups had shorter OS rates than those in the Low and Int groups, respectively (Int→Low, HR: 3.1, 95% CI: 1.0-9.6; High→Int, HR: 2.3, 95% CI: 1.5-4.7). Based on the 7 markers, we defined the engineered MCL international prognostic index (eMIPI), which was validated and confirmed in two independent cohorts; Conclusions: We developed and validated a ML-based prognostic model for MCL. Even when currently limited to baseline predictors, our approach has high scalability potential.

Establishment of a Predictive Model for GvHD-free, Relapse-free Survival after Allogeneic HSCT using Ensemble Learning

Stacked ensemble of machine-learning algorithms could establish more accurate prediction model for survival analysis than existing methods.

Stacked ensemble model can be applied to personalized prediction of HSCT outcomes from pretransplant characteristics.

Graft-versus-host disease-free, relapse-free survival (GRFS) is a useful composite end point that measures survival without relapse or significant morbidity after allogeneic hematopoietic stem cell transplantation (allo-HSCT). We aimed to develop a novel analytical method that appropriately handles right-censored data and competing risks to understand the risk for GRFS and each component of GRFS. This study was a retrospective data-mining study on a cohort of 2207 adult patients who underwent their first allo-HSCT within the Kyoto Stem Cell Transplantation Group, a multi-institutional joint research group of 17 transplantation centers in Japan. The primary end point was GRFS. A stacked ensemble of Cox Proportional Hazard (Cox-PH) regression and 7 machine-learning algorithms was applied to develop a prediction model. The median age for the patients was 48 years. For GRFS, the stacked ensemble model achieved better predictive accuracy evaluated by C-index than other state-of-the-art competing risk models (ensemble model: 0.670; Cox-PH: 0.668; Random Survival Forest: 0.660; Dynamic DeepHit: 0.646). The probability of GRFS after 2 years was 30.54% for the high-risk group and 40.69% for the low-risk group (hazard ratio compared with the low-risk group: 2.127; 95% CI, 1.19-3.80). We developed a novel predictive model for survival analysis that showed superior risk stratification to existing methods using a stacked ensemble of multiple machine-learning algorithms.

A novel Iowa-Mayo validated composite risk assessment tool for allogeneic stem cell transplantation survival outcome prediction

Allogeneic hematopoietic stem cell transplantation (HSCT) is a curative option for many hematologic conditions and is associated with considerable morbidity and mortality. Therefore, prognostic tools are essential to navigate the complex patient, disease, donor, and transplant characteristics that differentially influence outcomes. We developed a novel, comprehensive composite prognostic tool. Using a lasso-penalized Cox regression model (n = 273), performance status, HCT-CI, refined disease-risk index (rDRI), donor and recipient CMV status, and donor age were identified as predictors of disease-free survival (DFS). The results for overall survival (OS) were similar except for recipient CMV status not being included in the model. Models were validated in an external dataset (n = 378) and resulted in a c-statistic of 0.61 and 0.62 for DFS and OS, respectively. Importantly, this tool incorporates donor age as a variable, which has an important role in HSCT outcomes. This needs to be further studied in prospective models. An easy-to-use and a web-based nomogram can be accessed here: https://allohsctsurvivalcalc.iowa.uiowa.edu/ .

Data mining in clinical big data: the frequently used databases, steps, and methodological models

Many high quality studies have emerged from public databases, such as Surveillance, Epidemiology, and End Results (SEER), National Health and Nutrition Examination Survey (NHANES), The Cancer Genome Atlas (TCGA), and Medical Information Mart for Intensive Care (MIMIC); however, these data are often characterized by a high degree of dimensional heterogeneity, timeliness, scarcity, irregularity, and other characteristics, resulting in the value of these data not being fully utilized. Data-mining technology has been a frontier field in medical research, as it demonstrates excellent performance in evaluating patient risks and assisting clinical decision-making in building disease-prediction models. Therefore, data mining has unique advantages in clinical big-data research, especially in large-scale medical public databases. This article introduced the main medical public database and described the steps, tasks, and models of data mining in simple language. Additionally, we described data-mining methods along with their practical applications. The goal of this work was to aid clinical researchers in gaining a clear and intuitive understanding of the application of data-mining technology on clinical big-data in order to promote the production of research results that are beneficial to doctors and patients.

A Review on the Role of Artificial Intelligence in Stem Cell Therapy: An Initiative for Modern Medicines

Stem Cells (SCs) show a wide range of applications in the treatment of numerous diseases, including neurodegenerative diseases, diabetes, cardiovascular diseases, cancer, etc. SC related research has gained popularity owing to the unique characteristics of self-renewal and differentiation. Artificial Intelligence (AI), an emerging field of computer science and engineering, has shown potential applications in different fields like robotics, agriculture, home automation, healthcare, banking, and transportation since its invention. This review aims to describe the various applications of AI in SC biology, including understanding the behavior of SCs, recognizing individual cell type before undergoing differentiation, characterization of SCs using mathematical models and prediction of mortality risk associated with SC transplantation. This review emphasizes the role of neural networks in SC biology and further elucidates the concepts of machine learning and deep learning and their applications in SC research.

Predicting Intensive Care Unit Length of Stay After Acute Type A Aortic Dissection Surgery Using Machine Learning

Background: Patients with acute type A aortic dissection are usually transferred to the intensive care unit (ICU) after surgery. Prolonged ICU length of stay (ICU-LOS) is associated with higher level of care and higher mortality. We aimed to develop and validate machine learning models for predicting ICU-LOS after acute type A aortic dissection surgery.

Methods: A total of 353 patients with acute type A aortic dissection transferred to ICU after surgery from September 2016 to August 2019 were included. The patients were randomly divided into the training dataset (70%) and the validation dataset (30%). Eighty-four preoperative and intraoperative factors were collected for each patient. ICU-LOS was divided into four intervals (<4, 4–7, 7–10, and >10 days) according to interquartile range. Kendall correlation coefficient was used to identify factors associated with ICU-LOS. Five classic classifiers, Naive Bayes, Linear Regression, Decision Tree, Random Forest, and Gradient Boosting Decision Tree, were developed to predict ICU-LOS. Area under the curve (AUC) was used to evaluate the models' performance.

Results: The mean age of patients was 51.0 ± 10.9 years and 307 (87.0%) were males. Twelve predictors were identified for ICU-LOS, namely, D-dimer, serum creatinine, lactate dehydrogenase, cardiopulmonary bypass time, fasting blood glucose, white blood cell count, surgical time, aortic cross-clamping time, with Marfan's syndrome, without Marfan's syndrome, without aortic aneurysm, and platelet count. Random Forest yielded the highest performance, with an AUC of 0.991 (95% confidence interval [CI]: 0.978–1.000) and 0.837 (95% CI: 0.766–0.908) in the training and validation datasets, respectively.

Conclusions: Machine learning has the potential to predict ICU-LOS for acute type A aortic dissection. This tool could improve the management of ICU resources and patient-throughput planning, and allow better communication with patients and their families.

Predicting 30-day mortality after ST elevation myocardial infarction: Machine learning- based random forest and its external validation using two independent nationwide datasets

BACKGROUND

Various prognostic models for mortality prediction following ST-segment elevation myocardial infarction (STEMI) have been developed over the past two decades. Our group has previously demonstrated that machine learning (ML)-based models can outperform known risk scores for 30-day mortality post-STEMI. The study aimed to redevelop an ML-based random forest prediction model for 30-day mortality post-STEMI and externally validate it on a large cohort.

METHODS

This was a retrospective, supervised learning, data mining study developed on the Acute Coronary Syndrome Israeli Survey (ACSIS) registry and the Myocardial Ischemia National Audit Project (MINAP) for external validation. Patients included received reperfusion therapy for STEMI between 2006 and 2016. Discrimination and calibration performances were assessed for two developed models and compared with the Global Registry of Acute Cardiac Events (GRACE) score.

RESULTS

The ACSIS cohort (2,782 included /15,212 total) and MINAP cohort (22,693 included/735,000 total) were significantly different in most variables, yet similar in 30-day mortality rate (4.3-4.4%). Random forest models were developed on the ACSIS cohort with a full model including all 32 variables and a simple model including the 10 most important ones. Features' importance was calculated using the varImp function measuring how much each feature contributes to the data's homogeneity. Applying the optimized models on the MINAP validation cohort showed high discrimination of area under the curve (AUC) = 0.804 (0.786-0.822) for the full model, and AUC = 0.787 (0.748-0.780) using the simple model, compared with the GRACE risk score discrimination of AUC = 0.764 (0.748-0.780). All models were not well calibrated for the MINAP data. Following Platt scaling on 20% of the MINAP data, the random forest models calibration improved while the GRACE calibration did not change.

CONCLUSIONS

The random forest predictive model for 30-day mortality post STEMI, developed on the ACSIS national registry, has been validated in the MINAP large external cohort and can be applied early at admission for risk stratification. The model performed better than the commonly used GRACE score. Furthermore, to the best of our knowledge, this is the first externally validated ML-based model for STEMI.

A Deep Learning-Based Method for Forecasting Gold Price with Respect to Pandemics

The spread of COVID-19 has had a devastating impact on the world economy, international trade relations, and globalization. As this pandemic advances and new potential pandemics are on the horizon, a precise analysis of recent fluctuations of trade becomes necessary for international decisions and controlling the world in a similar crisis. The COVID-19 pandemic made a new pattern of trade in the world and affected how businesses work and trade with each other. It means that every potential pandemic or any unprecedented event in the world can change the market rules. This research develops a novel model to have a proper estimation of the stock market values with respect to the COVID-19 dataset using long short-term memory networks (LSTM). The goal of this study is to establish a model that can predict near future regarding the variable set of features. The nature of the features in each pandemic is completely different; therefore, prediction results for a pandemic by a specific model cannot be applied to other pandemics. Hence, recognizing and extracting the features which affect the pandemic is pivotal. In this study, we develop a framework that provides a better understanding of the features and feature selection process. Although the global impacts of COVID-19 are complicated, we are trying to show how additional features like COVID-19 cases can help to forecast in a real-world scenario, rather than relying solely on the history of tickers, which is used conventionally for prediction. This study is based on a preliminary analysis of features such as COVID-19 cases and other market tickers for enhancing forecasting models’ performance against fluctuations in the market. Our predictors are based on the market value data and COVID-19 pandemic daily time-series data (i.e. the number of new cases). In this study, we selected Gold price as a base for our forecasting task which can be replaced by any other markets. We have applied Convolutional Neural Networks (CNN) LSTM, vector sequence output LSTM, Bidirectional LSTM, and encoder–decoder LSTM on the dataset. The results of the vector sequence output LSTM achieved an MSE of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$6.0e-4$$\end{document}6.0e-4, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$8.0e-4$$\end{document}8.0e-4, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$2.0e-3$$\end{document}2.0e-3 on the validation set respectfully for 1 day, 2 days, and 30 days predictions in advance which are outperforming other proposed method in the literature.

Use of Machine Learning Models to Predict Death After Acute Myocardial Infarction

Importance

Accurate prediction of adverse outcomes after acute myocardial infarction (AMI) can guide the triage of care services and shared decision-making, and novel methods hold promise for using existing data to generate additional insights.

Objective

To evaluate whether contemporary machine learning methods can facilitate risk prediction by including a larger number of variables and identifying complex relationships between predictors and outcomes.

Design, Setting, and Participants

This cohort study used the American College of Cardiology Chest Pain-MI Registry to identify all AMI hospitalizations between January 1, 2011, and December 31, 2016. Data analysis was performed from February 1, 2018, to October 22, 2020.

Main Outcomes and Measures

Three machine learning models were developed and validated to predict in-hospital mortality based on patient comorbidities, medical history, presentation characteristics, and initial laboratory values. Models were developed based on extreme gradient descent boosting (XGBoost, an interpretable model), a neural network, and a meta-classifier model. Their accuracy was compared against the current standard developed using a logistic regression model in a validation sample.

Results

A total of 755 402 patients (mean [SD] age, 65 [13] years; 495 202 [65.5%] male) were identified during the study period. In independent validation, 2 machine learning models, gradient descent boosting and meta-classifier (combination including inputs from gradient descent boosting and a neural network), marginally improved discrimination compared with logistic regression (C statistic, 0.90 for best performing machine learning model vs 0.89 for logistic regression). Nearly perfect calibration in independent validation data was found in the XGBoost (slope of predicted to observed events, 1.01; 95% CI, 0.99-1.04) and the meta-classifier model (slope of predicted-to-observed events, 1.01; 95% CI, 0.99-1.02), with more precise classification across the risk spectrum. The XGBoost model reclassified 32 393 of 121 839 individuals (27%) and the meta-classifier model reclassified 30 836 of 121 839 individuals (25%) deemed at moderate to high risk for death in logistic regression as low risk, which were more consistent with the observed event rates.

Conclusions and Relevance

In this cohort study using a large national registry, none of the tested machine learning models were associated with substantive improvement in the discrimination of in-hospital mortality after AMI, limiting their clinical utility. However, compared with logistic regression, XGBoost and meta-classifier models, but not the neural network, offered improved resolution of risk for high-risk individuals.

Predicting Long-Term Mortality in TAVI Patients Using Machine Learning Techniques

Background: Whereas transcatheter aortic valve implantation (TAVI) has become the gold standard for aortic valve stenosis treatment in high-risk patients, it has recently been extended to include intermediate risk patients. However, the mortality rate at 5 years is still elevated. The aim of the present study was to develop a novel machine learning (ML) approach able to identify the best predictors of 5-year mortality after TAVI among several clinical and echocardiographic variables, which may improve the long-term prognosis. Methods: We retrospectively enrolled 471 patients undergoing TAVI. More than 80 pre-TAVI variables were collected and analyzed through different feature selection processes, which allowed for the identification of several variables with the highest predictive value of mortality. Different ML models were compared. Results: Multilayer perceptron resulted in the best performance in predicting mortality at 5 years after TAVI, with an area under the curve, positive predictive value, and sensitivity of 0.79, 0.73, and 0.71, respectively. Conclusions: We presented an ML approach for the assessment of risk factors for long-term mortality after TAVI to improve clinical prognosis. Fourteen potential predictors were identified with the organic mitral regurgitation (myxomatous or calcific degeneration of the leaflets and/or annulus) which showed the highest impact on 5 years mortality.

Comparison of Approaches for Prediction of Renal Replacement Therapy-Free Survival in Patients with Acute Kidney Injury

BACKGROUND/AIMS

Acute kidney injury (AKI) in critically ill patients is common, and continuous renal replacement therapy (CRRT) is a preferred mode of renal replacement therapy (RRT) in hemodynamically unstable patients. Prediction of clinical outcomes in patients on CRRT is challenging. We utilized several approaches to predict RRT-free survival (RRTFS) in critically ill patients with AKI requiring CRRT.

METHODS

We used the Medical Information Mart for Intensive Care (MIMIC-III) database to identify patients ≥18 years old with AKI on CRRT, after excluding patients who had ESRD on chronic dialysis, and kidney transplantation. We defined RRTFS as patients who were discharged alive and did not require RRT ≥7 days prior to hospital discharge. We utilized all available biomedical data up to CRRT initiation. We evaluated 7 approaches, including logistic regression (LR), random forest (RF), support vector machine (SVM), adaptive boosting (AdaBoost), extreme gradient boosting (XGBoost), multilayer perceptron (MLP), and MLP with long short-term memory (MLP + LSTM). We evaluated model performance by using area under the receiver operating characteristic (AUROC) curves.

RESULTS

Out of 684 patients with AKI on CRRT, 205 (30%) patients had RRTFS. The median age of patients was 63 years and their median Simplified Acute Physiology Score (SAPS) II was 67 (interquartile range 52-84). The MLP + LSTM showed the highest AUROC (95% CI) of 0.70 (0.67-0.73), followed by MLP 0.59 (0.54-0.64), LR 0.57 (0.52-0.62), SVM 0.51 (0.46-0.56), AdaBoost 0.51 (0.46-0.55), RF 0.44 (0.39-0.48), and XGBoost 0.43 (CI 0.38-0.47).

CONCLUSIONS

A MLP + LSTM model outperformed other approaches for predicting RRTFS. Performance could be further improved by incorporating other data types.

Machine Learning Models to Predict Major Adverse Cardiovascular Events After Orthotopic Liver Transplantation: A Cohort Study

OBJECTIVE

To develop machine learning models that can predict post-transplantation major adverse cardiovascular events (MACE), all-cause mortality, and cardiovascular mortality in patients undergoing liver transplantation (LT).

DESIGN

Retrospective cohort study.

SETTING

High-volume tertiary care center.

PARTICIPANTS

The study comprised 1,459 consecutive patients undergoing LT between January 2008 and December 2019.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

MACE, all-cause mortality, and cardiovascular mortality were modeled using logistic regression, least absolute shrinkage and selection surgery regression, random forests, support vector machine, and gradient-boosted modeling (GBM). All models were built by splitting data into training and testing cohorts, and performance was assessed using five-fold cross-validation based on the area under the receiver operating characteristic curve and Harrell's C statistic. A total of 1,459 patients were included in the final cohort; 1,425 (97.7%) underwent index transplantation, 963 (66.0%) were female, the median age at transplantation was 57 (11-70) years, and the median Model for End-Stage Liver Disease score was 20 (6-40). Across all outcomes, the GBM model XGBoost achieved the highest performance, with an area under the receiver operating curve of 0.71 (95% confidence interval [CI] 0.63-0.79) for MACE, a Harrell's C statistic of 0.64 (95% CI 0.57-0.73) for overall survival, and 0.72 (95% CI 0.59-0.85) for cardiovascular mortality over a mean follow-up of 4.4 years. Examination of Shapley values for the GBM model revealed that on the cohort-wide level, the top influential factors for postoperative MACE were age at transplantation, diabetes, serum creatinine, cirrhosis caused by nonalcoholic steatohepatitis, right ventricular systolic pressure, and left ventricular ejection fraction.

CONCLUSION

Machine learning models developed using data from a tertiary care transplantation center achieved good discriminant function in predicting post-LT MACE, all-cause mortality, and cardiovascular mortality. These models can support clinicians in recipient selection and help screen individuals who may be at elevated risk for post-transplantation MACE.

Machine learning models predict coagulopathy in spontaneous intracerebral hemorrhage patients in ER

AIMS

Coagulation abnormality is one of the primary concerns for patients with spontaneous intracerebral hemorrhage admitted to ER. Conventional laboratory indicators require hours for coagulopathy diagnosis, which brings difficulties for appropriate intervention within the optimal window. This study evaluates the possibility of building efficient coagulopathy prediction models using data mining and machine learning algorithms.

METHODS

A retrospective cohort enrolled 1668 cases with acute spontaneous intracerebral hemorrhage from three medical centers, excluding those under antithrombotic therapies. Coagulopathy-related clinical parameters were initially screened by univariate analysis. Two machine learning algorithms, the random forest and the support vector machine, were deployed via an approach of four-fold cross-validation to screen out the most important parameters contributing to the occurrence of coagulopathy. Model discrimination was assessed using metrics, including accuracy, precision, recall, and F1 score.

RESULTS

Albumin/globulin ratio, neutrophil count, lymphocyte percentage, aspartate transaminase, alanine transaminase, hemoglobin, platelet count, white blood cell count, neutrophil percentage, systolic and diastolic pressure were identified as major predictors to the occurrence of acute coagulopathy. Compared to support vector machine, the model based on the random forest algorithm showed better accuracy (93.1%, 95% confidence interval [CI]: 0.913-0.950), precision (92.4%, 95% CI: 0.897-0.951), F1 score (91.5%, 95% CI: 0.889-0.964), and recall score (93.6%, 95% CI: 0.909-0.964), and yielded higher area under the receiver operating characteristic curve (AU-ROC) (0.962, 95% CI: 0.942-0.982).

CONCLUSION

The constructed models exhibit good prediction accuracy and efficiency. It might be used in clinical practice to facilitate target intervention for acute coagulopathy in patients with spontaneous intracerebral hemorrhage.

Using a machine learning algorithm to predict acute graft-versus-host disease following allogeneic transplantation

Acute graft-versus-host disease (aGVHD) is 1 of the critical complications that often occurs following allogeneic hematopoietic stem cell transplantation (HSCT). Thus far, various types of prediction scores have been created using statistical calculations. The primary objective of this study was to establish and validate the machine learning-dependent index for predicting aGVHD. This was a retrospective cohort study that involved analyzing databases of adult HSCT patients in Japan. The alternating decision tree (ADTree) machine learning algorithm was applied to develop models using the training cohort (70%). The ADTree algorithm was confirmed using the hazard model on data from the validation cohort (30%). Data from 26 695 HSCT patients transplanted from allogeneic donors between 1992 and 2016 were included in this study. The cumulative incidence of aGVHD was 42.8%. Of >40 variables considered, 15 were adapted into a model for aGVHD prediction. The model was tested in the validation cohort, and the incidence of aGVHD was clearly stratified according to the categorized ADTree scores; the cumulative incidence of aGVHD was 29.0% for low risk and 58.7% for high risk (hazard ratio, 2.57). Predicting scores for aGVHD also demonstrated the link between the risk of development aGVHD and overall survival after HSCT. The machine learning algorithms produced clinically reasonable and robust risk stratification scores. The relatively high reproducibility and low impacts from the interactions among the variables indicate that the ADTree algorithm, along with the other data-mining approaches, may provide tools for establishing risk score.

Machine Learning Prediction Models for In-Hospital Mortality After Transcatheter Aortic Valve Replacement

OBJECTIVES

This study sought to develop and compare an array of machine learning methods to predict in-hospital mortality after transcatheter aortic valve replacement (TAVR) in the United States.

BACKGROUND

Existing risk prediction tools for in-hospital complications in patients undergoing TAVR have been designed using statistical modeling approaches and have certain limitations.

METHODS

Patient data were obtained from the National Inpatient Sample database from 2012 to 2015. The data were randomly divided into a development cohort (n = 7,615) and a validation cohort (n = 3,268). Logistic regression, artificial neural network, naive Bayes, and random forest machine learning algorithms were applied to obtain in-hospital mortality prediction models.

RESULTS

A total of 10,883 TAVRs were analyzed in our study. The overall in-hospital mortality was 3.6%. Overall, prediction models' performance measured by area under the curve were good (>0.80). The best model was obtained by logistic regression (area under the curve: 0.92; 95% confidence interval: 0.89 to 0.95). Most obtained models plateaued after introducing 10 variables. Acute kidney injury was the main predictor of in-hospital mortality ranked with the highest mean importance in all the models. The National Inpatient Sample TAVR score showed the best discrimination among available TAVR prediction scores.

CONCLUSIONS

Machine learning methods can generate robust models to predict in-hospital mortality for TAVR. The National Inpatient Sample TAVR score should be considered for prognosis and shared decision making in TAVR patients.

Machine Learning Models with Preoperative Risk Factors and Intraoperative Hypotension Parameters Predict Mortality After Cardiac Surgery

OBJECTIVES

Machine learning models used to predict postoperative mortality rarely include intraoperative factors. Several intraoperative factors like hypotension (IOH), vasopressor-inotropes, and cardiopulmonary bypass (CPB) time are significantly associated with postoperative outcomes. The authors explored the ability of machine learning models incorporating intraoperative risk factors to predict mortality after cardiac surgery.

DESIGN

Retrospective study.

SETTING

Tertiary hospital.

PARTICIPANTS

A total of 5,015 adults who underwent cardiac surgery from 2008 to 2016.

INTERVENTION

None.

MEASUREMENTS AND MAIN RESULTS

The intraoperative phase was divided into the following: (1) CPB, (2) outside CPB, and (3) total surgery for quantifying IOH only. Phase-specific IOH parameters (area under the curve for mean arterial pressure <65 mmHg), vasopressor-inotropes (norepinephrine equivalents), duration, and cross-clamp time, along with preoperative risk factors ,were incorporated into the models. The primary outcome was mortality. The following 5 models were applied to 3 intraoperative phases separately: (1) logistic regression, (2) random forests, (3) neural networks, (4) support vector machines, and (5) extreme gradient boosting (XGB). Mortality was predicted using area under the receiver operating characteristic curve. Of 5,015 patients included, 112 (2.2%) died. XGB model from the outside-CPB phase predicted mortality better with area under the receiver operating characteristic curve, 95% confidence interval (CI): 0.88(0.83-0.94); positive predictive value, 0.10(0.06-0.15); specificity 0.85 (0.83-0.87) and sensitivity 0.75 (0.57-0.90).

CONCLUSION

XGB machine learning model from IOH outside the CPB phase seemed to offer a better discrimination, sensitivity, specificity, and positive predictive value compared with other models. Machine learning models incorporating intraoperative adverse factors might offer better predictive ability for risk stratification and triaging of patients after cardiac surgery.

Machine learning and artificial intelligence in haematology

Digitalization of the medical record and integration of genomic methods into clinical practice have resulted in an unprecedented wealth of data. Machine learning is a subdomain of artificial intelligence that attempts to computationally extract meaningful insights from complex data structures. Applications of machine learning in haematological scenarios are steadily increasing. However, basic concepts are often unfamiliar to clinicians and investigators. The purpose of this review is to provide readers with tools to interpret and critically appraise machine learning literature. We begin with the elucidation of standard terminology and then review examples in haematology. Guidelines for designing and evaluating machine-learning studies are provided. Finally, we discuss limitations of the machine-learning approach.

External validation and comparison of multiple prognostic scores in allogeneic hematopoietic stem cell transplantation

Clinical decisions in allogeneic hematopoietic stem cell transplantation (allo-HSCT) are supported by the use of prognostic scores for outcome prediction. Scores vary in their features and in the composition of development cohorts. We sought to externally validate and compare the performance of 8 commonly applied scoring systems on a cohort of allo-HSCT recipients. Among 528 patients studied, acute myeloid leukemia was the leading transplant indication (44%) and 46% of patients had a matched sibling donor. Most models successfully grouped patients into higher and lower risk strata, supporting their use for risk classification. However, discrimination varied (2-year overall survival area under the receiver operating characteristic curve [AUC]: revised Pretransplantation Assessment of Mortality [rPAM], 0.64; PAM, 0.63; revised Disease Risk Index [rDRI], 0.62; Endothelial Activation and Stress Index [EASIx], 0.60; combined European Society for Blood and Marrow Transplantation [EBMT]/Hematopoietic Cell Transplantation-specific Comorbidity Index [HCT-CI], 0.58; EBMT, 0.58; Comorbidity-Age, 0.58; HCT-CI, 0.55); AUC ranges from 0.5 (random) to 1.0 (perfect prediction). rPAM and PAM, which had the greatest predictive capacity across all outcomes, are comprehensive models including patient, disease, and transplantation information. Interestingly, EASIx, a biomarker-driven model, had comparable performance for nonrelapse mortality (NRM; 2-year AUC, 0.65) but no predictive value for relapse (2-year AUC, 0.53). Overall, allo-HSCT prognostic systems may be useful for risk stratification, but individual prediction remains a challenge, as reflected by the scores' limited discriminative capacity.

Preoperative Prediction of Lymph Node Metastasis in Patients With Early-T-Stage Non-small Cell Lung Cancer by Machine Learning Algorithms

Background: Lymph node metastasis (LNM) is difficult to precisely predict before surgery in patients with early-T-stage non-small cell lung cancer (NSCLC). This study aimed to develop machine learning (ML)-based predictive models for LNM.

Methods: Clinical characteristics and imaging features were retrospectively collected from 1,102 NSCLC ≤ 2 cm patients. A total of 23 variables were included to develop predictive models for LNM by multiple ML algorithms. The models were evaluated by the receiver operating characteristic (ROC) curve for predictive performance and decision curve analysis (DCA) for clinical values. A feature selection approach was used to identify optimal predictive factors.

Results: The areas under the ROC curve (AUCs) of the 8 models ranged from 0.784 to 0.899. Some ML-based models performed better than models using conventional statistical methods in both ROC curves and decision curves. The random forest classifier (RFC) model with 9 variables introduced was identified as the best predictive model. The feature selection indicated the top five predictors were tumor size, imaging density, carcinoembryonic antigen (CEA), maximal standardized uptake value (SUV_max), and age.

Conclusions: By incorporating clinical characteristics and radiographical features, it is feasible to develop ML-based models for the preoperative prediction of LNM in early-T-stage NSCLC, and the RFC model performed best.

Machine Learning Algorithms for the Prediction of Central Lymph Node Metastasis in Patients With Papillary Thyroid Cancer

BACKGROUND

Central lymph node metastasis (CLNM) occurs frequently in patients with papillary thyroid cancer (PTC), but performing prophylactic central lymph node dissection is still controversial. There are no reliable models for predicting CLNM. This study aimed to develop predictive models for CLNM by machine learning (ML) algorithms.

METHODS

Patients with PTC who underwent initial thyroid resection at our hospital between January 2018 and December 2019 were enrolled. A total of 22 variables, including clinical characteristics and ultrasonography (US) features, were used for conventional univariate and multivariate analysis and to construct ML-based models. A 5-fold cross validation strategy was used for validation and a feature selection approach was applied to identify risk factors.

RESULTS

The areas under the receiver operating characteristic curve (AUC) of 7 models ranged from 0.680 to 0.731. All models performed significantly better than US (AUC=0.623) in predicting CLNM (P<0.05). In decision curve, most of the models also performed better than US. The gradient boosting decision tree model with 7 variables was identified as the best model because of its best performance in both ROC (AUC=0.731) and decision curves. Based on multivariate analysis and feature selection, young age, male sex, low serum thyroid peroxidase antibody and US features such as suspected lymph nodes, microcalcification and tumor size > 1.1 cm were the most contributing predictors for CLNM.

CONCLUSIONS

It is feasible to develop predictive models of CLNM in PTC patients by incorporating clinical characteristics and US features. The ML algorithm may be a useful tool for the prediction of lymph node metastasis in thyroid cancer.

Computational prediction of implantation outcome after embryo transfer

The aim of this study is to develop a computational prediction model for implantation outcome after an embryo transfer cycle. In this study, information of 500 patients and 1360 transferred embryos, including cleavage and blastocyst stages and fresh or frozen embryos, from April 2016 to February 2018, were collected. The dataset containing 82 attributes and a target label (indicating positive and negative implantation outcomes) was constructed. Six dominant machine learning approaches were examined based on their performance to predict embryo transfer outcomes. Also, feature selection procedures were used to identify effective predictive factors and recruited to determine the optimum number of features based on classifiers performance. The results revealed that random forest was the best classifier (accuracy = 90.40% and area under the curve = 93.74%) with optimum features based on a 10-fold cross-validation test. According to the Support Vector Machine-Feature Selection algorithm, the ideal numbers of features are 78. Follicle stimulating hormone/human menopausal gonadotropin dosage for ovarian stimulation was the most important predictive factor across all examined embryo transfer features. The proposed machine learning-based prediction model could predict embryo transfer outcome and implantation of embryos with high accuracy, before the start of an embryo transfer cycle.

A Review of Machine Learning Approaches in Assisted Reproductive Technologies

INTRODUCTION

Assisted reproductive technologies (ART) are recent improvements in infertility treatment. However, there is no significant increase in pregnancy rates with the aid of ART. Costly and complex process of ART's makes them as challenging issues. Computational prediction models could predict treatment outcome, before the start of an ART cycle.

AIM

This review provides an overview on machine learning-based prediction models in ART.

METHODS

This article was executed based on a literature review through scientific databases search such as PubMed, Scopus, Web of Science and Google Scholar.

RESULTS

We identified 20 papers reporting on machine learning-based prediction models in IVF or ICSI settings. All of the models were validated only by internal validation. Therefore, external validation of the models and the impact analysis of them were the missing parts of the all studies.

CONCLUSION

Machine learning-based prediction models provide a clinical decision support tool for both clinicians and patients and lead to improvement in ART success rates.

Patient-based prediction algorithm of relapse after allo-HSCT for acute Leukemia and its usefulness in the decision-making process using a machine learning approach

Although allogeneic hematopoietic stem cell transplantation (allo‐HSCT) is a curative therapy for high‐risk acute leukemia (AL), some patients still relapse. Since patients simultaneously have many prognostic factors, difficulties are associated with the construction of a patient‐based prediction algorithm of relapse. The alternating decision tree (ADTree) is a successful classification method that combines decision trees with the predictive accuracy of boosting. It is a component of machine learning (ML) and has the capacity to simultaneously analyze multiple factors. Using ADTree, we attempted to construct a prediction model of leukemia relapse within 1 year of transplantation. With the model of training data (n = 148), prediction accuracy, the AUC of ROC, and the κ‐statistic value were 78.4%, 0.746, and 0.508, respectively. The false positive rate (FPR) of the relapse prediction was as low as 0.134. In an evaluation of the model with validation data (n = 69), prediction accuracy, AUC, and FPR of the relapse prediction were similar at 71.0%, 0.667, and 0.216, respectively. These results suggest that the model is generalized and highly accurate. Furthermore, the output of ADTree may visualize the branch point of treatment. For example, the selection of donor types resulted in different relapse predictions. Therefore, clinicians may change treatment options by referring to the model, thereby improving outcomes. The present results indicate that ML, such as ADTree, will contribute to the decision‐making process in the diversified allo‐HSCT field and be useful for preventing the relapse of leukemia.

Prediction of Recurrence after Transsphenoidal Surgery for Cushing's Disease: The Use of Machine Learning Algorithms

BACKGROUND

There are no reliable predictive models for recurrence after transsphenoidal surgery (TSS) for Cushing's disease (CD).

OBJECTIVES

This study aimed to develop machine learning (ML)-based predictive models for CD recurrence after initial TSS and to evaluate their performance.

METHOD

A total of 354 CD patients were included in this retrospective, supervised learning, data mining study. Predictive models for recurrence were developed according to 17 variables using 7 algorithms. Models were evaluated based on the area under the receiver operating characteristic curve (AUC).

RESULTS

All patients were followed up for over 12 months (mean ± SD 43.80 ± 35.61). The recurrence rate was 13.0%. Age (p < 0.001), postoperative morning serum cortisol nadir (p = 0.002), and postoperative (p < 0.001) and preoperative (p = 0.04) morning adrenocorticotropin (ACTH) level were significantly related to recurrence. AUCs of the 7 models ranged from 0.608 to 0.781. The best performance (AUC = 0.781, 95% CI 0.706, 0.856) appeared when 8 variables were introduced to the random forest (RF) algorithm, which was much better than that of logistic regression (AUC = 0.684, p = 0.008) and that of using only postoperative morning serum cortisol (AUC = 0.635, p < 0.001). According to the feature selection algorithms, the top 3 predictors were age, postoperative serum cortisol, and postoperative ACTH.

CONCLUSIONS

Using ML-based models for prediction of the recurrence after initial TSS for CD is feasible, and RF performs best. The performance of most of ML-based models was significantly better than that of some conventional models.