Using machine learning to improve risk prediction in durable left ventricular assist devices

Abstract

Risk models have historically displayed only moderate predictive performance in estimating mortality risk in left ventricular assist device therapy. This study evaluated whether machine learning can improve risk prediction for left ventricular assist devices. Primary durable left ventricular assist devices reported in the Interagency Registry for Mechanically Assisted Circulatory Support between March 1, 2006 and December 31, 2016 were included. The study cohort was randomly divided 3:1 into training and testing sets. Logistic regression and machine learning models (extreme gradient boosting) were created in the training set for 90-day and 1-year mortality and their performance was evaluated after bootstrapping with 1000 replications in the testing set. Differences in model performance were also evaluated in cases of concordance versus discordance in predicted risk between logistic regression and extreme gradient boosting as defined by equal size patient tertiles. A total of 16,120 patients were included. Calibration metrics were comparable between logistic regression and extreme gradient boosting. C-index was improved with extreme gradient boosting (90-day: 0.707 [0.683–0.730] versus 0.740 [0.717–0.762] and 1-year: 0.691 [0.673–0.710] versus 0.714 [0.695–0.734]; each p<0.001). Net reclassification index analysis similarly demonstrated an improvement of 48.8% and 36.9% for 90-day and 1-year mortality, respectively, with extreme gradient boosting (each p<0.001). Concordance in predicted risk between logistic regression and extreme gradient boosting resulted in substantially improved c-index for both logistic regression and extreme gradient boosting (90-day logistic regression 0.536 versus 0.752, 1-year logistic regression 0.555 versus 0.726, 90-day extreme gradient boosting 0.623 versus 0.772, 1-year extreme gradient boosting 0.613 versus 0.742, each p<0.001). These results demonstrate that machine learning can improve risk model performance for durable left ventricular assist devices, both independently and as an adjunct to logistic regression.

Early Identification of High-Risk Patients with HCC Relapse: A Spatial Immune Scoring System Enabled by Artificial Intelligence

“…It used CD57 to identify NK cells 25,26 instead of CD56 due to better specificity 60-62 (Extended Data Fig. 1e) and superior prognostic value in HCC 10 . Furthermore, we present the largest cohort of mIHC-stained WSIs that integrates spatial technologies and ML to focus specifically on immune profiles, unlike previous studies that relied on H&E-stained WSIs with ML that are potentially biased towards tumor morphology 17,41,63 .…”

“…Through mIHC staining analysis, we determined the percentages of SPON2 + , ZFP36L2 + , ZFP36 + , VIM + , or HLA-DRB1 + cells at IF and TC for use as predictor variables in building machine learning models that are predictive of HCC relapse outcomes. We evaluated a battery of machine learning models, including a generalized linear model 5 , a support vector machine 7 with a linear kernel or a radial kernel, a decision tree model 8 , a random forest model 9 , and an extreme gradient boosting model 10 .…”

“…Particularly, the redistribution of immune cells within the tumor microenvironment (TME) plays a crucial role in the surveillance and control of HCC 3,4 . High natural killer (NK) cell infiltration and activity are closely associated with improved HCC prognosis [5][6][7][8][9][10] , with intratumoral NK cells holding greater prognostic value over peripheral NK cells 10 , especially at the invasive front (IF). The distribution of immune cells within tumours is closely related to metastatic recurrence [11][12][13][14] , supporting the hypothesis that the spatial distribution of immune cells across tumor regions influences HCC recurrence 15,16 .…”

A Comparison among Different Machine Learning Pretest Approaches to Predict Stress-Induced Ischemia at PET/CT Myocardial Perfusion Imaging

“…The joint PET evaluation improves had a good performance (AUROC = 86%), and the SVM algorithm outperformed the other methods evaluated. In a study [ 30 ], including a total of 16,120 patients, ML improved one-year risk discrimination in predicting durable left ventricular assist devices as compared to logistic regression (C-index 71% vs. 69%, P < 0.001); however, calibration metrics were comparable. Globally, these studies confirm limited value of current clinical models to accurately predict the presence of myocardial ischemia at stress MPI [ 31 ].…”

The future of Cardiothoracic surgery in Artificial intelligence

“…A study by Kilic et al revealed that there is an increase in the use of durable left ventricular assist devices (LVAD) being implanted in the United States but there was no widely used risk stratification tool for LVAD therapy. By utilizing the Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) database they studied the 90-day and 1-year mortality rates following primary LVAD implantation using both logistic regression and machine learning approaches and found that machine learning models were both well-calibrated and had improved discriminatory capability as compared to logistic regression [ 29 ]. Wang et al developed and validated a machine learning method that can predict the amount of red blood cell transfusions required for a cardiothoracic surgery and showed excellent results [ 30 ].…”