Thrombus Imaging Characteristics to Predict Early Recanalization in Anterior Circulation Large Vessel Occlusion Stroke.

The early management of transferred patients with a large vessel occlusion (LVO) stroke could be improved by identifying patients who are likely to recanalize early. We aim to predict early recanalization based on patient clinical and thrombus imaging characteristics. We included 81 transferred anterior-circulation LVO patients with an early recanalization, defined as the resolution of the LVO or the migration to a distal location not reachable with endovascular treatment upon repeated radiological imaging. We compared their clinical and imaging characteristics with all (322) transferred patients with a persistent LVO in the MR CLEAN Registry. We measured distance from carotid terminus to thrombus (DT), thrombus length, density, and perviousness on baseline CT images. We built logistic regression models to predict early recanalization. We validated the predictive ability by computing the median area-under-the-curve (AUC) of the receiver operating characteristics curve for 100 5-fold cross-validations. The administration of intravenous thrombolysis (IVT), longer transfer times, more distal occlusions, and shorter, pervious, less dense thrombi were characteristic of early recanalization. After backward elimination, IVT administration, DT and thrombus density remained in the multivariable model, with an AUC of 0.77 (IQR 0.72-0.83). Baseline thrombus imaging characteristics are valuable in predicting early recanalization and can potentially be used to optimize repeated imaging workflow.

Temporal validation of 30-day mortality prediction models for transcatheter aortic valve implantation using statistical process control - An observational study in a national population.

Background: Various mortality prediction models for Transcatheter Aortic Valve Implantation (TAVI) have been developed in the past years. The effect of time on the performance of such models, however, is unclear given the improvements in the procedure and changes in patient selection, potentially jeopardizing the usefulness of the prediction models in clinical practice. We aim to explore how time affects the performance and stability of different types of prediction models of 30-day mortality after TAVI. Methods: We developed both parametric (Logistic Regression) and non-parametric (XGBoost) models to predict 30-day mortality after TAVI using data from the Netherlands Heart Registration. The models were trained with data from 2013 to the beginning of 2016 and pre-control charts from Statistical Process Control were used to analyse how time affects the models' performance on independent data from the mid of 2016 to the end of 2019. The area under the Receiver Operating Characteristics curve (AUC) was used to evaluate the models in terms of discrimination and the Brier Score (BS), which is related to calibration, in terms of accuracy of the predicted probabilities. To understand the extent to which refitting the models contribute to the models' stability, we also allowed the models to be updated over time. Results: We included data from 11,291 consecutive TAVI patients from hospitals in the Netherlands. The parametric model without re-training had a median AUC of 0.64 (IQR 0.54-0.73) and BS of 0.028 (IQR 0.021-0.035). For the non-parametric model, the median AUC was 0.63 (IQR 0.48-0.68) and BS was 0.027 (IQR 0.021-0.036). Over time, the developed parametric model was stable in terms of AUC and unstable in terms of BS. The non-parametric model was considered unstable in both AUC and BS. Repeated model refitting resulted in stable models in terms of AUC and decreased the variability of BS, although BS was still unstable. The refitted parametric model had a median AUC of 0.66 (IQR 0.57-0.73) and BS of 0.027 (IQR 0.020-0.035) while the non-parametric model had a median AUC of 0.66 (IQR 0.57-0.74) and BS of 0.027 (IQR 0.023-0.035). Conclusions: The temporal validation of the TAVI 30-day mortality prediction models showed that the models refitted over time are more stable and accurate when compared to the frozen models. This highlights the importance of repeatedly refitted models over time to improve or at least maintain their performance stability. The non-parametric approach did not show improvement over the parametric approach.

An integrated approach to geographic validation helped scrutinize prediction model performance and its variability.

OBJECTIVES: To illustrate in-depth validation of prediction models developed on multicenter data. METHODS: For each hospital in a multicenter registry, we evaluated predictive performance of a 30-day mortality prediction model for transcatheter aortic valve implantation (TAVI) using the Netherlands heart registration (NHR) dataset. We measured discrimination and calibration per hospital in a leave-center-out analysis (LCOA). Meta-analysis was used to calculate I2 values per performance metric from the LCOA and to compute mean and confidence interval (CI) estimates. Case mix differences between studies were inspected using the framework of Debray et al. for understanding external validation. We also aimed to discover subgroups (SGs) with high model prediction error (PE) and their distribution over the centers. RESULTS: We studied 16 hospitals with 11,599 TAVI patients with an early mortality of 3.7%. The models' area under the curve (AUCs) had a wide range between hospitals from 0.59 to 0.79, and miscalibration occurred in seven hospitals. Mean AUC from meta-analysis was 0.68 (95% CI 0.65-0.70). I2 values were 0%, 74%, and 0% for AUC, calibration intercept and slope, respectively. Between-hospital case-mix differences were substantial, and model transportability was low. One SG was discovered with marked global PE and was associated with poor performance on validation centers. CONCLUSION: The illustrated combination of approaches provides useful insights to inspect multicenter-based prediction models, and it exposes their limitations in transportability and performance variability when applied to different populations.

Local and Distributed Machine Learning for Inter-hospital Data Utilization: An Application for TAVI Outcome Prediction.

Background: Machine learning models have been developed for numerous medical prognostic purposes. These models are commonly developed using data from single centers or regional registries. Including data from multiple centers improves robustness and accuracy of prognostic models. However, data sharing between multiple centers is complex, mainly because of regulations and patient privacy issues. Objective: We aim to overcome data sharing impediments by using distributed ML and local learning followed by model integration. We applied these techniques to develop 1-year TAVI mortality estimation models with data from two centers without sharing any data. Methods: A distributed ML technique and local learning followed by model integration was used to develop models to predict 1-year mortality after TAVI. We included two populations with 1,160 (Center A) and 631 (Center B) patients. Five traditional ML algorithms were implemented. The results were compared to models created individually on each center. Results: The combined learning techniques outperformed the mono-center models. For center A, the combined local XGBoost achieved an AUC of 0.67 (compared to a mono-center AUC of 0.65) and, for center B, a distributed neural network achieved an AUC of 0.68 (compared to a mono-center AUC of 0.64). Conclusion: This study shows that distributed ML and combined local models techniques, can overcome data sharing limitations and result in more accurate models for TAVI mortality estimation. We have shown improved prognostic accuracy for both centers and can also be used as an alternative to overcome the problem of limited amounts of data when creating prognostic models.

Prediction of Atrial Fibrillation Recurrence after Thoracoscopic Surgical Ablation Using Machine Learning Techniques.

Thoracoscopic surgical ablation (SA) for atrial fibrillation (AF) has shown to be an effective treatment to restore sinus rhythm in patients with advanced AF. Identifying patients who will not benefit from this procedure would be valuable to improve personalized AF therapy. Machine learning (ML) techniques may assist in the improvement of clinical prediction models for patient selection. The aim of this study is to investigate how available baseline characteristics predict AF recurrence after SA using ML techniques. One-hundred-sixty clinical baseline variables were collected from 446 AF patients undergoing SA in our tertiary referral center. Multiple ML models were trained on five outcome measurements, including either all or a number of key variables selected by using the least absolute shrinkage and selection operator (LASSO). There was no difference in model performance between different ML techniques or outcome measurements. Variable selection significantly improved model performance (AUC: 0.73, 95% CI: 0.68-0.77). Subgroup analysis showed a higher model performance in younger patients (<55 years, AUC: 0.82 vs. >55 years, AUC 0.66). Recurrences of AF after SA can be predicted best when using a selection of baseline characteristics, particularly in young patients.

Patient-tailored Contrast Delivery Protocols for Computed Tomography Coronary Angiography: Lower Contrast Dose and Better Image Quality.

PURPOSE: The first objective of this study was to evaluate the efficacy of a patient-tailored contrast delivery protocol for coronary computed tomography angiography (CTCA), in terms of diagnostic coronary attenuation and total iodine load (TIL), by adjusting the iodine delivery rate (IDR) via dilution for body weight and tube voltage (kV), as compared with a protocol with a fixed bolus of contrast in a clinical setting. The secondary objective was to assess the association between the test-bolus data and luminal attenuation in CTCA. MATERIALS AND METHODS: Patients who underwent CTCA with fixed IDR contrast delivery (cohort 1) or with IDR adjusted for body weight and kV settings (70 to 120 kV) (cohort 2) were included, and compared for intravascular luminal attenuation and TIL. The association between intravascular luminal attenuation and test-bolus scan data was investigated with linear regression. RESULTS: In cohort 1 (176 patients), the mean luminal attenuation differed markedly between kV categories, whereas in cohort 2 (154 patients), there were no marked differences. The mean TIL reduced significantly (20.1±1.2 g in cohort 1, 17.7±3.0 g in cohort 2, P<0.001). The peak height of the test-bolus scan was independently associated with luminal attenuation in the ascending aorta, with a 0.58 HU increase per HU peak-height increase (SE=0.18, P<0.001). CONCLUSION: Clinical implementation of a patient-tailored contrast delivery protocol for CTCA, adjusted for body weight and kV, improves luminal attenuation and significantly reduces the TIL. The peak height of the test-bolus scan is associated with luminal attenuation in the ascending aorta in the CTCA scan.

Machine Learning for Predicting Mortality in Transcatheter Aortic Valve Implantation: An Inter-Center Cross Validation Study.

Current prognostic risk scores for transcatheter aortic valve implantation (TAVI) do not benefit yet from modern machine learning techniques, which can improve risk stratification of one-year mortality of patients before TAVI. Despite the advancement of machine learning in healthcare, data sharing regulations are very strict and typically prevent exchanging patient data, without the involvement of ethical committees. A very robust validation approach, including 1300 and 631 patients per center, was performed to validate a machine learning model of one center at the other external center with their data, in a mutual fashion. This was achieved without any data exchange but solely by exchanging the models and the data processing pipelines. A dedicated exchange protocol was designed to evaluate and quantify the model's robustness on the data of the external center. Models developed with the larger dataset offered similar or higher prediction accuracy on the external validation. Logistic regression, random forest and CatBoost lead to areas under curve of the ROC of 0.65, 0.67 and 0.65 for the internal validation and of 0.62, 0.66, 0.68 for the external validation, respectively. We propose a scalable exchange protocol which can be further extended on other TAVI centers, but more generally to any other clinical scenario, that could benefit from this validation approach.

Improving electrocardiogram-based detection of rare genetic heart disease using transfer learning: An application to phospholamban p.Arg14del mutation carriers.

The pathogenic mutation p.Arg14del in the gene encoding Phospholamban (PLN) is known to cause cardiomyopathy and leads to increased risk of sudden cardiac death. Automatic tools might improve the detection of patients with this rare disease. Deep learning is currently the state-of-the-art in signal processing but requires large amounts of data to train the algorithms. In situations with relatively small amounts of data, like PLN, transfer learning may improve accuracy. We propose an ECG-based detection of the PLN mutation using transfer learning from a model originally trained for sex identification. The sex identification model was trained with 256,278 ECGs and subsequently finetuned for PLN detection (155 ECGs of patients with PLN) with two control groups: a balanced age/sex matched group and a randomly selected imbalanced population. The data was split in 10 folds and 20% of the training data was used for validation and early stopping. The models were evaluated with the area under the receiver operating characteristic curve (AUROC) of the testing data. We used gradient activation for explanation of the prediction models. The models trained with transfer learning outperformed the models trained from scratch for both the balanced (AUROC 0.87 vs AUROC 0.71) and imbalanced (AUROC 0.0.90 vs AUROC 0.65) population. The proposed approach was able to improve the accuracy of a rare disease detection model by transfer learning information from a non-manual annotated and abundant label with only limited data available.

Computer versus cardiologist: Is a machine learning algorithm able to outperform an expert in diagnosing a phospholamban p.Arg14del mutation on the electrocardiogram?

BACKGROUND: Phospholamban (PLN) p.Arg14del mutation carriers are known to develop dilated and/or arrhythmogenic cardiomyopathy, and typical electrocardiographic (ECG) features have been identified for diagnosis. Machine learning is a powerful tool used in ECG analysis and has shown to outperform cardiologists. OBJECTIVES: We aimed to develop machine learning and deep learning models to diagnose PLN p.Arg14del cardiomyopathy using ECGs and evaluate their accuracy compared to an expert cardiologist. METHODS: We included 155 adult PLN mutation carriers and 155 age- and sex-matched control subjects. Twenty-one PLN mutation carriers (13.4%) were classified as symptomatic (symptoms of heart failure or malignant ventricular arrhythmias). The data set was split into training and testing sets using 4-fold cross-validation. Multiple models were developed to discriminate between PLN mutation carriers and control subjects. For comparison, expert cardiologists classified the same data set. The best performing models were validated using an external PLN p.Arg14del mutation carrier data set from Murcia, Spain (n = 50). We applied occlusion maps to visualize the most contributing ECG regions. RESULTS: In terms of specificity, expert cardiologists (0.99) outperformed all models (range 0.53-0.81). In terms of accuracy and sensitivity, experts (0.28 and 0.64) were outperformed by all models (sensitivity range 0.65-0.81). T-wave morphology was most important for classification of PLN p.Arg14del carriers. External validation showed comparable results, with the best model outperforming experts. CONCLUSION: This study shows that machine learning can outperform experienced cardiologists in the diagnosis of PLN p.Arg14del cardiomyopathy and suggests that the shape of the T wave is of added importance to this diagnosis.

A morphology based deep learning model for atrial fibrillation detection using single cycle electrocardiographic samples.

BACKGROUND: Deep learning (DL) has shown promising results in improving atrial fibrillation (AF) detection algorithms. However, these models are often criticized because of their "black box" nature. AIM: To develop a morphology based DL model to discriminate AF from sinus rhythm (SR), and to visualize which parts of the ECG are used by the model to derive to the right classification. METHODS: We pre-processed raw data of 1469 ECGs in AF or SR, of patients with a history AF. Input data was generated by normalizing all single cycles (SC) of one ECG lead to SC-ECG samples by 1) centralizing the R wave or 2) scaling from R-to- R wave. Different DL models were trained by splitting the data in a training, validation and test set. By using a DL based heat mapping technique we visualized those areas of the ECG used by the classifier to come to the correct classification. RESULTS: The DL model with the best performance was a feedforward neural network trained by SC-ECG samples on a R-to-R wave basis of lead II, resulting in an accuracy of 0.96 and F1-score of 0.94. The onset of the QRS complex proved to be the most relevant area for the model to discriminate AF from SR. CONCLUSION: The morphology based DL model developed in this study was able to discriminate AF from SR with a very high accuracy. DL model visualization may help clinicians gain insights into which (unrecognized) ECG features are most sensitive to discriminate AF from SR.