Elastic Net Regularization Paths for All Generalized Linear Models.

The lasso and elastic net are popular regularized regression models for supervised learning. Friedman, Hastie, and Tibshirani (2010) introduced a computationally efficient algorithm for computing the elastic net regularization path for ordinary least squares regression, logistic regression and multinomial logistic regression, while Simon, Friedman, Hastie, and Tibshirani (2011) extended this work to Cox models for right-censored data. We further extend the reach of the elastic net-regularized regression to all generalized linear model families, Cox models with (start, stop] data and strata, and a simplified version of the relaxed lasso. We also discuss convenient utility functions for measuring the performance of these fitted models.

Feature-weighted elastic net: using "features of features" for better prediction.

In some supervised learning settings, the practitioner might have additional information on the features used for prediction. We propose a new method which leverages this additional information for better prediction. The method, which we call the feature-weighted elastic net ("fwelnet"), uses these "features of features" to adapt the relative penalties on the feature coefficients in the elastic net penalty. In our simulations, fwelnet outperforms the lasso in terms of test mean squared error and usually gives an improvement in true positive rate or false positive rate for feature selection. We also apply this method to early prediction of preeclampsia, where fwelnet outperforms the lasso in terms of 10-fold cross-validated area under the curve (0.86 vs. 0.80). We also provide a connection between fwelnet and the group lasso and suggest how fwelnet might be used for multi-task learning.

Validation of a Machine Learning Model That Outperforms Clinical Risk Scoring Systems for Upper Gastrointestinal Bleeding.

BACKGROUND & AIMS: Scoring systems are suboptimal for determining risk in patients with upper gastrointestinal bleeding (UGIB); these might be improved by a machine learning model. We used machine learning to develop a model to calculate the risk of hospital-based intervention or death in patients with UGIB and compared its performance with other scoring systems. METHODS: We analyzed data collected from consecutive unselected patients with UGIB from medical centers in 4 countries (the United States, Scotland, England, and Denmark; n = 1958) from March 2014 through March 2015. We used the data to derive and internally validate a gradient-boosting machine learning model to identify patients who met a composite endpoint of hospital-based intervention (transfusion or hemostatic intervention) or death within 30 days. We compared the performance of the machine learning prediction model with validated pre-endoscopic clinical risk scoring systems (the Glasgow-Blatchford score [GBS], admission Rockall score, and AIMS65). We externally validated the machine learning model using data from 2 Asia-Pacific sites (Singapore and New Zealand; n = 399). Performance was measured by area under receiver operating characteristic curve (AUC) analysis. RESULTS: The machine learning model identified patients who met the composite endpoint with an AUC of 0.91 in the internal validation set; the clinical scoring systems identified patients who met the composite endpoint with AUC values of 0.88 for the GBS (P = .001), 0.73 for Rockall score (P < .001), and 0.78 for AIMS65 score (P < .001). In the external validation cohort, the machine learning model identified patients who met the composite endpoint with an AUC of 0.90, the GBS with an AUC of 0.87 (P = .004), the Rockall score with an AUC of 0.66 (P < .001), and the AIMS65 with an AUC of 0.64 (P < .001). At cutoff scores at which the machine learning model and GBS identified patients who met the composite endpoint with 100% sensitivity, the specificity values were 26% with the machine learning model versus 12% with GBS (P < .001). CONCLUSIONS: We developed a machine learning model that identifies patients with UGIB who met a composite endpoint of hospital-based intervention or death within 30 days with a greater AUC and higher levels of specificity, at 100% sensitivity, than validated clinical risk scoring systems. This model could increase identification of low-risk patients who can be safely discharged from the emergency department for outpatient management.

Early identification of patients with acute gastrointestinal bleeding using natural language processing and decision rules.

BACKGROUND AND AIM: Guidelines recommend risk stratification scores in patients presenting with gastrointestinal bleeding (GIB), but such scores are uncommonly employed in practice. Automation and deployment of risk stratification scores in real time within electronic health records (EHRs) would overcome a major impediment. This requires an automated mechanism to accurately identify ("phenotype") patients with GIB at the time of presentation. The goal is to identify patients with acute GIB by developing and evaluating EHR-based phenotyping algorithms for emergency department (ED) patients. METHODS: We specified criteria using structured data elements to create rules for identifying patients and also developed multiple natural language processing (NLP)-based approaches for automated phenotyping of patients, tested them with tenfold cross-validation for 10 iterations (n = 7144) and external validation (n = 2988) and compared them with a standard method to identify patient conditions, the Systematized Nomenclature of Medicine. The gold standard for GIB diagnosis was the independent dual manual review of medical records. The primary outcome was the positive predictive value. RESULTS: A decision rule using GIB-specific terms from ED triage and ED review-of-systems assessment performed better than the Systematized Nomenclature of Medicine on internal validation and external validation (positive predictive value = 85% confidence interval:83%-87% vs 69% confidence interval:66%-72%; P < 0.001). The syntax-based NLP algorithm and Bidirectional Encoder Representation from Transformers neural network-based NLP algorithm had similar performance to the structured-data fields decision rule. CONCLUSIONS: An automated decision rule employing GIB-specific triage and review-of-systems terms can be used to trigger EHR-based deployment of risk stratification models to guide clinical decision making in real time for patients with acute GIB presenting to the ED.

Reluctant Generalised Additive Modelling.

Sparse generalised additive models (GAMs) are an extension of sparse generalised linear models that allow a model's prediction to vary non-linearly with an input variable. This enables the data analyst build more accurate models, especially when the linearity assumption is known to be a poor approximation of reality. Motivated by reluctant interaction modelling, we propose a multi-stage algorithm, called reluctant generalised additive modelling (RGAM), that can fit sparse GAMs at scale. It is guided by the principle that, if all else is equal, one should prefer a linear feature over a non-linear feature. Unlike existing methods for sparse GAMs, RGAM can be extended easily to binary, count and survival data. We demonstrate the method's effectiveness on real and simulated examples.

Neural network predicts need for red blood cell transfusion for patients with acute gastrointestinal bleeding admitted to the intensive care unit.

Acute gastrointestinal bleeding is the most common gastrointestinal cause for hospitalization. For high-risk patients requiring intensive care unit stay, predicting transfusion needs during the first 24 h using dynamic risk assessment may improve resuscitation with red blood cell transfusion in admitted patients with severe acute gastrointestinal bleeding. A patient cohort admitted for acute gastrointestinal bleeding (N = 2,524) was identified from the Medical Information Mart for Intensive Care III (MIMIC-III) critical care database and separated into training (N = 2,032) and internal validation (N = 492) sets. The external validation patient cohort was identified from the eICU collaborative database of patients admitted for acute gastrointestinal bleeding presenting to large urban hospitals (N = 1,526). 62 demographic, clinical, and laboratory test features were consolidated into 4-h time intervals over the first 24 h from admission. The outcome measure was the transfusion of red blood cells during each 4-h time interval. A long short-term memory (LSTM) model, a type of Recurrent Neural Network, was compared to a regression-based models on time-updated data. The LSTM model performed better than discrete time regression-based models for both internal validation (AUROC 0.81 vs 0.75 vs 0.75; P < 0.001) and external validation (AUROC 0.65 vs 0.56 vs 0.56; P < 0.001). A LSTM model can be used to predict the need for transfusion of packed red blood cells over the first 24 h from admission to help personalize the care of high-risk patients with acute gastrointestinal bleeding.