One-year mortality of colorectal cancer patients: development and validation of a prediction model using linked national electronic data.

BACKGROUND: The existing literature does not provide a prediction model for mortality of all colorectal cancer patients using contemporary national hospital data. We developed and validated such a model to predict colorectal cancer death within 90, 180 and 365 days after diagnosis. METHODS: Cohort study using linked national cancer and death records. The development population included 27,480 patients diagnosed in England in 2015. The test populations were diagnosed in England in 2016 (n = 26,411) and Wales in 2015-2016 (n = 3814). Predictors were age, gender, socioeconomic status, referral source, performance status, tumour site, TNM stage and treatment intent. Cox regression models were assessed using Brier scores, c-indices and calibration plots. RESULTS: In the development population, 7.4, 11.7 and 17.9% of patients died from colorectal cancer within 90, 180 and 365 days after diagnosis. T4 versus T1 tumour stage had the largest adjusted association with the outcome (HR 4.67; 95% CI: 3.59-6.09). C-indices were 0.873-0.890 (England) and 0.856-0.873 (Wales) in the test populations, indicating excellent separation of predicted risks by outcome status. Models were generally well calibrated. CONCLUSIONS: The model was valid for predicting short-term colorectal cancer mortality. It can provide personalised information to support clinical practice and research.

Linear Deconfounded Score Method: Scoring DAGs With Dense Unobserved Confounding.

This article deals with the discovery of causal relations from a combination of observational data and qualitative assumptions about the nature of causality in the presence of unmeasured confounding. We focus on applications where unobserved variables are known to have a widespread effect on many of the observed ones, which makes the problem particularly difficult for constraint-based methods, because most pairs of variables are conditionally dependent given any other subset, rendering the causal effect unidentifiable. In this article, we show that under the principle of independent mechanisms, unobserved confounding in this setting leaves a statistical footprint in the observed data distribution that allows for disentangling spurious and causal effects. Using this insight, we demonstrate that a sparse linear Gaussian directed acyclic graph (DAG) among observed variables may be recovered approximately and propose a simple adjusted score-based causal discovery algorithm that may be implemented with general-purpose solvers and scales to high-dimensional problems. We find, in addition, that despite the conditions we pose to guarantee causal recovery, performance in practice is robust to large deviations in model assumptions, and extensions to nonlinear structural models are possible.

Logistic regression and machine learning predicted patient mortality from large sets of diagnosis codes comparably.

OBJECTIVE: The objective of the study was to compare the performance of logistic regression and boosted trees for predicting patient mortality from large sets of diagnosis codes in electronic healthcare records. STUDY DESIGN AND SETTING: We analyzed national hospital records and official death records for patients with myocardial infarction (n = 200,119), hip fracture (n = 169,646), or colorectal cancer surgery (n = 56,515) in England in 2015-2017. One-year mortality was predicted from patient age, sex, and socioeconomic status, and 202 to 257 International Classification of Diseases 10th Revision codes recorded in the preceding year or not (binary predictors). Performance measures included the c-statistic, scaled Brier score, and several measures of calibration. RESULTS: One-year mortality was 17.2% (34,520) after myocardial infarction, 27.2% (46,115) after hip fracture, and 9.3% (5,273) after colorectal surgery. Optimism-adjusted c-statistics for the logistic regression models were 0.884 (95% confidence interval [CI]: 0.882, 0.886), 0.798 (0.796, 0.800), and 0.811 (0.805, 0.817). The equivalent c-statistics for the boosted tree models were 0.891 (95% CI: 0.889, 0.892), 0.804 (0.802, 0.806), and 0.803 (0.797, 0.809). Model performance was also similar when measured using scaled Brier scores. All models were well calibrated overall. CONCLUSION: In large datasets of electronic healthcare records, logistic regression and boosted tree models of numerous diagnosis codes predicted patient mortality comparably.

Predicting the Risk of Inpatient Hypoglycemia With Machine Learning Using Electronic Health Records.

OBJECTIVE: We analyzed data from inpatients with diabetes admitted to a large university hospital to predict the risk of hypoglycemia through the use of machine learning algorithms. RESEARCH DESIGN AND METHODS: Four years of data were extracted from a hospital electronic health record system. This included laboratory and point-of-care blood glucose (BG) values to identify biochemical and clinically significant hypoglycemic episodes (BG ≤3.9 and ≤2.9 mmol/L, respectively). We used patient demographics, administered medications, vital signs, laboratory results, and procedures performed during the hospital stays to inform the model. Two iterations of the data set included the doses of insulin administered and the past history of inpatient hypoglycemia. Eighteen different prediction models were compared using the area under the receiver operating characteristic curve (AUROC) through a 10-fold cross validation. RESULTS: We analyzed data obtained from 17,658 inpatients with diabetes who underwent 32,758 admissions between July 2014 and August 2018. The predictive factors from the logistic regression model included people undergoing procedures, weight, type of diabetes, oxygen saturation level, use of medications (insulin, sulfonylurea, and metformin), and albumin levels. The machine learning model with the best performance was the XGBoost model (AUROC 0.96). This outperformed the logistic regression model, which had an AUROC of 0.75 for the estimation of the risk of clinically significant hypoglycemia. CONCLUSIONS: Advanced machine learning models are superior to logistic regression models in predicting the risk of hypoglycemia in inpatients with diabetes. Trials of such models should be conducted in real time to evaluate their utility to reduce inpatient hypoglycemia.

A Hierarchical Bayesian Model for Personalized Survival Predictions.

We study the problem of personalizing survival estimates of patients in heterogeneous populations for clinical decision support. The desiderata are to improve predictions by making them personalized to the patient-at-hand, to better understand diseases and their risk factors, and to provide interpretable model outputs to clinicians. To enable accurate survival prognosis in heterogeneous populations we propose a novel probabilistic survival model which flexibly captures individual traits through a hierarchical latent variable formulation. Survival paths are estimated by jointly sampling the location and shape of the individual survival distribution resulting in patient-specific curves with quantifiable uncertainty estimates. An understanding of model predictions is paramount in medical practice where decisions have major social consequences. We develop a personalized interpreter that can be used to test the effect of covariates on each individual patient, in contrast to traditional methods that focus on population average effects. We extensively validated the proposed approach in various clinical settings, with a special focus on cardiovascular disease.