Search

Abstract Since it emerged in December of 2019, COVID-19 has placed a huge burden on medical care in countries throughout the world, as it led to a huge number of hospitalizations and mortalities. Many medical centers were overloaded, as their intensive care units and auxiliary protection resources proved insufficient, which made the effective allocation of medical resources an urgent matter. This study describes learned survival prediction models that could help medical professionals make effective decisions regarding patient triage and resource allocation. We created multiple data subsets from a publicly available COVID-19 epidemiological dataset to evaluate the effectiveness of various combinations of covariates—age, sex, geographic location, and chronic disease status—in learning survival models (here, “Individual Survival Distributions”; ISDs) for hospital discharge and also for death events. We then supplemented our datasets with demographic and economic information to obtain potentially more accurate survival models. Our extensive experiments compared several ISD models, using various measures. These results show that the “gradient boosting Cox machine” algorithm outperformed the competing techniques, in terms of these performance evaluation metrics, for predicting both an individual’s likelihood of hospital discharge and COVID-19 mortality. Our curated datasets and code base are available at our Github repository for reproducing the results reported in this paper and for supporting future research.

Abstract Lung cancer is the leading cause of human cancer mortality due to the lack of early diagnosis technology. The low-dose computed tomography scan (LDCT) is one of the main techniques to screen cancers. However, LDCT still has a risk of radiation exposure and it is not suitable for the general public. In this study, plasma metabolic profiles of lung cancer were performed using a comprehensive metabolomic method with different liquid chromatography methods coupled with a Q-Exactive high-resolution mass spectrometer. Metabolites with different polarities (amino acids, fatty acids, and acylcarnitines) can be detected and identified as differential metabolites of lung cancer in small volumes of plasma. Logistic regression models were further developed to identify cancer stages and types using those significant biomarkers. Using the Variable Importance in Projection (VIP) and the area under the curve (AUC) scores, we have successfully identified the top 5, 10, and 20 metabolites that can be used to differentiate lung cancer stages and types. The discrimination accuracy and AUC score can be as high as 0.829 and 0.869 using the five most significant metabolites. This study demonstrated that using 5 + metabolites (Palmitic acid, Heptadecanoic acid, 4-Oxoproline, Tridecanoic acid, Ornithine, and etc.) has the potential for early lung cancer screening. This finding is useful for transferring the diagnostic technology onto a point-of-care device for lung cancer diagnosis and prognosis.

We propose a method to predict when a woman will develop breast cancer (BCa) from her lifestyle and health history features. To address this objective, we use data from the Alberta’s Tomorrow Project of 18,288 women to train Individual Survival Distribution (ISD) models to predict an individual’s Breast-Cancer-Onset (BCaO) probability curve. We show that our three-step approach–(1) filling missing data with multiple imputations by chained equations, followed by (2) feature selection with the multivariate Cox method, and finally, (3) using MTLR to learn an ISD model–produced the model with the smallest L1-Hinge loss among all calibrated models with comparable C-index. We also identified 7 actionable lifestyle features that a woman can modify and illustrate how this model can predict the quantitative effects of those changes–suggesting how much each will potentially extend her BCa-free time. We anticipate this approach could be used to identify appropriate interventions for individuals with a higher likelihood of developing BCa in their lifetime.

Alzheimer’s dementia (AD) is a chronic neurodegenerative illness that manifests in a gradual decline of cognitive function. Early identification of AD is essential for managing the ensuing cognitive deficits, which may lead to a better prognostic outcome. Speech data can serve as a window into cognitive functioning and can be used to screen for early signs of AD. This paper describes methods for learning models using speech samples from the DementiaBank database, for identifying which subjects have Alzheimer’s dementia. We consider two machine learning tasks: 1) binary classification to distinguish patients from healthy controls, and 2) regression to estimate each subject’s Mini-Mental State Examination (MMSE) score. To develop models that can use acoustic and/or language features, we explore a variety of dimension reduction techniques, training algorithms, and fusion strategies. Our best performing classification model, using language features with dimension reduction and regularized logistic regression, achieves an accuracy of 85.4% on a held-out test set. On the regression task, a linear regression model trained on a reduced set of language features achieves a root mean square error (RMSE) of 5.62 on the test set. These results demonstrate the promise of using machine learning for detecting cognitive decline from speech in AD patients.