ABSTRACT
Although lung cancer survival status and survival length predictions have primarily been studied individually, a scheme that leverages both fields in an interpretable way for physicians remains elusive. We propose a two-phase data analytic framework that is capable of classifying survival status for 0.5-, 1-, 1.5-, 2-, 2.5-, and 3-year time-points (phase I) and predicting the number of survival months within 3 years (phase II) using recent Surveillance, Epidemiology, and End Results data from 2010 to 2017. In this study, we employ three analytical models (general linear model, extreme gradient boosting, and artificial neural networks), five data balancing techniques (synthetic minority oversampling technique (SMOTE), relocating safe level SMOTE, borderline SMOTE, adaptive synthetic sampling, and majority weighted minority oversampling technique), two feature selection methods (least absolute shrinkage and selection operator (LASSO) and random forest), and the one-hot encoding approach. By implementing a comprehensive data preparation phase, we demonstrate that a computationally efficient and interpretable method such as GLM performs comparably to more complex models. Moreover, we quantify the effects of individual features in phase I and II by exploiting GLM coefficients. To the best of our knowledge, this study is the first to (a) implement a comprehensive data processing approach to develop performant, computationally efficient, and interpretable methods in comparison to black-box models, (b) visualize top factors impacting survival odds by utilizing the change in odds ratio, and (c) comprehensively explore short-term lung cancer survival using a two-phase approach.
Subject(s)
Lung Neoplasms , Humans , Linear Models , Neural Networks, ComputerABSTRACT
Recent investigations on spinel CoMn2O4 have shown its potential for applications in water splitting and fuel cell technologies as it exhibits strong catalytic behavior through oxygen reduction reactivity. To further understand this material, we report for the first time the synthesis of single-crystalline Co1+x Mn2-x O4 thin films using molecular beam epitaxy. By varying sample composition, we establish links between cation stoichiometry and material properties using in-situ x-ray photoelectron spectroscopy, x-ray diffraction, scanning transmission electron microscopy, x-ray absorption spectroscopy, and spectroscopic ellipsometry. Our results indicate that excess Co ions occupy tetrahedral interstitial sites at lower excess Co stoichiometries, and become substitutional for octahedrally-coordinated Mn at higher Co levels. We compare these results with density functional theory models of stoichiometric CoMn2O4 to understand how the Jahn-Teller distortion and hybridization in Mn-O bonds impact the ability to hole dope the material with excess Co. The findings provide important insights into CoMn2O4 and related spinel oxides that are promising candidates for inexpensive oxygen reduction reaction catalysts.
ABSTRACT
Recent investigations on spinel CoMn2O4have shown its potential for applications in water splitting and fuel cell technologies as it exhibits strong catalytic behavior through oxygen reduction reactivity. To further understand this material, we report for the first time the synthesis of single-crystalline Co1+xMn2-xO4thin films using molecular beam epitaxy. By varying sample composition, we establish links between cation stoichiometry and material properties using in-situ x-ray photoelectron spectroscopy, x-ray diffraction, scanning transmission electron microscopy, x-ray absorption spectroscopy, and spectroscopic ellipsometry. Our results indicate that excess Co ions occupy tetrahedral interstitial sites at lower excess Co stoichiometries, and become substitutional for octahedrally-coordinated Mn at higher Co levels. We compare these results with density functional theory models of stoichiometric CoMn2O4to understand how the Jahn-Teller distortion and hybridization in Mn-O bonds impact the ability to hole dope the material with excess Co. The findings provide important insights into CoMn2O4and related spinel oxides that are promising candidates for inexpensive oxygen reduction reaction catalysts.
ABSTRACT
We characterize the structure-property relationship of alkali metal elements in oxygen-passivated graphene pores using the density functional theory that accounts for quantum mechanical effects and charge transfer. Our description is based on the structural and electronic properties of the system and shows common trends among the different alkali metals and pores. We find that these nanopores which serve as docking sites for alkali metal elements give the strongest binding when the size of the pore is similar to the element's van der Waals radius. A linear correlation between the binding energy and the energy location of the alkali element valence state is found for all elements and pores. Analysis of the charge transfer reveals that alkali adsorption increases the local charge in the perimeters of the pore by amounts that depend on the geometry. Moreover, charge distributions in pristine graphene resemble those of an ideal conductor despite its semimetallic character and atomic thickness while oscillations in the vicinity of O-passivated nanopores are observed. Our results suggest that charge transfer is localized within a few nanometers of the pore and, therefore, allude to high density energy storage. The outcomes of this work are significant towards the application of porous graphene as effective membranes for ion filtration of water and electrode applications.
