Logic Forest: an ensemble classifier for discovering logical combinations of binary markers

Novel Machine Learning Approach to Identify Preoperative Risk Factors Associated With Super-Utilization of Medicare Expenditure Following Surgery

Importance

Typically defined as the top 5% of health care users, super-utilizers are responsible for an estimated 40% to 55% of all health care costs. Little is known about which factors may be associated with increased risk of long-term postoperative super-utilization.

Objective

To identify clusters of patients with distinct constellations of clinical and comorbid patterns who may be associated with an elevated risk of super-utilization in the year following elective surgery.

Design, Setting, and Participants

A retrospective longitudinal cohort study of 1 049 160 patients who underwent abdominal aortic aneurysm repair, coronary artery bypass graft, colectomy, total hip arthroplasty, total knee arthroplasty, or lung resection were identified from the 100% Medicare inpatient and outpatient Standard Analytic Files at all inpatient facilities performing 1 or more of the evaluated surgical procedures from 2013 to 2015. Data from 2012 to 2016 were used to evaluate expenditures in the year preceding and following surgery. Using a machine learning approach known as Logic Forest, comorbidities and interactions of comorbidities that put patients at an increased chance of becoming a super-utilizer were identified. All comorbidities, as defined by the Charlson (range, 0-24) and Elixhauser (range, 0-29) comorbidity indices, were used in the analysis. Higher scores indicated higher comorbidity burden. Data analysis was completed on November 16, 2018.

Main Outcome and Measures

Super-utilization of health care in the year following surgery.

Results

In total, 1 049 160 patients met inclusion criteria and were included in the analytic cohort. Their median (interquartile range) age was 73 (69-78) years, and approximately 40% were male. Super-utilizers comprised 4.8% of the overall cohort (n = 79 746) yet incurred 31.7% of the expenditures. Although the difference in overall expenditures per person between super-utilizers ($4049) and low users ($2148) was relatively modest prior to surgery, the difference in expenditures between super-utilizers ($79 698) vs low users ($2977) was marked in the year following surgery. Risk factors associated with super-utilization of health care included hemiplegia/paraplegia (odds ratio, 5.2; 95% CI, 4.4-6.2), weight loss (odds ratio, 3.5; 95% CI, 2.9-4.2), and congestive heart failure with chronic kidney disease stages I to IV (odds ratio, 3.4; 95% CI, 3.0-3.9).

Conclusions and Relevance

Super-utilizers comprised only a small fraction of the surgical population yet were responsible for a disproportionate amount of Medicare expenditure. Certain subpopulations were associated with super-utilization of health care following surgical intervention despite having lower overall use in the preoperative period.

An Analytic Approach Using Candidate Gene Selection and Logic Forest to Identify Gene by Environment Interactions (G × E) for Systemic Lupus Erythematosus in African Americans

Development and progression of many human diseases, such as systemic lupus erythematosus (SLE), are hypothesized to result from interactions between genetic and environmental factors. Current approaches to identify and evaluate interactions are limited, most often focusing on main effects and two-way interactions. While higher order interactions associated with disease are documented, they are difficult to detect since expanding the search space to all possible interactions of p predictors means evaluating 2^p − 1 terms. For example, data with 150 candidate predictors requires considering over 10⁴⁵ main effects and interactions. In this study, we present an analytical approach involving selection of candidate single nucleotide polymorphisms (SNPs) and environmental and/or clinical factors and use of Logic Forest to identify predictors of disease, including higher order interactions, followed by confirmation of the association between those predictors and interactions identified with disease outcome using logistic regression. We applied this approach to a study investigating whether smoking and/or secondhand smoke exposure interacts with candidate SNPs resulting in elevated risk of SLE. The approach identified both genetic and environmental risk factors, with evidence suggesting potential interactions between exposure to secondhand smoke as a child and genetic variation in the ITGAM gene associated with increased risk of SLE.

Key Considerations and Methods in the Study of Gene-Environment Interactions

With increased involvement of genetic data in most epidemiological investigations, gene-environment (G × E) interactions now stand as a topic, which must be meticulously assessed and thoroughly understood. The level, mode, and outcomes of interactions between environmental factors and genetic traits have the capacity to modulate disease risk. These must, therefore, be carefully evaluated as they have the potential to offer novel insights on the "missing heritability problem", reaching beyond our current limitations. First, we review a definition of G × E interactions. We then explore how concepts such as the early manifestation of the genetic components of a disease, the heterogeneity of complex traits, the clear definition of epidemiological strata, and the effect of varying physiological conditions can affect our capacity to detect (or miss) G × E interactions. Lastly, we discuss the shortfalls of regression models to study G × E interactions and how other methods such as the ReliefF algorithm, pattern recognition methods, or the LASSO (Least Absolute Shrinkage and Selection Operator) method can enable us to more adequately model G × E interactions. Overall, we present the elements to consider and a path to follow when studying genetic determinants of disease in order to uncover potential G × E interactions.

Developmental Coordination Disorder, An Umbrella Term for Motor Impairments in Children: Nature and Co-Morbid Disorders

Background: Developmental Coordination Disorder (DCD) defines a heterogeneous class of children exhibiting marked impairment in motor coordination as a general group of deficits in fine and gross motricity (subtype mixed group) common to all research studies, and with a variety of other motor disorders that have been little investigated. No consensus about symptoms and etiology has been established.

Methods: Data from 58 children aged 6 to 13 years with DCD were collected on DSM-IV criteria, similar to DSM-5 criteria. They had no other medical condition and inclusion criteria were strict (born full-term, no medication, no occupational/physical therapy). Multivariate statistical methods were used to evidence relevant interactions between discriminant features in a general DCD subtype group and to highlight specific co-morbidities. The study examined age-calibrated standardized scores from completed assessments of psychological, neuropsychological, and neuropsychomotor functions, and more specifically the presence of minor neurological dysfunctions (MND) including neurological soft signs (NSS), without evidence of focal neurological brain involvement. These were not considered in most previous studies.

Results: Findings show the salient DCD markers for the mixed subtype (imitation of gestures, digital perception, digital praxia, manual dexterity, upper, and lower limb coordination), vs. surprising co-morbidities, with 33% of MND with mild spasticity from phasic stretch reflex (PSR), not associated with the above impairments but rather with sitting tone (p = 0.004) and dysdiadochokinesia (p = 0.011). PSR was not specific to a DCD subtype but was related to increased impairment of coordination between upper and lower limbs and manual dexterity. Our results highlight the major contribution of an extensive neuro-developmental assessment (mental and physical).

Discussion: The present study provides important new evidence in favor of a complete physical neuropsychomotor assessment, including neuromuscular tone examination, using appropriate standardized neurodevelopmental tools (common tasks across ages with age-related normative data) in order to distinguish motor impairments gathered under the umbrella term of developmental coordination disorders (subcortical vs. cortical). Mild spasticity in the gastrocnemius muscles, such as phasic stretch reflex (PSR), suggests disturbances of the motor pathway, increasing impairment of gross and fine motricity. These findings contribute to understanding the nature of motor disorders in DCD by taking account of possible co-morbidities (corticospinal tract disturbances) to improve diagnosis and adapt treatment programmes in clinical practice.

LBoost: A boosting algorithm with application for epistasis discovery

Many human diseases are attributable to complex interactions among genetic and environmental factors. Statistical tools capable of modeling such complex interactions are necessary to improve identification of genetic factors that increase a patient's risk of disease. Logic Forest (LF), a bagging ensemble algorithm based on logic regression (LR), is able to discover interactions among binary variables predictive of response such as the biologic interactions that predispose individuals to disease. However, LF's ability to recover interactions degrades for more infrequently occurring interactions. A rare genetic interaction may occur if, for example, the interaction increases disease risk in a patient subpopulation that represents only a small proportion of the overall patient population. We present an alternative ensemble adaptation of LR based on boosting rather than bagging called LBoost. We compare the ability of LBoost and LF to identify variable interactions in simulation studies. Results indicate that LBoost is superior to LF for identifying genetic interactions associated with disease that are infrequent in the population. We apply LBoost to a subset of single nucleotide polymorphisms on the PRDX genes from the Cancer Genetic Markers of Susceptibility Breast Cancer Scan to investigate genetic risk for breast cancer. LBoost is publicly available on CRAN as part of the LogicForest package, http://cran.r-project.org/.

Patterns and processes in crop domestication: an historical review and quantitative analysis of 203 global food crops

Domesticated food crops are derived from a phylogenetically diverse assemblage of wild ancestors through artificial selection for different traits. Our understanding of domestication, however, is based upon a subset of well-studied 'model' crops, many of them from the Poaceae family. Here, we investigate domestication traits and theories using a broader range of crops. We reviewed domestication information (e.g. center of domestication, plant traits, wild ancestors, domestication dates, domestication traits, early and current uses) for 203 major and minor food crops. Compiled data were used to test classic and contemporary theories in crop domestication. Many typical features of domestication associated with model crops, including changes in ploidy level, loss of shattering, multiple origins, and domestication outside the native range, are less common within this broader dataset. In addition, there are strong spatial and temporal trends in our dataset. The overall time required to domesticate a species has decreased since the earliest domestication events. The frequencies of some domestication syndrome traits (e.g. nonshattering) have decreased over time, while others (e.g. changes to secondary metabolites) have increased. We discuss the influences of the ecological, evolutionary, cultural and technological factors that make domestication a dynamic and ongoing process.

Refining developmental coordination disorder subtyping with multivariate statistical methods

BACKGROUND

With a large number of potentially relevant clinical indicators penalization and ensemble learning methods are thought to provide better predictive performance than usual linear predictors. However, little is known about how they perform in clinical studies where few cases are available. We used Random Forests and Partial Least Squares Discriminant Analysis to select the most salient impairments in Developmental Coordination Disorder (DCD) and assess patients similarity.

METHODS

We considered a wide-range testing battery for various neuropsychological and visuo-motor impairments which aimed at characterizing subtypes of DCD in a sample of 63 children. Classifiers were optimized on a training sample, and they were used subsequently to rank the 49 items according to a permuted measure of variable importance. In addition, subtyping consistency was assessed with cluster analysis on the training sample. Clustering fitness and predictive accuracy were evaluated on the validation sample.

RESULTS

Both classifiers yielded a relevant subset of items impairments that altogether accounted for a sharp discrimination between three DCD subtypes: ideomotor, visual-spatial and constructional, and mixt dyspraxia. The main impairments that were found to characterize the three subtypes were: digital perception, imitations of gestures, digital praxia, lego blocks, visual spatial structuration, visual motor integration, coordination between upper and lower limbs. Classification accuracy was above 90% for all classifiers, and clustering fitness was found to be satisfactory.

CONCLUSIONS

Random Forests and Partial Least Squares Discriminant Analysis are useful tools to extract salient features from a large pool of correlated binary predictors, but also provide a way to assess individuals proximities in a reduced factor space. Less than 15 neuro-visual, neuro-psychomotor and neuro-psychological tests might be required to provide a sensitive and specific diagnostic of DCD on this particular sample, and isolated markers might be used to refine our understanding of DCD in future studies.

Systems biology data analysis methodology in pharmacogenomics

Pharmacogenetics aims to elucidate the genetic factors underlying the individual's response to pharmacotherapy. Coupled with the recent (and ongoing) progress in high-throughput genotyping, sequencing and other genomic technologies, pharmacogenetics is rapidly transforming into pharmacogenomics, while pursuing the primary goals of identifying and studying the genetic contribution to drug therapy response and adverse effects, and existing drug characterization and new drug discovery. Accomplishment of both of these goals hinges on gaining a better understanding of the underlying biological systems; however, reverse-engineering biological system models from the massive datasets generated by the large-scale genetic epidemiology studies presents a formidable data analysis challenge. In this article, we review the recent progress made in developing such data analysis methodology within the paradigm of systems biology research that broadly aims to gain a 'holistic', or 'mechanistic' understanding of biological systems by attempting to capture the entirety of interactions between the components (genetic and otherwise) of the system.

Power of data mining methods to detect genetic associations and interactions

BACKGROUND

Genetic association studies, thus far, have focused on the analysis of individual main effects of SNP markers. Nonetheless, there is a clear need for modeling epistasis or gene-gene interactions to better understand the biologic basis of existing associations. Tree-based methods have been widely studied as tools for building prediction models based on complex variable interactions. An understanding of the power of such methods for the discovery of genetic associations in the presence of complex interactions is of great importance. Here, we systematically evaluate the power of three leading algorithms: random forests (RF), Monte Carlo logic regression (MCLR), and multifactor dimensionality reduction (MDR).

METHODS

We use the algorithm-specific variable importance measures (VIMs) as statistics and employ permutation-based resampling to generate the null distribution and associated p values. The power of the three is assessed via simulation studies. Additionally, in a data analysis, we evaluate the associations between individual SNPs in pro-inflammatory and immunoregulatory genes and the risk of non-Hodgkin lymphoma.

RESULTS

The power of RF is highest in all simulation models, that of MCLR is similar to RF in half, and that of MDR is consistently the lowest.

CONCLUSIONS

Our study indicates that the power of RF VIMs is most reliable. However, in addition to tuning parameters, the power of RF is notably influenced by the type of variable (continuous vs. categorical) and the chosen VIM.

Travelling the world of gene-gene interactions

Over the last few years, main effect genetic association analysis has proven to be a successful tool to unravel genetic risk components to a variety of complex diseases. In the quest for disease susceptibility factors and the search for the 'missing heritability', supplementary and complementary efforts have been undertaken. These include the inclusion of several genetic inheritance assumptions in model development, the consideration of different sources of information, and the acknowledgement of disease underlying pathways of networks. The search for epistasis or gene-gene interaction effects on traits of interest is marked by an exponential growth, not only in terms of methodological development, but also in terms of practical applications, translation of statistical epistasis to biological epistasis and integration of omics information sources. The current popularity of the field, as well as its attraction to interdisciplinary teams, each making valuable contributions with sometimes rather unique viewpoints, renders it impossible to give an exhaustive review of to-date available approaches for epistasis screening. The purpose of this work is to give a perspective view on a selection of currently active analysis strategies and concerns in the context of epistasis detection, and to provide an eye to the future of gene-gene interaction analysis.