Normalization transformations have recently experienced a resurgence in popularity in the era of machine learning, particularly in data preprocessing. However, the classical methods that can be adapted to cross-validation are not always effective. We introduce Ordered Quantile (ORQ) normalization, a one-to-one transformation that is designed to consistently and effectively transform a vector of arbitrary distribution into a vector that follows a normal (Gaussian) distribution. In the absence of ties, ORQ normalization is guaranteed to produce normally distributed transformed data. Once trained, an ORQ transformation can be readily and effectively applied to new data. We compare the effectiveness of the ORQ technique with other popular normalization methods in a simulation study where the true data generating distributions are known. We find that ORQ normalization is the only method that works consistently and effectively, regardless of the underlying distribution. We also explore the use of repeated cross-validation to identify the best normalizing transformation when the true underlying distribution is unknown. We apply our technique and other normalization methods via the bestNormalize R package on a car pricing data set. We built bestNormalize to evaluate the normalization efficacy of many candidate transformations; the package is freely available via the Comprehensive R Archive Network.

Firearm injuries disproportionately affect young, male, non-White populations, causing substantial individual and societal burden. Annual costs for hospitalized firearm injuries have not been widely described, as most previous cost studies have focused on lifetime costs. We examined a nationally-representative database of hospitalizations in the US to estimate per-hospital and overall hospital costs for firearm injuries by intent, type of weapon, and payer source. We conducted a retrospective cohort study of all firearm injury hospitalizations in the National Inpatient Sample from 2003 through 2013. The National Inpatient Sample, maintained by the Healthcare Utilization Project, is a stratified and weighted national sample of more than 20% of all hospitals. All admissions for firearm injuries were identified through Ecodes, yielding a weighted total of 336,785 for the study period. Average annual per-patient and overall hospital costs were estimated using generalized linear modelling, controlling for patient and hospital variables. Costs by intent, firearm type, and payer sources were estimated. Annually from 2003 through 2013, 30,617 hospital admissions were for firearm injuries, for an annual rate of 10.1 admissions per 100,000 US population. More than 80% of hospitalizations were among individuals aged 15–44, and rates were nine times higher for males than females and nearly ten times higher for the Black than the White population. More than 60% of admissions were for assaults, and 70% of the injuries that had a known firearm type were from handguns. The average annual admission cost was $622 million. The highest per-admission costs were for injuries from assault weapons ($32,237 per admission) and for legal intervention ($33,462 per admission), but the highest total costs were for unspecific firearm type ($373 million) and assaults ($389 million). A quarter of firearm injury hospitalizations were among the uninsured, yielding average annual total costs of $155 million. Hospitals can project that government insurance will be the highest source for firearm injury reimbursement, and depending on healthcare access laws, that many of their firearm injury admissions will not be covered by insurance.

Extracorporeal membrane oxygenation (ECMO) has been increasingly studied as a life support modality, but it is unclear if its use has changed over time. Recent publication shows no significant trend in use of ECMO over time; however, this report does not include more recent data. We performed trend analysis to determine if and when the use of ECMO changed in the past decade. We identified hospitalizations (2000–2011) in the Nationwide Inpatient Sample during which ECMO was recorded. We used a segmented linear regression model to determine trend and to identify a temporal change point when rate of ECMO use increased. ECMO use gradually grew until 2007, at which time there was a dramatic increase in the rate (p = 0.0003). There was no difference in mortality after 2007 (p = 0.3374), but there was longer length of stay (p = 0.0001) and smaller percentage of women (p = 0.005). There has been a marked increase in ECMO use since 2007. As ECMO use becomes more common, further study regarding indications, cost-effectiveness, and outcomes is warranted to guide optimal use.

Diagnostic and treatment approaches for sarcoidosis have changed dramatically over the past decade. Yet, the most recent reports of trends in hospitalizations of sarcoidosis patients are over ten years old. The objectives of this study were to determine the incidence of sarcoidosis among hospitalized patients and to analyze recent trends and seasonality of hospitalizations in sarcoidosis patients.