RESUMO
Given functional data from a survival process with time-dependent covariates, we derive a smooth convex representation for its nonparametric log-likelihood functional and obtain its functional gradient. From this we devise a generic gradient boosting procedure for estimating the hazard function nonparametrically. An illustrative implementation of the procedure using regression trees is described to show how to recover the unknown hazard. The generic estimator is consistent if the model is correctly specified; alternatively an oracle inequality can be demonstrated for tree-based models. To avoid overfitting, boosting employs several regularization devices. One of them is step-size restriction, but the rationale for this is somewhat mysterious from the viewpoint of consistency. Our work brings some clarity to this issue by revealing that step-size restriction is a mechanism for preventing the curvature of the risk from derailing convergence.
RESUMO
INTRODUCTION: Few studies of the demographics, natural history, and clinical management of inclusion body myositis (IBM) have been performed in a large patient population. To more accurately define these characteristics, we developed and distributed a questionnaire to patients with IBM. METHODS: A cross-sectional, self-reporting survey was conducted. RESULTS: The mean age of the 916 participants was 70.4 years, the male-to-female ratio was 2:1, and the majority reported difficulty with ambulation and activities of daily living. The earliest symptoms included impaired use and weakness of arms and legs. The mean time from first symptoms to diagnosis was 4.7 years. Half reported that IBM was their initial diagnosis. A composite functional index negatively associated with age and disease duration, and positively associated with participation in exercise. CONCLUSIONS: These data are valuable for informing patients how IBM manifestations are expected to impair daily living and indicate that self-reporting could be used to establish outcome measures in clinical trials.
Assuntos
Demografia , Miosite de Corpos de Inclusão/diagnóstico , Miosite de Corpos de Inclusão/epidemiologia , Atividades Cotidianas , Adulto , Distribuição por Idade , Idoso , Idoso de 80 Anos ou mais , Estudos Transversais , Pessoas com Deficiência , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Miosite de Corpos de Inclusão/complicações , América do Norte/epidemiologia , AutorrelatoRESUMO
The proliferation of medical monitoring devices makes it possible to track health vitals at high frequency, enabling the development of dynamic health risk scores that change with the underlying readings. Survival analysis, in particular hazard estimation, is well-suited to analyzing this stream of data to predict disease onset as a function of the time-varying vitals. This paper introduces the software package BoXHED (pronounced 'box-head') for nonparametrically estimating hazard functions via gradient boosting. BoXHED 1.0 is a novel tree-based implementation of the generic estimator proposed in Lee et al. (2017), which was designed for handling time-dependent covariates in a fully nonparametric manner. BoXHED is also the first publicly available software implementation for Lee et al. (2017). Applying it to a cardiovascular disease dataset from the Framingham Heart Study reveals novel interaction effects among known risk factors, potentially resolving an open question in clinical literature.
RESUMO
1) To show how to exploit the information contained in the trajectories of time-varying patient clinical data for dynamic predictions of mortality in the ICU; and 2) to demonstrate the additional predictive value that can be achieved by incorporating this trajectory information. DESIGN: Observational, retrospective study of patient medical records for training and testing of statistical learning models using different sets of predictor variables. SETTING: Medical ICU at the Yale-New Haven Hospital. SUBJECTS: Electronic health records of 3,763 patients admitted to the medical ICU between January 2013 and January 2015. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Six-hour mortality predictions for ICU patients were generated and updated every 6 hours by applying the random forest classifier to patient time series data from the prior 24 hours. The time series were processed in different ways to create two main models: 1) manual extraction of the summary statistics used in the literature (min/max/median/first/last/number of measurements) and 2) automated extraction of trajectory features using machine learning. Out-of-sample area under the receiver operating characteristics curve and area under the precision-recall curve ("precision" refers to positive predictive value and "recall" to sensitivity) were used to evaluate the predictive performance of the two models. For 6-hour prediction and updating, the second model achieved area under the receiver operating characteristics curve and area under the precision-recall curve of 0.905 (95% CI, 0.900-0.910) and 0.381 (95% CI, 0.368-0.394), respectively, which are statistically significantly higher than those achieved by the first model, with area under the receiver operating characteristics curve and area under the precision-recall curve of 0.896 (95% CI, 0.892-0.900) and 0.905 (95% CI, 0.353-0.379). The superiority of the second model held true for 12-hour prediction/updating as well as for 24-hour prediction/updating. CONCLUSIONS: We show that statistical learning techniques can be used to automatically extract all relevant shape features for use in predictive modeling. The approach requires no additional data and can potentially be used to improve any risk model that uses some form of trajectory information. In this single-center study, the shapes of the clinical data trajectories convey information about ICU mortality risk beyond what is already captured by the summary statistics currently used in the literature.
RESUMO
Background. Health savings accounts (HSAs) are tax-advantaged savings accounts available only to households with high-deductible health insurance. This article provides initial answers to two questions: 1) How should a household budget for its annual HSA contributions? 2) Do current contribution limits provide households with the flexibility to use HSAs efficiently? To answer these questions, we formulate the household's problem as one of determining a contribution strategy for minimizing total expected discounted medical costs. Methods. We use the 2002-2014 Medical Expenditure Panel Survey to develop a novel data-driven model for forecasting a household's health care costs based on its current cost percentile and other characteristics. A dynamic policy, in which the contribution each year brings the HSA balance up to a household-specific threshold, is derived. This is compared to a simpler static policy in which the target HSA balance is simply the plan's out-of-pocket maximum, with contributions in any year capped by a limit. Results. We find that: 1) the dynamic policy can save a household up to 19% in costs compared to the static one that is a proxy for typical contribution behavior; and 2) the recommended contribution amounts for 9% to 11% of households in a given year materially exceed what is currently allowed by the federal government. Conclusions. The dynamic policy derived from our data-analytic framework is able to unlock significant tax savings for health care consumers. To allow all households to use HSAs in a tax-efficient manner, a two-tiered contribution policy is needed: Allow unlimited contributions up to some balance, and then impose restrictions thereafter. The resulting impact on overall tax receipts is estimated to be well below what is currently allowed by legislation.
RESUMO
PURPOSE: End-of-life care for patients with advanced cancer is aggressive and costly. Oncologists inconsistently estimate life expectancy and address goals of care. Currently available prognostication tools are based on subjective clinical assessment. An objective prognostic tool could help oncologists and patients decide on a realistic plan for end-of-life care. We developed a predictive model (Imminent Mortality Predictor in Advanced Cancer [IMPAC]) for short-term mortality in hospitalized patients with advanced cancer. METHODS: Electronic health record data from 669 patients with advanced cancer who were discharged from Yale Cancer Center/Smilow Cancer Hospital were extracted. Statistical learning techniques were used to develop a tool to estimate survival probabilities. Patients were randomly split into training (70%) and validation (30%) sets 20 times. We tested the predictive properties of IMPAC for mortality at 30, 60, 90, and 180 days past the day of admission. RESULTS: For mortality within 90 days at a 40% sensitivity level, IMPAC has close to 60% positive predictive value. Patients estimated to have a greater than 50% chance of death within 90 days had a median survival time of 47 days. Patients estimated to have a less than 50% chance of death had a median survival of 290 days. Area under the receiver operating characteristic curve for IMPAC averaged greater than .70 for all time horizons tested. Estimated potential cost savings per patient was $15,413 (95% CI, $9,162 to $21,665) in 2014 constant dollars. CONCLUSION: IMPAC, a novel prognostic tool, can generate life expectancy probabilities in real time and support oncologists in counseling patients about end-of-life care. Potentially avoidable costs are significant.
Assuntos
Neoplasias/mortalidade , Neoplasias/patologia , Idoso , Custos e Análise de Custo , Registros Eletrônicos de Saúde , Feminino , Mortalidade Hospitalar , Humanos , Masculino , Pessoa de Meia-Idade , Neoplasias/epidemiologia , Neoplasias/terapia , Prognóstico , Curva ROC , Assistência Terminal , Fatores de TempoRESUMO
OBJECTIVE: To determine whether profit status is associated with differences in hospital days per patient, an outcome that may also be influenced by provider financial goals. DATA SOURCES: United States Renal Data System Standard Analysis Files and Centers for Medicare and Medicaid Services cost reports. DESIGN: We compared the number of hospital days per patient per year across for-profit and nonprofit dialysis facilities during 2003. To address possible referral bias in the assignment of patients to dialysis facilities, we used an instrumental variable regression method and adjusted for selected patient-specific factors, facility characteristics such as size and chain affiliation, as well as metrics of market competition. DATA EXTRACTION METHODS: All patients who received in-center hemodialysis at any time in 2003 and for whom Medicare was the primary payer were included (N=170,130; roughly two-thirds of the U.S. hemodialysis population). Patients dialyzed at hospital-based facilities and patients with no dialysis facilities within 30 miles of their residence were excluded. RESULTS: Overall, adjusted hospital days per patient were 17+/-5 percent lower in nonprofit facilities. The difference between nonprofit and for-profit facilities persisted with the correction for referral bias. There was no association between hospital days per patient per year and chain affiliation, but larger facilities had inferior outcomes (facilities with 73 or more patients had a 14+/-1.7 percent increase in hospital days relative to facilities with 35 or fewer patients). Differences in outcomes among for-profit and nonprofit facilities translated to 1,600 patient-years in hospital that could be averted each year if the hospital utilization rates in for-profit facilities were to decrease to the level of their nonprofit counterparts. CONCLUSIONS: Hospital days per patient-year were statistically and clinically significantly lower among nonprofit dialysis providers. These findings suggest that the indirect incentives in Medicare's current payment system may provide insufficient incentive for for-profit providers to achieve optimal patient outcomes.
Assuntos
Instituições Privadas de Saúde/organização & administração , Hospitalização/estatística & dados numéricos , Tempo de Internação/estatística & dados numéricos , Diálise Renal/estatística & dados numéricos , Instituições Filantrópicas de Saúde/organização & administração , Pesquisa sobre Serviços de Saúde , Humanos , Formulário de Reclamação de Seguro/estatística & dados numéricos , Marketing de Serviços de Saúde , Medicare , Afiliação Institucional , Avaliação de Resultados em Cuidados de Saúde , Propriedade , Encaminhamento e Consulta/estatística & dados numéricos , Análise de Regressão , Reembolso de Incentivo , Risco Ajustado , Viagem , Estados UnidosRESUMO
Transfusion therapy of liver transplant patients remains a challenge. High volumes of intraoperative blood transfusion have been shown to increase the risk of poor graft or patient survival. We conducted a retrospective study of 209 consecutive liver transplant cases at our institution. Only patients receiving their first liver transplant, with no other simultaneous organ transplants, were included. Cox proportional hazard modeling was used to identify clinical variables correlated with postoperative patient mortality. Statistically significant variables for poor patient survival were the number of red blood cell and plasma units transfused, a history of red blood cell alloantibodies, and the immunosuppressive regimen used. History of pregnancy also approached statistical significance but was less robust than the other 3 variables. Our findings suggest that blood transfusion and immune modulation greatly affect the survival of patients after liver transplantation.