Repeated measures random forests (RMRF): Identifying factors associated with nocturnal hypoglycemia.

Nocturnal hypoglycemia is a common phenomenon among patients with diabetes and can lead to a broad range of adverse events and complications. Identifying factors associated with hypoglycemia can improve glucose control and patient care. We propose a repeated measures random forest (RMRF) algorithm that can handle nonlinear relationships and interactions and the correlated responses from patients evaluated over several nights. Simulation results show that our proposed algorithm captures the informative variable more often than naïvely assuming independence. RMRF also outperforms standard random forest and extremely randomized trees algorithms. We demonstrate scenarios where RMRF attains greater prediction accuracy than generalized linear models. We apply the RMRF algorithm to analyze a diabetes study with 2524 nights from 127 patients with type 1 diabetes. We find that nocturnal hypoglycemia is associated with HbA1c, bedtime blood glucose (BG), insulin on board, time system activated, exercise intensity, and daytime hypoglycemia. The RMRF can accurately classify nights at high risk of nocturnal hypoglycemia.

Random forests of interaction trees for estimating individualized treatment effects in randomized trials.

Assessing heterogeneous treatment effects is a growing interest in advancing precision medicine. Individualized treatment effects (ITEs) play a critical role in such an endeavor. Concerning experimental data collected from randomized trials, we put forward a method, termed random forests of interaction trees (RFIT), for estimating ITE on the basis of interaction trees. To this end, we propose a smooth sigmoid surrogate method, as an alternative to greedy search, to speed up tree construction. The RFIT outperforms the "separate regression" approach in estimating ITE. Furthermore, standard errors for the estimated ITE via RFIT are obtained with the infinitesimal jackknife method. We assess and illustrate the use of RFIT via both simulation and the analysis of data from an acupuncture headache trial.

Covariate-modulated local false discovery rate for genome-wide association studies.

MOTIVATION: Genome-wide association studies (GWAS) have largely failed to identify most of the genetic basis of highly heritable diseases and complex traits. Recent work has suggested this could be because many genetic variants, each with individually small effects, compose their genetic architecture, limiting the power of GWAS, given currently obtainable sample sizes. In this scenario, Bonferroni-derived thresholds are severely underpowered to detect the vast majority of associations. Local false discovery rate (fdr) methods provide more power to detect non-null associations, but implicit assumptions about the exchangeability of single nucleotide polymorphisms (SNPs) limit their ability to discover non-null loci. METHODS: We propose a novel covariate-modulated local false discovery rate (cmfdr) that incorporates prior information about gene element-based functional annotations of SNPs, so that SNPs from categories enriched for non-null associations have a lower fdr for a given value of a test statistic than SNPs in unenriched categories. This readjustment of fdr based on functional annotations is achieved empirically by fitting a covariate-modulated parametric two-group mixture model. The proposed cmfdr methodology is applied to a large Crohn's disease GWAS. RESULTS: Use of cmfdr dramatically improves power, e.g. increasing the number of loci declared significant at the 0.05 fdr level by a factor of 5.4. We also demonstrate that SNPs were declared significant using cmfdr compared with usual fdr replicate in much higher numbers, while maintaining similar replication rates for a given fdr cutoff in de novo samples, using the eight Crohn's disease substudies as independent training and test datasets. Availability an implementation: https://sites.google.com/site/covmodfdr/ CONTACT: : wes.stat@gmail.com SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Frailty Modeling via the Empirical Bayes Hastings Sampler.

Studies of ocular disease and analyses of time to disease onset are complicated by the correlation expected between the two eyes from a single patient. We overcome these statistical modeling challenges through a nonparametric Bayesian frailty model. While this model suggests itself as a natural one for such complex data structures, model fitting routines become overwhelmingly complicated and computationally intensive given the nonparametric form assumed for the frailty distribution and baseline hazard function. We consider empirical Bayesian methods to alleviate these difficulties through a routine that iterates between frequentist, data-driven estimation of the cumulative baseline hazard and Markov chain Monte Carlo estimation of the frailty and regression coefficients. We show both in theory and through simulation that this approach yields consistent estimators of the parameters of interest. We then apply the method to the short-wave automated perimetry (SWAP) data set to study risk factors of glaucomatous visual field deficits.

A Combinatorial Optimization Framework for Scoring Students in University Admissions.

BACKGROUND AND OBJECTIVES: Selecting applications for college admission is critical for university operation and development. This paper leverages machine learning techniques to support enrollment management teams through data-informed decision-making in this otherwise laborious admissions processing. RESEARCH DESIGN AND MEASURES: Two aspects of university admissions are considered. An ensemble learning approach, through the SuperLearner algorithm, is used to predict student show (yield) rate. The goal is to improve prediction accuracy to minimize over- or under-enrollment. A combinatorial optimization framework is proposed to weigh academic performance and experiential factors for ranking and selecting students for admission. This framework uses simulated annealing, and an efficacy study is presented to evaluate performance. RESULTS: The proposed framework is illustrated for selecting an incoming class by optimizing predicted graduation rate and by developing an eligibility index. Each example presents a selection process under potential academic performance and experiential factor targets a university may place on an admitted class. R code is provided for higher education researchers and practitioners to apply the proposed methods in their own settings.

Measurement error of dietary self-report in intervention trials.

Dietary intervention trials aim to change dietary patterns of individuals. Participating in such trials could impact dietary self-report in divergent ways: Dietary counseling and training on portion-size estimation could improve self-report accuracy; participant burden could increase systematic error. Such intervention-associated biases could complicate interpretation of trial results. The authors investigated intervention-associated biases in reported total carotenoid intake using data on 3,088 breast cancer survivors recruited between 1995 and 2000 and followed through 2006 in the Women's Healthy Eating and Living Study, a randomized intervention trial. Longitudinal data from 2 self-report methods (24-hour recalls and food frequency questionnaires) and a plasma carotenoid biomarker were collected. A flexible measurement error model was postulated. Parameters were estimated in a Bayesian framework by using Markov chain Monte Carlo methods. Results indicated that the validity (i.e., correlation with "true" intake) of both self-report methods was significantly higher during follow-up for intervention versus nonintervention participants (4-year validity estimates: intervention = 0.57 for food frequency questionnaires and 0.58 for 24-hour recalls; nonintervention = 0.42 for food frequency questionnaires and 0.48 for 24-hour recalls). However, within- and between-instrument error correlations during follow-up were higher among intervention participants, indicating an increase in systematic error. Diet interventions can impact measurement errors of dietary self-report. Appropriate statistical methods should be applied to examine intervention-associated biases when interpreting results of diet trials.

Visual field quality control in the Ocular Hypertension Treatment Study (OHTS).

OBJECTIVE: To report the impact of visual field quality control (QC) procedures on the rates of visual field unreliability, test parameter errors, and visual field defects attributed to testing artifacts in the Ocular Hypertension Treatment Study (OHTS). METHODS: OHTS technicians were certified for perimetry and were required to submit 2 sets of visual fields that met study criteria before testing study participants. The OHTS Visual Field Reading Center (VFRC) evaluated 46,777 visual fields completed by 1618 OHTS participants between February 1994 and December 2003. Visual field QC errors, rates of unreliability, and defects attributed to testing artifacts were assessed. The OHTS QC system addressed 3 areas of clinic performance: (1) test parameter errors, (2) patient data errors, and (3) shipment errors. A visual field was classified as unreliable if any of the reliability indices exceeded the 33% limit. Clinical sites were immediately contacted by the VFRC via fax, e-mail, and/or phone and instructed on how to prevent further testing errors on fields with defects attributed to testing artifacts. MAIN OUTCOME MEASURES: QC errors (test parameter errors) and unreliability rates. RESULTS: A total of 2.4% (1136/ 46,777) of the visual fields were unreliable and 0.23% (107/46,777) had incorrect test parameters. Visual field defects attributed to testing artifacts occurred in approximately 1% (483/46,777) of the visual fields. CONCLUSIONS: Prompt transmission of visual fields to the VFRC for ongoing and intensive QC monitoring and rapid feedback to technicians helps to reduce the frequency of unreliable visual fields and incorrect testing parameters. Visual field defects attributed to testing artifacts were infrequent in the OHTS.

Complications of facial suspension sutures.

BACKGROUND: Barbed suture suspension has gained popularity as an inexpensive, "non-surgical" facial rejuvenation procedure. Initial studies on this procedure have reported minimal complications. OBJECTIVE: In this article, we review and evaluate several cases involving complications resulting from barbed suture suspension. METHODS: Four case histories of patients who required removal of suspension sutures because of symptomatic suture, failure, settling, or infection are summarized. The suspension sutures involved were examined after removal and compared to suture withdrawn through bovine muscle. RESULTS: Two patients reported chronic foreign body sensation. One patient had facial weakness. One patient had obvious dimpling. One patient had localized infection resulting from suture placement. On removal, the barbs on the suspension sutures appeared to be reflected against the direction of lift. Additionally, some sutures were clearly mobile. Forced withdrawal of new suture through bovine muscle did not result in similar anteversion of suture barbs. This would suggest the barb orientation noted at surgery was not the result of the removal process. Identification can be difficult, as the suture in some cases is clear. Once the sutures had been identified, they were easily pulled out without the need for additional surgical access. CONCLUSIONS: Suture suspension may gradually fail due to eversion of suture barbs or migration through tissues. Although the complication rate for this technique has been reported to be low, we believe that surgeons performing this procedure should be aware the rate may be higher than indicated in the literature.

Asymmetries and visual field summaries as predictors of glaucoma in the ocular hypertension treatment study.

PURPOSE: To evaluate whether baseline visual field data and asymmetries between eyes predict the onset of primary open-angle glaucoma (POAG) in Ocular Hypertension Treatment Study (OHTS) participants. METHODS: A new index, mean prognosis (MP), was designed for optimal combination of visual field thresholds, to discriminate between eyes that developed POAG from eyes that did not. Baseline intraocular pressure (IOP) in fellow eyes was used to construct measures of IOP asymmetry. Age-adjusted baseline thresholds were used to develop indicators of visual field asymmetry and summary measures of visual field defects. Marginal multivariate failure time models were constructed that relate the new index MP, IOP asymmetry, and visual field asymmetry to POAG onset for OHTS participants. RESULTS: The marginal multivariate failure time analysis showed that the MP index is significantly related to POAG onset (P < 0.0001) and appears to be a more highly significant predictor of POAG onset than either mean deviation (MD; P = 0.17) or pattern standard deviation (PSD; P = 0.046). A 1-mm Hg increase in IOP asymmetry between fellow eyes is associated with a 17% increase in risk for development of POAG. When threshold asymmetry between eyes existed, the eye with lower thresholds was at a 37% greater risk of development of POAG, and this feature was more predictive of POAG onset than the visual field index MD, though not as strong a predictor as PSD. CONCLUSIONS: The MP index, IOP asymmetry, and binocular test point asymmetry can assist in clinical evaluation of eyes at risk of development of POAG.

The association between glaucomatous visual fields and optic nerve head features in the Ocular Hypertension Treatment Study.

PURPOSE: To determine the association between change from baseline in the optic nerve head (ONH) and the visual field (VF) during follow-up of ocular hypertension participants in the Ocular Hypertension Treatment Study. DESIGN: Longitudinal randomized clinical trial. PARTICIPANTS: One hundred sixty-eight eyes of 152 ocular hypertensive participants ages 40 to 80 years. METHODS: Cox hazard models were applied to 3232 eyes, which included 81 eyes that reached a study end point by developing a glaucomatous VF (VF end point) and 128 eyes that reached a study end point by developing an optic disc change (optic disc end point). MAIN OUTCOME MEASURES: Primary open-angle glaucoma end point as determined by changes in the VF or optic disc. RESULTS: Forty-one eyes reached an end point by both VF and optic disc criteria; 40 eyes reached only a VF end point, and 87 reached only an optic disc end point. Times to reach isolated disc or field end points were similar. Visual field end points were more likely (P<0.0001) in eyes that showed the following ONH features: an ONH hemorrhage, thinning of the optic disc rim, or enlargement of the horizontal cup-to-disc (C/D) ratio. Optic disc end points were more likely (P<0.0001) in eyes that showed the following VF features: some evidence of a nasal step or a partial arcuate VF defect, or an increase in the pattern standard deviation (PSD). CONCLUSIONS: Both the VF and the optic disc must be monitored with equal diligence, because either may show the first evidence of glaucomatous damage. Changes in the ONH based on stereophotographic observation (rim thinning, hemorrhage, or a slight increase in C/D ratio) and VF changes (evidence of a nasal step/partial arcuate defect or an increase in PSD) suggest that these cases have an increased risk of developing glaucoma. Confirmation of such subtle findings should be sought through repeat testing and correlation with other clinical results.

Normal visual field test results following glaucomatous visual field end points in the Ocular Hypertension Treatment Study.

OBJECTIVE: To compare the occurrence of normal visual field (VF) test results following 2 vs 3 consecutive, abnormal, reliable test results in the Ocular Hypertension Treatment Study. METHODS: The Ocular Hypertension Treatment Study is a prospective, multicenter follow-up study as part of a longitudinal randomized clinical trial. Sixty-four (68 eyes) of 1636 participants developed a VF primary open-angle glaucoma (POAG) end point. We compared the proportion of normal VF test results after a VF POAG end point among eyes whose VF abnormality was confirmed by 2 (n = 9 eyes) vs 3 (n = 59 eyes) consecutive, abnormal, reliable VF test results. RESULTS: The proportion of VF test results that were normal subsequent to a VF POAG end point in eyes whose abnormality was confirmed by 2 consecutive, abnormal, reliable test results was significantly higher (73 [66%] of 110) compared with eyes whose abnormality was confirmed by 3 consecutive, abnormal, reliable test results (46 [12%] of 381) (P = .01). CONCLUSIONS: A VF POAG end point confirmed by 3 consecutive, abnormal, reliable VF test results appears to have greater specificity and stability than either 1 or 2 consecutive, abnormal, reliable VF test results. However, some eyes whose VF POAG end point was confirmed by 3 consecutive, abnormal, reliable test results nonetheless had 1 or more normal test results on follow-up.

Sensitivity, specificity and other diagnostic measures with multiple sites per unit.

Medical researchers often study the performance of diagnostic tools. The performance of the diagnostic tool is evaluated by calculating sensitivity, specificity and similar measures. Sensitivity is defined as the probability of a positive diagnostic test in a patient with the illness or injury for which the test serves as a diagnostic tool. Specificity is the probability of a negative diagnostic test in a patient free of the disease or injury. For some conditions, the test is applied to several anatomic locations or at different times in a patient and each test can be positive or negative. The findings of the tests performed on the same patient are correlated and any summary measure should account for this correlation. We propose a summary measure for the subject that combines the results from all tests performed on the same patient while accounting for the correlation. The measure is easily calculated and requires no specific statistical software. We also provide an estimate of the standard error and confidence intervals. The method is illustrated with a study, conducted at the University of California, Davis School of Medicine, of the sensitivity of magnetic resonance (MR) imaging and plain film (PF) radiography in detecting injuries in the cervical spine, such as fractures and soft tissue injuries.

Penetrance of the fragile X-associated tremor/ataxia syndrome in a premutation carrier population.

CONTEXT: Premutation expansions (55-200 CGG repeats) of the fragile X mental retardation 1 (FMR1) gene are frequent in the general population, with estimated prevalences of 1 per 259 females and 1 per 813 males. Several articles have recently described the presence of late-onset neurological symptoms in male carriers of premutation (FMR1) alleles. The main clinical features described in this newly identified syndrome are cerebellar ataxia and intention tremor. Additional documented symptoms include short-term memory loss, executive functional deficits, cognitive decline, parkinsonism, peripheral neuropathy, lower-limb proximal muscle weakness, and autonomic dysfunction. OBJECTIVE: To study the penetrance of the fragile X-associated tremor/ataxia syndrome (FXTAS) among premutation carriers. DESIGN, SETTING, AND PARTICIPANTS: Family-based study of 192 individuals (premutation carriers and controls) whose families belong to the Northern or Southern California Fragile X Associations. Data were collected (March 2002-April 2003) through a survey and a standardized neurological examination, which was videotaped and subsequently scored in a blinded fashion. MAIN OUTCOME MEASURES: Penetrance of intention tremor and ataxia among adult carriers (aged > or =50 years) of premutation expansions of the FMR1 gene. RESULTS: Data from the survey of 192 individuals demonstrated an age-related penetrance of the combination of reported intention tremor and gait ataxia in male carriers (17%, 38%, 47%, and 75% [lower-bound estimates] for participants aged 50-59, 60-69, 70-79, and > or =80 years, respectively). The male carrier group had an age-adjusted 13-fold increased risk (95% confidence interval, 3.9-25.4; P =.003) of combined intention tremor and gait ataxia when compared with male controls. The clinical examination data from 93 individuals demonstrated that male carriers experienced more difficulties on each of 3 standardized neurological rating scales compared with controls (P<.05). Female carrier scores were also higher than those of female controls (P<.05) on 2 of the 3 neurological rating scales, but no participant was identified with probable or definite FXTAS. CONCLUSIONS: The study demonstrates that older male carriers of premutation alleles of the FMR1 gene are at high risk of developing FXTAS. Since male premutation carriers are relatively common in the general population, older men with ataxia and intention tremor should be screened for the FMR1 mutation, especially if these signs are accompanied by parkinsonism, autonomic dysfunction, or cognitive decline, regardless of family history.

Predicting progressive glaucomatous optic neuropathy using baseline standard automated perimetry data.

PURPOSE: To test the hypothesis that specific locations and patterns of threshold findings within the visual field have predictive value for progressive glaucomatous optic neuropathy (pGON). METHODS: Age-adjusted standard automated perimetry thresholds, along with other clinical variables gathered at the initial examination of 168 individuals with high-risk ocular hypertension or early glaucoma, were used as predictors in a classification tree model. The classification variable was a determination of pGON, based on longitudinally gathered stereo optic nerve head photographs. Only data for the worse eye of each individual were included. Data from 100 normal subjects were used to test the specificity of the models. RESULTS: Classification tree models suggest that patterns of baseline visual field findings are predictive of pGON with sensitivity 65% and specificity 87% on average. Average specificity when data from normal subjects were run on the models was 69%. CONCLUSIONS: Classification trees can be used to determine which visual field locations are most predictive of poorer prognosis for pGON. Spatial patterns within the visual field convey useable predictive information, in most cases when thresholds are still well within the classically defined normal range.

Comparing SF-36 scores across three groups of women with different health profiles.

BACKGROUND: The widespread use of the Medical Outcomes Study (MOS) 36-item Short-Form Health Survey (SF-36) facilitates the comparison of health-related quality of life (HRQL) across independent studies. OBJECTIVES: To compare the scores of eight scales and two summary scales of the SF-36 across participants in the Women's Healthy Eating and Living (WHEL) trial, the Women's Health Initiative-Dietary Modification trial (WHI-DM), and the MOS, and to illustrate the use of effect sizes for interpreting the importance of group differences. METHODS: WHEL and WHI-DM are both multi-center dietary interventions; only data from the UC Davis sites were used in our study. WHEL participants had a recent history of breast cancer, WHI-DM participants were healthy, postmenopausal women, and women in the MOS had a history of hypertension, diabetes, heart disease, or depression. General linear models were used to identify statistically significant differences in scale scores. Meaningful differences were determined by effect sizes computed using a common within-group standard deviation (SD) and SDs from normative data. RESULTS: After adjusting for age and marital status, SF-36 scores for the WHI-DM and WHEL samples were similar and both had statistically significantly higher scores than the MOS sample. Relative to the WHEL or WHI-DM studies, MOS scores for scales related to the physical domain were clearly meaningfully lower whereas scale scores related to the mental health domain were potentially meaningfully lower. CONCLUSIONS: The HRQL of breast cancer survivors is comparable to that of healthy women and better than that of women with chronic health conditions, particularly with respect to physical health. This study illustrated the use of ranges of effects sizes for aiding the interpretation of SF-36 scores differences across independent studies.

Adult facial growth: applications to aesthetic surgery.

BACKGROUND: Reshaping of the face with age is a result of volume change and loss of support. It is not well understood which tissues are involved in this process. Recent publications suggest that adult bone growth may have a significant role. OBJECTIVE: We report a longitudinal cephalometric analysis of midfacial growth in adults to determine the role of bone in facial aging. METHODS: The Behrents modification of the Bolton Cephalometric study in patients up to age 83 was reviewed. A trigonometric analysis targeted orbital and anterior maxillary growth. RESULTS: Facial bone growth is shown to continue throughout adulthood. Anterior descent creates increased bone projection. CONCLUSIONS: The appearance of facial aging is caused by attrition of soft tissue volume and loss of support. The deficiency of maxillary bone projection seen in some patients, with tear trough depression and negative vector eyelid (polar bear), preexists adulthood and is unmasked with age.