An online framework for survival analysis: reframing Cox proportional hazards model for large data sets and neural networks.

In many biomedical applications, outcome is measured as a "time-to-event" (e.g., disease progression or death). To assess the connection between features of a patient and this outcome, it is common to assume a proportional hazards model and fit a proportional hazards regression (or Cox regression). To fit this model, a log-concave objective function known as the "partial likelihood" is maximized. For moderate-sized data sets, an efficient Newton-Raphson algorithm that leverages the structure of the objective function can be employed. However, in large data sets this approach has two issues: (i) The computational tricks that leverage structure can also lead to computational instability; (ii) The objective function does not naturally decouple: Thus, if the data set does not fit in memory, the model can be computationally expensive to fit. This additionally means that the objective is not directly amenable to stochastic gradient-based optimization methods. To overcome these issues, we propose a simple, new framing of proportional hazards regression: This results in an objective function that is amenable to stochastic gradient descent. We show that this simple modification allows us to efficiently fit survival models with very large data sets. This also facilitates training complex, for example, neural-network-based, models with survival data.

Methods for Feature Selection in Down-Selection of Vaccine Regimens Based on Multivariate Immune Response Endpoints.

In clinical trials, it is often of interest to compare and order several candidate regimens based on multiple endpoints. For example, in HIV vaccine development, immune response profiles induced by vaccination are key for selecting vaccine regimens to advance to efficacy evaluation. Motivated by the need to rank and choose a few vaccine regimens based on their immunogenicity in phase I trials, Huang et al. (Biostatistics 18(2):230-243, 2017) proposed a ranking/filtering/selection algorithm that down-selects vaccine regimens to satisfy the superiority and non-redundancy criteria, based on multiple immune response endpoints. In practice, many candidate immune response endpoints can be correlated with each other. An important question that remains to be addressed is how to choose a parsimonious set of the available immune response endpoints to effectively compare regimens. In this paper, we propose novel algorithms for selecting immune response endpoints to be used in regimen down-selection, based on importance weights assigned to individual endpoints and their correlation structure. We show through extensive simulation studies that pre-selection of endpoints can substantially improve performance of the subsequent regimen down-selection process. The application of the proposed method is demonstrated using a real example in HIV vaccine research, although the methods are also applicable in general to clinical research for dimension reduction when comparing regimens based on multiple candidate endpoints.

Physical Activity in Adults With an Amputation as Assessed With a Self-Reported Exercise Vital Sign.

BACKGROUND: Physical activity (PA) is important for the prevention and treatment of numerous chronic medical conditions. Individuals with a limb amputation face unique challenges for staying physically active. There are few studies evaluating PA of civilians with amputation in the United States. OBJECTIVE: To evaluate self-reported PA in persons with an amputation in the outpatient setting using a standardized exercise vital sign (EVS) and correlate PA with demographic information, amputation characteristics, and disease burden. DESIGN: Cross-sectional observational study. SETTING: Outpatient rehabilitation clinic at a tertiary care institution. INTERVENTIONS: N/A. PARTICIPANTS: Two hundred twenty-nine patients with limb amputation. MAIN OUTCOME MEASUREMENTS: EVS (self-reported weekly participation in moderate to vigorous intensity exercise), disease burden using a modified Charlson Comorbidity Index (CCI), possession of a prosthetic limb, amputation level, time from amputation, body mass index (BMI), gender, race, and age. RESULTS: A total of 28.8% of patients with limb amputation self-reported exercising at or above 150 min/wk as recommended by the United States Department of Health and Human Services (HHS); 31.8% of patients with transfemoral amputations, 27.8% with transtibial amputations, and 36% with upper extremity amputations reported exercising the recommended amount. Those with a prosthesis exercised 0.91 h/wk more than those without a prosthesis (95% CI 0.01, 1.8, P = .047), and female patients exercised 1.09 h/wk less than male patients (95% confidence interval [CI] 1.69-0.49, P < .001). Increasing age (P = .045), CCI (P = .006), and BMI (P = .005) all had a small but significant correlation with lower EVS. There was no statistically significant correlation between EVS and amputation level, race, or time from amputation. CONCLUSIONS: Less than one-third of patients with an amputation meet HHS recommendations for aerobic exercise. Male patients, those with a prosthesis, lower CCI, lower BMI, and younger age reported higher PA rates. Assessing EVS can help clinicians to identify patients with amputation that are not sufficiently active and may benefit from PA counseling and prescription.

Brief Ultrasound-Aided Teaching to Improve the Accuracy and Confidence of Resident Musculoskeletal Palpation.

BACKGROUND: Multiple studies demonstrate poor musculoskeletal palpation accuracy of physiatry residents. With the growing use of ultrasound clinically, it could serve as a powerful educational tool. However, there are no published studies examining the efficacy of ultrasound-aided education at the bedside, where much learning takes place during postgraduate training. OBJECTIVE: To determine if brief ultrasound-aided teaching improves residents' accuracy and confidence of musculoskeletal palpation. DESIGN: Before/After Trial. SETTING: Academic Institution. PARTICIPANTS: Ten physical medicine and rehabilitation residents were voluntary participants. INTERVENTION: Each resident was given a survey to assess confidence in palpating the hook of the hamate and medial calcaneal tubercle. They then attempted to palpate and place a marker over these two structures in a model patient. Marker placement accuracy was verified by ultrasound. Faculty spent less than 1 minute per landmark, sonographically teaching its correct localization. The resident, after moving to a new model patient, was then re-tested on accurate marker placement for both landmarks. A repeat survey to assess confidence was administered. MAIN OUTCOME MEASUREMENTS: Preintervention and postintervention accuracy and confidence of correct localization of both landmarks. RESULTS: We used McNemar's and Wilcoxon's tests to determine whether the intervention improved the location accuracy and confidence of successful palpation, respectively. Confidence level of correct localization of two landmarks improved on average from 4.3 to 8.15 on a scale of 10 after intervention. Of 20 palpation attempts, 11 correctly placed the marker postintervention after an incorrect placement, 8 made no change, and one changed from correct to incorrect placement. The odds of improvement in accuracy was 11 (95% confidence interval [CI] 1.60-473.47, P < .01). Both tests showed significant improvements after the intervention at the significance level of .05. CONCLUSIONS: Brief, ultrasound-aided teaching increases accuracy and confidence of residents' musculoskeletal palpation.

Plasma metabolomics profiles suggest beneficial effects of a low-glycemic load dietary pattern on inflammation and energy metabolism.

BACKGROUND: Low-glycemic load dietary patterns, characterized by consumption of whole grains, legumes, fruits, and vegetables, are associated with reduced risk of several chronic diseases. METHODS: Using samples from a randomized, controlled, crossover feeding trial, we evaluated the effects on metabolic profiles of a low-glycemic whole-grain dietary pattern (WG) compared with a dietary pattern high in refined grains and added sugars (RG) for 28 d. LC-MS-based targeted metabolomics analysis was performed on fasting plasma samples from 80 healthy participants (n = 40 men, n = 40 women) aged 18-45 y. Linear mixed models were used to evaluate differences in response between diets for individual metabolites. Kyoto Encyclopedia of Genes and Genomes (KEGG)-defined pathways and 2 novel data-driven analyses were conducted to consider differences at the pathway level. RESULTS: There were 121 metabolites with detectable signal in >98% of all plasma samples. Eighteen metabolites were significantly different between diets at day 28 [false discovery rate (FDR) < 0.05]. Inositol, hydroxyphenylpyruvate, citrulline, ornithine, 13-hydroxyoctadecadienoic acid, glutamine, and oxaloacetate were higher after the WG diet than after the RG diet, whereas melatonin, betaine, creatine, acetylcholine, aspartate, hydroxyproline, methylhistidine, tryptophan, cystamine, carnitine, and trimethylamine were lower. Analyses using KEGG-defined pathways revealed statistically significant differences in tryptophan metabolism between diets, with kynurenine and melatonin positively associated with serum C-reactive protein concentrations. Novel data-driven methods at the metabolite and network levels found correlations among metabolites involved in branched-chain amino acid (BCAA) degradation, trimethylamine-N-oxide production, and ß oxidation of fatty acids (FDR < 0.1) that differed between diets, with more favorable metabolic profiles detected after the WG diet. Higher BCAAs and trimethylamine were positively associated with homeostasis model assessment-insulin resistance. CONCLUSIONS: These exploratory metabolomics results support beneficial effects of a low-glycemic load dietary pattern characterized by whole grains, legumes, fruits, and vegetables, compared with a diet high in refined grains and added sugars on inflammation and energy metabolism pathways. This trial was registered at clinicaltrials.gov as NCT00622661.