Bagging cross-validated bandwidths with application to big data

Hall & Robinson (2009) proposed and analysed the use of bagged cross-validation to choose the bandwidth of a kernel density estimator. They established that bagging greatly reduces the noise inherent in ordinary cross-validation, and hence leads to a more efficient bandwidth selector. The asymptotic theory of Hall & Robinson (2009) assumes that $N$, the number of bagged subsamples, is $\infty$. We expand upon their theoretical results by allowing $N$ to be finite, as it is in practice. Our results indicate an important difference in the rate of convergence of the bagged cross-validation bandwidth for the cases $N=\infty$ and $N<\infty$. Simulations quantify the improvement in statistical efficiency and computational speed that can result from using bagged cross-validation as opposed to a binned implementation of ordinary cross-validation. The performance of the bagged bandwidth is also illustrated on a real, very large, dataset. Finally, a byproduct of our study is the correction of errors appearing in the Hall & Robinson (2009) expression for the asymptotic mean squared error of the bagging selector.

Falsely incompatible B-cell flow cytometry crossmatch after pronase treatment: a case report

This report presents a falsely incompatible B cell crossmatch by flow cytometry in a lung transplant recipient. The patient was a 35-year-old Caucasian male with end-stage lung disease secondary to cystic fibrosis whose pretransplantation serologic workup did not disclose the presence of anti-HLA class II antibodies by single antigen bead testing. Unexpectedly, crossmatch of recipient sera with pronase-treated donor lymphocytes resulted in antibody binding to B cells only. The positive reactivity was reproducible in pronase-treated autologous B cells. Recipient sera did not react with nontreated donor or autologous lymphocytes. Herein, we describe our approach to this unexpected crossmatch result and consider the implications of false-positive crossmatch results on transplantation.

Skeletal muscle ryanodine receptor channels are activated by the fungal metabolite, gliotoxin

Interactions between the reactive disulfide fungal metabolite, gliotoxin (GTX), and rabbit skeletal ryanodine receptor (RyR) calcium release channels have been examined. RyRs in terminal cisternae vesicles formed a covalent complex with 100 microm (35)S-GTX, which was reversed by 1 mm dithiothreitol (DTT) or 1 mm glutathione. GTX (80-240 microm), added to either cytoplasmic (cis) or luminal (trans) solutions, increased the rate of Ca(2+) release from SR vesicles and the frequency of opening of single RyR channels in lipid bilayers. Channel activation was reversed upon addition of 2 mm DTT to the cis solution, showing that the activation was due to an oxidation reaction (2 mm DTT added to the cis solution in the absence of GTX did not affect RyR activity). Furthermore, RyRs were not activated by trans GTX if the cis chamber contained DTT, suggesting that GTX oxidized a site in or near the membrane. In contrast to cis DTT, 2 mm DTT in the trans solution increased RyR activity when added either alone or with 200 microm trans GTX. The results suggest that (i) GTX increases RyR channel activity by oxidizing cysteine residues that are close to the membrane and located on RyR, or associated proteins, and (ii) a disulfide bridge or nitrosothiol, accessible only from the luminal solution, normally suppresses RyR channel activity. Some of the actions of GTX in altering Ca(2+) homeostatsis might depend on its modification of RyR calcium channels.

Changes in the mechanisms involved in uterine contractions during pregnancy in guinea-pigs

1. The mechanisms involved in contraction in guinea-pig myometrium were compared at mid- and late pregnancy. Tension was recorded simultaneously with either membrane potential or cytoplasmic calcium ([Ca2+]i) in strips exposed briefly to prostaglandin F2alpha (PGF). 2. PGF-induced increases in tension were underpinned by action potentials followed by sustained depolarization and biphasic increases in [Ca2+]i at mid- (peak, 879 +/- 199 nM; sustained, 298 +/- 35 nM, n = 11) and late pregnancy (peak, 989 +/- 302 nM; sustained 178 +/- 33 nM, n = 8). 3. At mid- and late pregnancy, nifedipine (10-6 M) reduced (a) the PGF-induced increase in tension to 84 and 35 %, (b) the level attained during the depolarization by 2 and 12 mV and (c) the peak rise in [Ca2+]i to 42 and 17 %. The sustained rises in [Ca2+]i were resistant to nifedipine. 4. In Ca2+-free solution (containing 1 mM EGTA), PGF elicited an increase in tension that was 26 % of that in 2.5 mM Ca2+ and an increase in [Ca2+]i (24 % of the sustained level) at mid-pregnancy but no increase in tension or [Ca2+]i at term. 5. At both stages of pregnancy, PGF decreased the level of [Ca2+]i required to elicit increases in tension comparable to those evoked by high K+o. The slope of the tension-[Ca2+]i curves were steeper in mid- than in late pregnancy. 6. In conclusion, at mid-pregnancy, the contractile response of the guinea-pig myometrium to PGF involves Ca2+ influx through L-type voltage-operated Ca2+ channels (VOCCs) and by receptor-operated mechanisms, release of Ca2+ from intracellular stores, and an increase in the sensitivity of the contractile apparatus to Ca2+. At term the situation is different: a modest increase in the sensitivity of the contractile apparatus to Ca2+ persists and there is a major reliance on Ca2+ influx through VOCCs.

Nitric oxide activates or inhibits skeletal muscle ryanodine receptors depending on its concentration, membrane potential and ligand binding

We show that rabbit skeletal RyR channels in lipid bilayers can be activated or inhibited by NO, in a manner that depends on donor concentration, membrane potential and the presence of channel agonists. 10 microm S-nitroso-N-acetyl-penicillamine (SNAP) increased RyR activity at -40 mV within 15 sec of addition to the cis chamber, with a 2-fold increase in frequency of channel opening (F(o)). 10 microm SNAP did not alter activity at +40 mV and did not further activate RyRs previously activated by 2 mm cis ATP at +40 or -40 mV. In contrast to the increase in F(o) with 10 microm SNAP, 1 mm SNAP caused a 2-fold reduction in F(o) but a 1.5-fold increase in mean open time (T(o)) at -40 mV in the absence of ATP. 1 mm SNAP or 0.5 mm sodium nitroprusside (SNP) induced approximately 3-fold reductions in F(o) and T(o) at +40 or -40 mV when channels were activated by 2 mm cis ATP or in channels activated by 6.5 microm peptide A at -40 mV (peptide A corresponds to part of the II-III loop of the skeletal dihydropyridine receptor). Both SNAP-induced activation and SNAP/SNP-induced inhibition were reversed by 2 mm dithiothreitol. The results suggest that S-Nitrosylation or oxidation of at least three classes of protein thiols by NO each produced characteristic changes in RyR activity. We propose that, in vivo, initial release of NO activates RyRs, but stronger release increases [NO] and inhibits RyR activity and contraction.

Preventive foot care. A user-friendly system for patients and physicians

Our user-friendly foot skin and toenail grading system is simple to understand for both patients and physicians. Current medical practice dictates that primary care physicians deliver the most comprehensive care possible for their patients. This includes preventive care and documentation of outcomes. Our approach simplifies evaluation and management of the majority of foot skin and toenail conditions in compromised hosts. Reimbursement for the extra care is justified, and authorized billing codes exist. Our system helps to prevent the physical, emotional, and financial costs associated with severe foot wounds and nail disorders. We remind our patients that proper foot care is as important in preventing foot complications as drug therapy is in preventing complications in other organs. Our approach to foot skin and toenail care is largely prophylactic and can be easily instituted in a primary care practice.

The calibration of gas volume measuring devices at continuous and pulsatile flows

A gas circuit that was capable of passing continuous or pulsatile flows via a 350 L Collins chain-compensated gasometer was built and evaluated. Various turbine volume transducers and dry gas meters were tested with gas compositions and flows that mimicked: a) inspired pulsatile flow over the physiological range and, b) mixed expirate being withdrawn from a Douglas bag. We found the Collins gasometer to be very accurate throughout its elevation, but its mixing fan is not required and atmospheric air should be left to saturate and the added water vapour calculated. Dry gas meters can be accurate to within 1% when calibrated (60 to 150 L/min), but require at least 25 L to be passed through them. The Morgan Ventilometer is an extremely reproducible device (coefficient of variation 0-0.2%, n = 60), but an increase in calibration syringe rate will elevate the calibration factor and reduce the percentage accuracy (one unit increase in calibration factor reduces accuracy by 0.6 - 1.0%). The optimal calibration syringe rate appears to be 30 - 45 b/min. Entrainment through the attached respiratory tubing can also alter the validity of the Ventilometer's calibration procedure.

Precision and accuracy of Morgan ventilometers at continuous and sinusoidal flows

The P.K. Morgan Mark II Ventilometer comprises a turbine transducer and a control/readout unit, which contains either a VENTX5A or VENTX6A programme. This equipment is used to measure minute ventilation during oxygen consumption tests. The precision and accuracy of six of these devices (a VENTX5A with two turbines, and two VENTX6As each with two different turbines) were examined during continuous and pulsatile flows; accuracy was affected by the syringe rate during the calibration procedure. The reference standard was a 350 l Collins chain-compensated gasometer. At continuous flows, the two VENTX5A devices showed accuracies of 99.5-100.5% over the range 60-160 l.min-1 whereas those for the four VENTX6As were 99.3-102.1% over the range 60-200 l.min-1. The mean accuracies of sinusoidal flows for the VENTX5A ranged between 96.1-101.3% for minute volumes spanning 15-200 l.min-1; comparable data for the VENTX6As were 95.9-98.9%. These mean accuracies improved to 98.1-101.3% for the VENTX5A at measured minute volumes > 80 l.min-1. Omission of the two lowest tidal volumes (1 and 1.5 l) for the VENTX6As narrowed the range of accuracy to 96.9-98.5%. The coefficients of variation for five trials at each minute ventilation were 0-0.2%. We conclude that, while the P.K. Morgan Mark II Ventilometer has excellent precision, its accuracy is dependent on the calibration syringe rate, minute ventilation and program.

Evaluation of five small turbine-type respirometers used in adult anesthesia

OBJECTIVE

The purpose of this study was to determine the accuracy of five rotating vane or turbine-type respirometers--fdE Magtrak Respiratory Monitor, fdE Haloscale Standard Wright Respirometer, Micro-Medical Pocket Monitor II, and Ohmeda 5400 and 5410 Volume Monitors--which are marketed as suitable for measuring expired gas volume in anesthetized adult patients.

METHODS

The devices were tested for accuracy at simulated minute ventilations ranging from 2.4 to 15 L/min with three gas mixtures: air, air saturated with water vapor, and N2O/O2 (70:30).

RESULTS

In the usual range for adult minute ventilation during general anesthesia with intermittent positive-pressure ventilation (5-12 L/min), all devices were on average accurate to within 12%. At the lowest minute volume readings (2.4 L/min) all devices except the Pocket Monitor tended to have low readings. No significant differences in readings were introduced by either saturation with water vapor or introduction of 70% N2O, except with ventilation below 4 L/min with both Ohmeda respirometers.

CONCLUSIONS

All monitors showed reasonable clinical accuracy in the usual adult minute ventilation range during general anesthesia with intermittent positive-pressure ventilation (> 5 L/min) when used to measure air, water vapor-saturated air, or N2O/O2 mixtures.

The accuracy of dry gas meters at continuous and sinusoidal flows

Dry gas meters are used in physiology laboratories to measure minute ventilation. The accuracy of these meters must be known since an error, of for instance 5%, affects the subsequent calculations for oxygen consumption by the same amount. Two precalibrated DTM-325 dry gas meters (American Meter Co., Philadelphia, PA, USA) were therefore tested for accuracy against a 350 l Collins chain-compensated gasometer. Provided that at least 25 l was passed per measurement then: a. continuous flows (air saturated with water vapour) between 60 and 150 l.min-1 were measured with an error of < 1%; and b. sinusoidal flows (ambient air) between 8 and 100 l.min-1 were misread by < 1% and the error was still within 2% at 140 l.min-1. It may, therefore, be concluded that the two precalibrated dry gas meters studied are valid volume measuring devices.

The accuracy of dry gas meters at continuous and sinusoidal flows

Dry gas meters are used in physiology laboratories to measure minute ventilation. The accuracy of these meters must be known since an error, of for instance 5%, affects the subsequent calculations for oxygen consumption by the same amount. Two precalibrated DTM-325 dry gas meters (American Meter Co., Philadelphia, PA, USA) were therefore tested for accuracy against a 350 l Collins chain-compensated gasometer. Provided that at least 25 l was passed per measurement then: a. continuous flows (air saturated with water vapour) between 60 and 150 l.min-1 were measured with an error of < 1%; and b. sinusoidal flows (ambient air) between 8 and 100 l.min-1 were misread by < 1% and the error was still within 2% at 140 l.min-1. It may, therefore, be concluded that the two precalibrated dry gas meters studied are valid volume measuring devices.

Enhanced psychophysiological responses of type A coronary patients during type A-relevant imagery

This study used imagery instructions to examine cardiac patients' physiological responses to Type A-relevant situations. Twenty Type A and Type B patients, classified by the Jenkins Activity Survey (JAS), were presented both "Type A" and "Neutral" imagery scripts, followed by the administration of the Structured Interview (SI). Subjects rated each scene for vividness and their emotional reaction to the image. Heart period, pulse transit time, finger temperature, finger pulse volume, frontalis EMG, neck EMG, forearm flexor EMG, and forearm extensor EMG were monitored throughout the experiment. Results showed that (a) Type A's had greater heart period and neck EMG responses to Type A scenes compared to Neutral scenes, (b) Type A's had faster pulse transit times and lower finger temperatures than Type B's throughout the imagery task and the SI, (c) both Type A's and Type B's rated Type A scenes as more emotionally arousing than Neutral scenes, and (d) although the pattern of results was the same when subjects were classified by the SI, results were no longer statistically significant.

Predicting differential response to EMG biofeedback and relaxation training: the role of cognitive structure

Analyzed treatment outcome data for 102 headache patients who had been assigned randomly to receive either EMG biofeedback (N = 70) or relaxation training (N = 32). It was hypothesized that a measure of cognitive structure would differentially predict success at the two training tasks and that patients high in cognitive structure would derive more benefit from the more structured relaxation task and less benefit from the biofeedback task. This hypothesis was confirmed (p less than .01). The analysis also demonstrated that relaxation training was significantly more effective than biofeedback (p less than .05) and that mixed headache patients improved significantly less (p less than .05) than either migraine or muscle-contraction headache patients.

A comparison of frontal EMG biofeedback and neck EMG biofeedback in the treatment of muscle-contraction headache

EMG biofeedback from the frontal area (FFB) was compared to EMG biofeedback from the neck (NFB) in the treatment of chronic muscle-contraction headache. Both treatment groups (N = 10) evidenced significant decreases in reported headache activity, with the NFB group also significantly reducing medication consumption. An analysis of EMG changes suggested that subjects were able to produce large within-session changes in EMG activity during initial sessions, with the major effect of additional training being an increase in speed with which these changes occurred. In neither group, however, did changes in EMG activity correspond closely to changes in reported headache activity.

A cost containment success story

At Northwest Hospital in Seattle, cost containment is not a stopgap program but a principle of sound management. Since 1974, an aggressive cost containment effort has enabled the hospital to increase its earnings in the face of declining revenues and to reduce its rates two years in a row.