Median Prediction Error Research Articles

Prophylactic norepinephrine infusion for postspinal anesthesia hypotension in patients undergoing cesarean section: A randomized, controlled, dose-finding trial.

Prophylactic norepinephrine infusion effectively lowers the incidence of postspinal anesthesia hypotension. The optimal prophylactic dose of norepinephrine remains undefined. The purpose of this study was to investigate the optimal prophylactic dose of norepinephrine to prevent postspinal anesthesia hypotension in patients undergoing cesarean section. Prospective, double-blind, randomized, controlled, dose-finding trial. Operating room from March 2020 to June 2020. Ninety nine primipara or multipara women 18 - 40 years of age, singleton pregnancy ≥ 37 weeks, American Society of Anesthesiologists' physical status 1 or 2, scheduled for elective cesarean section under spinal anesthesia were included in this study. Patients were randomly assigned into groups to receive normal saline (NS) or one of four different prophylactic doses (0.025 [NE25], 0.05 [NE50], 0.075 [NE75], and 0.1 [NE100] ug/kg/min) of norepinephrine. The primary end point was the incidence of postspinal anesthesia hypotension (systolic blood pressure [SBP] < 80% of baseline) within 15 min after spinal anesthesia. Secondary outcomes included the overall stability of SBP control versus baseline (median performance error [MDPE] and median absolute performance error [MDAPE]), the dose that would be effective in preventing postspinal anesthesia hypotension in 50% (effective dose, ED 50) and 90% (ED90) of patients, other adverse events (bradycardia, nausea, vomiting, hypertension, and the total additional bolus of norepinephrine and atropine), and neonatal outcomes (blood gas values and Apgar scores). The incidence of postspinal anesthesia hypotension in NS, NE25, NE50, NE75, and NE100 groups was 68.42% (13/19), 40.00% (8/20), 20.00% (4/20), 15.00% (3/20), and 10.00% (2/20), respectively. With increasing prophylactic doses of norepinephrine, the incidence of postspinal anesthesia hypotension decreased (p < 0.001), SBP was maintained closer to the baseline (MDPE, p < 0.001; MDAPE, p = 0.001), and the total additional bolus of norepinephrine decreased (p < 0.001). The ED50 and ED90 values of norepinephrine were 0.016 (95% CI: -0.014 - 0.033) and 0.088 (95% CI: 0.068 - 0.133) ug/kg/min, respectively. Other adverse effects, neonatal outcomes, and the total additional bolus of atropine did not differ among the five groups. A prophylactic dose of 0.05 or 0.075 μg/kg/min norepinephrine prevents postspinal anesthesia hypotension in patients undergoing cesarean section.

Refractive Outcomes and Anterior Chamber Depth after Cataract Surgery in Eyes with and without Previous Pars Plana Vitrectomy

ABSTRACT Purpose: To compare the differences in refractive outcome and anterior chamber depth (ACD) after phacoemulsification between eyes with and without previous pars plana vitrectomy (PPV). Materials and Methods: Patients who had significant cataracts after PPV were included in the study group, and patients with a matched axial length (AL) who had cataracts without PPV were selected as the control group. The performance of new generation intraocular lens (IOL) power calculation formulas (Barrett Universal II, Kane, Ladas Super formulas), and the traditional formulas (SRK/T, Holladay 1, Hoffer Q, Haigis) with and without the Wang–Koch (WK) AL adjustment were compared between the two groups. The postoperative ACD was measured using the Scheimpflug imaging system with manual correction at least three months after surgery. Results: In total, there were 193 eyes from 193 patients in each group. The mean prediction errors (MEs) of the new generation formulas had no significant systemic bias in the study group; the hyperopic shift was displayed in the traditional formulas for eyes with AL > 26mm. However, the difference of MEs between the two groups among all the formulas were not significant. The absolute prediction error (MAE) and median prediction error (Med AE) in the study group were larger than those in the control group among all the formulas. The postoperative ACD of the study group was deeper but not significant than that of the control group. Conclusions: There was no refractive shift in vitrectomized eyes compared with non-vitrectomized eyes no matter in new generation formulas or traditional vergence formulas. The prediction error among all the formulas in vitrectomized eyes were significantly higher than those in non–vitrectomized eyes. The ACD after phacoemulsification in vitrectomized eyes was not significantly different from non–vitrectomized eyes.

A Comparison of Intraocular Lens Power Calculation Formulas in High Myopia.

To comparatively evaluate the accuracy of newer intraocular lens (IOL) calculation formulas and common third-generation formulas after Wang-Koch adjustment in the prediction of postoperative refraction in highly myopic eyes. This was a retrospective study including eyes with high myopia that had uncomplicated cataract surgery with implantation of an AcrySof MA60MA IOL (power range: -5.00 to +5.00 diopters [D]) (Alcon Laboratories, Inc). All patients underwent optical biometry (Carl Zeiss IOLMaster 500 and IOLMaster 700, and Allegro Biograph) and the postoperative spherical equivalent for the implanted IOL was estimated using SRK/T, Holladay 1 (both Wang-Koch adjusted), Haigis, Barrett Universal II, Kane, Ladas, and Hill-RBF v2.0 formulas. Outcomes included the median absolute prediction error (MedAE) and the proportion of eyes within ±0.25, ±0.50, and ±1.00 D of the preoperative prediction. Eighty-two eyes with a mean axial length of 30.89 ± 1.85 mm were included. The MedAE in ascending order was Hill-RBF v2.0 0.31 D, Kane 0.33 D, Barrett 0.36 D, Holladay Iwk 0.37 D, SRK/Twk 0.37 D, Holladay Iwk 0.43 D, HaigisULIB 0.54 D, and Ladas 0.61 D. The formula with the lowest MedAE (Hill-RBF v2.0) yielded a prediction error within ±0.25, ±0.50, and ±1.00 D in 43.1%, 70.6%, and 94.1% of cases, respectively. Recent formulas such as Barrett Universal II, Kane, and Hill-RBF v2.0 and Wang-Koch adjusted formulas perform well in this subset of patients with high myopia. The Hill-RBF v2.0 formula had the lowest MedAE and highest proportion of eyes within ±0.25, ±0.50, and ±1.00 D of the predicted target. [J Refract Surg. 2021;37(3):207-211.].

Precision Dosing for Tacrolimus Using Genotypes and Clinical Factors in Kidney Transplant Recipients of European Ancestry.

Genetic variation in the CYP3A4 and CYP3A5 (CYP3A4/5) genes, which encode the key enzymes in tacrolimus metabolism, is associated with tacrolimus clearance and dose requirements. Tacrolimus has a narrow therapeutic index with high intra- and intersubject variability, in part because of genetic variation. High tacrolimus clearance and low trough concentration are associated with a greater risk for rejection, whereas high troughs are associated with calcineurin-induced toxicity. The objective of this study was to develop a model of tacrolimus clearance with a dosing equation accounting for genotypes and clinical factors in adult kidney transplant recipients of European ancestry that could preemptively guide dosing. Recipients receiving immediate-release tacrolimus for maintenance immunosuppression from 2 multicenter studies were included. Participants in the GEN03 study were used for tacrolimus model development (n = 608 recipients) and was validated by prediction performance in the DeKAF Genomics study (n = 1361 recipients). Nonlinear mixed-effects modeling was used to develop the apparent oral tacrolimus clearance (CL/F) model. CYP3A4/5 genotypes and clinical covariates were tested for their influence on CL/F. The predictive performance of the model was determined by assessing the bias (median prediction error [ME] and median percentage error [MPE]) and the precision (root median squared error [RMSE]) of the model. CYP3A5*3, CYP3A4*22, corticosteroids, calcium channel blocker and antiviral drug use, age, and diabetes significantly contributed to the interindividual variability of oral tacrolimus apparent clearance. The bias (ME, MPE) and precision (RMSE) of the final model was good, 0.49ng/mL, 6.5%, and 3.09ng/mL, respectively. Prospective testing of this equation is warranted.

Prediction accuracy of standard and total keratometry by swept-source optical biometer for multifocal intraocular lens power calculation

We aimed to compare the refractive outcomes of cataract surgery with diffractive multifocal intraocular lenses (IOLs) using standard keratometry (K) and total keratometry (TK). In this retrospective observational case series study, a total of 302 patients who underwent cataract surgery with multifocal IOL implantation were included. Predicted refractive outcomes were calculated based on the current standard formulas and a new formula developed for TK using K and TK, which were obtained from a swept-source optical biometer. At 2-month postoperatively, median absolute prediction errors (MedAEs) and proportion of eyes within ± 0.50 diopters (D) of predicted postoperative spherical equivalent (SE) refraction were analyzed. There was no significant difference between MedAEs or proportion of eyes within ± 0.50D of predicted refraction from K and TK in each formula. In TFNT00 and 839MP IOL cases, there was no difference between MedAEs from K and TK using any formula. In 829MP IOL cases, MedAE from TK was significantly larger than that from K in Barrett Universal II/Barrett TK Universal II (P = 0.033). In 677MY IOL cases, MedAE from TK was significantly larger than that from K in Haigis (P = 0.020) and Holladay 2 (P = 0.006) formulas. In the subgroup analysis for IOL, there was no difference between the proportion of eyes within ± 0.50 D of predicted refraction from K and TK using any formula. TFNT00 and 839MP IOLs were favorable with TK, with 677MY IOL with K and 829MP IOL being in a neutral position, which necessitates the study that investigates the accuracy of the new TK technology.

Project hyperopic power prediction: accuracy of 13 different concepts for intraocular lens calculation in short eyes

PurposeTo evaluate the accuracy of intraocular lens (IOL) power calculation in a patient cohort with short axial eye length to assess the performance of IOL power calculation schemes in strong...

A recurrent neural network using historical data to predict time series indoor PM2.5 concentrations for residential buildings

Due to the severe outdoor PM2.5 pollution in China, many people have installed air-cleaning systems in homes. To make the systems run automatically and intelligently, we developed a recurrent neural network (RNN) that uses historical data to predict the future indoor PM2.5 concentration. The RNN architecture includes an autoencoder and a recurrent part. We used data measured in an apartment over the course of an entire year to train and test the RNN. The data include indoor/outdoor PM2.5 concentration, environmental parameters and time of day. By comparing three different input strategies, we found that a strategy employing historical PM2.5 and time of day as inputs performed best. With this strategy, the model can be applied to predict the relatively stable trend of indoor PM2.5 concentration in advance. When the input length is 2h and the prediction horizon is 30min, the median prediction error is 8.3µg/m3 for the whole test set. For times with indoor PM2.5 concentrations between (20,50]µg/m3 and (50,100]µg/m3 , the median prediction error is 8.3 and 9.2µg/m3 , respectively. The low prediction error between the ground-truth and predicted values shows that the RNN can predict indoor PM2.5 concentrations with satisfactory performance.

Relationship between blood pressure stability and mortality in cardiac surgery patients: retrospective cohort study.

Performance measurement variables can be applied in clinical practice to evaluate hemodynamic instability. This study aimed to evaluate the relationship between the performance measurement of mean arterial pressure during cardiac surgery using cardiopulmonary bypass and postoperative mortality. A retrospective cohort study of patients who underwent cardiac surgery requiring cardiopulmonary bypass between 2013 and 2016 was conducted. The median performance error (MDPE) and median absolute performance error (MDAPE) were calculated using the preoperative mean arterial pressure as a reference, and intraoperative mean arterial pressures as measured values. Multivariable logistic regression analyses were performed using performance measurement variables to predict 30-day mortality. Overall survival according to performance measurement variables was evaluated using Cox proportional hazard models and Kaplan-Meier survival curves were generated to compare survival probability. Among 1203 patients, 110 (9.1%) died after surgery, and the 30-day mortality rate was 2.3% (28/1203). After adjusting for confounders, MDPE and MDAPE were significant mean arterial pressure derived predictors of 30-day mortality and overall survival. Intraoperative hypotension measured by performance measurement variables was independently associated with 30-day and overall mortality after cardiac surgery requiring cardiopulmonary bypass. Kaplan-Meier survival curves showed lower survival probability in patients with higher MDAPE during the pre- and post- cardiopulmonary bypass periods (P < 0.001 by log-rank test). Intraoperative hypotension measured by performance measurement variables was independently associated with 30-day and overall mortality after cardiac surgery requiring CPB. We propose that performance measurement variables are useful for quantifying the degree of intraoperative hypotension and predicting survival following cardiac surgery.Trial registration: ClinicalTrials.gov, identifier: NCT03785132.

Effect of Hemodilution on Fluctuation of Cerebral Oxygen Saturation during Cardiopulmonary Bypass in Children with Cyanotic Congenital Heart Disease

Background: In patients with cyanotic congenital heart disease (CHD), cerebral oxygenation may be maintained by elevations in hematocrit (Hct). Hemodilution accompanying cardiopulmonary bypass (CPB), however, can disrupt cerebral oxygen balance, leading to fluctuations in cerebral oxygen saturation (ScO2). The present study investigated the effects of Hct changes on the fluctuation of ScO2 during CPB in cyanotic CHD using performance measurement (PM). Methods: Children with CHD (51 acyanotic and 46 cyanotic) who had undergone cardiac surgery using CPB were enrolled. Median performance error (MDPE), median absolute performance error (MDAPE), and wobble parameters of ScO2 were calculated before (reference value), during, and after CPB. Correlations of PM parameters with Hct and reductions in Hct (ΔHct) were also evaluated. Results: Before CPB, patients with cyanotic CHD had lower MDPE and larger wobble than those with acyanotic CHD, although mean ScO2 did not differ significantly between the two groups. During CPB, ScO2 of acyanotic CHD increased as ΔHct increased, but PM variables were not associated with ΔHct. In cyanotic CHD, MDPE (r = −0.324, p = 0.032) and MDAPE (r = 0.339, p = 0.024) correlated significantly with ΔHct during CPB. After CPB, MDPE (r = 0.574, p = 0.025) and MDAPE (r = −0.543, p = 0.036) were significantly correlated with Hct in children with cyanotic CHD who underwent palliative surgery. Conclusion: Therefore, ScO2 fluctuation during CPB in children with cyanotic CHD may be affected by the decrease in Hct, suggesting that excessive hemodilution can negatively influence the maintenance of cerebral oxygenation in these patients.

Identifiability and physical interpretability of hybrid, gray-box models - a case study

Model identifiability concerns the uniqueness of uncertain model parameters to be estimated from available process data and is often thought of as a prerequisite for the physical interpretability of a model. Nevertheless, model identifiability may be challenging to obtain in practice due to both stochastic and deterministic uncertainties, e.g. low data variability, noisy measurements, erroneous model structure, and stochasticity and locality of the optimization algorithm. For gray-box, hybrid models, model identifiability is rarely obtainable due to a high number of parameters. We illustrate through an industrial case study - modeling of a production choke valve in a petroleum well - that physical interpretability may be preserved even for non-identifiable models with adequate parameter regularization in the estimation problem. To this end, in a real industrial scenario, it may be beneficial for the model's predictive performance to develop hybrid over mechanistic models, as the model flexibility is higher. Modeling of six petroleum wells on the asset Edvard Grieg using historical production data show a 35\% reduction in the median prediction error across the wells comparing a hybrid to a mechanistic model. On the other hand, both the predictive performance and physical interpretability of the developed models are influenced by the available data. The findings encourage research into online learning and other hybrid model variants to improve the results.

Predicting Pulmonary Distension in a Virtual Patient Model for Mechanical Ventilation

Abstract Recruitment maneuvers (RMs) following with positive-end-expiratory-pressure (PEEP) have proved effective in recruiting lung volume and preventing alveoli collapse. To date, standards for optimal patient-specific PEEP are unknown, resulting in variability in care and reduced outcomes, both indicating the need for personalized care. This research extends a well-validated virtual patient model by adding novel elements to model, which is able to utilize bedside available respiratory data, without increasing modelling complexity, to predict patient-specific lung distension and thus to minimise barotrauma risk. Prediction accuracy and robustness are validated against clinical data from 18 volume controlled ventilation (VCV) patients at 7 different baseline PEEP levels (0 to 12cmH2O), where predictions were made up to 12cmH2O of PEEP ahead. Using an exponential basis function set for prediction yields an absolute median peak inspiratory pressure prediction error of 1.50cmH2O for 623 prediction cases. Comparing predicted and clinically measured distension prediction in VCV demonstrated consistent, robust high accuracy with R2=0.90 (623 predictions), which is a measurable improvement in prediction error compared to predictions without using the proposed distension function (R2=0.82). Moreover, the R2 value increases to 0.93-0.95 if only clinically relevant ΔPEEP steps (2-6cmH2O) are considered with an overall median absolute error in peak pressure prediction of 1.04cmH2O. Overall, the results demonstrate the potential and significance for accurately capturing distension mechanics, allowing better risk assessment, as well as extending and more fully validating this virtual mechanical ventilation patient model.

PowerNet: a smart energy forecasting architecture based on neural networks

Electricity demand forecasting is a critical task for efficient, reliable and economical operation of the power grid, which is one of the most essential building blocks of smart cities. Accurate forecasting allows grid operators to properly maintain the balance of supply and demand as well as to optimize operational cost for generation and transmission. This article proposes a novel neural network architecture PowerNet which can incorporate multiple heterogeneous features such as historical energy consumption data, weather data and calendar information for the demand forecasting task. Using real-world smart meter dataset, we conduct an extensive evaluation to show the advantages of PowerNet over recently-proposed machine learning methods such as Gradient Boosting Tree (GBT), Support Vector Regression (SVR), Random Forest (RF) and Gated Recurrent Unit (GRU). PowerNet demonstrates notable performance in reducing both the median and worst-case prediction errors when forecasting demands of individual residential households. We further provide empirical results concerning the two operational considerations that are crucial when using PowerNet in practice: the time horizon the model can predict with a decent accuracy and the frequency of training the model to retain its modeling capability. Finally, we briefly discuss a multi-layer anomaly/electricity-theft detection approach based on PowerNet demand forecasting.

A Generalized Bioavailability Model (gBAM) for Predicting Chronic Copper Toxicity to Freshwater Fish

The generalized bioavailability model (gBAM) has been proposed as an alternative to the biotic ligand model (BLM) for modeling bioavailability and chronic toxicity of copper (Cu). The gBAM combines a log-linear effect of pH on free Cu2+ ion toxicity with BLM-type parameters for describing the protective effects of major cations (calcium [Ca]2+ , magnesium [Mg]2+ , and sodium [Na]+ ). In the present study, a Windermere Humic Aqueous Model (WHAM) VII-based gBAM for fish was parametrized based on an existing chronic (30-d) dataset of juvenile rainbow trout (Oncorhynchus mykiss). The model, with defined parameters (pH slope parameter [SpH ] = 0.4449 and biotic ligand competition constants [log KCaBL = 4.0, log KMgBL = 3.4, and log KNaBL = 3.0]), was shown to accurately predict the effects of pH, dissolved organic carbon, Ca, and Mg on chronic Cu toxicity to juvenile rainbow trout at the effect levels relevant for environmental risk assessment (i.e., median prediction error of 1.3-fold for 10 and 20% lethal concentrations). The gBAM predicted the effect of pH more accurately than a previously published Cu BLM for juvenile rainbow trout, especially at pH > 8. We also evaluated the cross-species and cross-life stage applicability of the newly developed juvenile rainbow trout gBAM using existing chronic Cu toxicity data with early life stages of fathead minnow (Pimephales promelas) and rainbow trout. We did this because using a single bioavailability model for all fish species and life stages is practical from a regulatory point of view. Although the early life stage datasets exhibit considerable uncertainties, 91% of the considered toxicity values at the effect levels most relevant in European environmental regulations (10% effect on survival or growth) were predicted within a 2-fold error. Overall, the chronic Cu gBAM we developed is a valuable alternative for the existing chronic Cu BLM for rainbow trout and performs sufficiently well to be used in risk assessment according to currently accepted standards of bioavailability model performance (from the current European regulatory point of view). However, our analysis also suggests that bioavailability relations differ between different fish life stages and between endpoints (e.g., mortality vs growth), which is currently not accounted for in environmental risk assessments. Environ Toxicol Chem 2020;39:2424-2436. © 2020 SETAC.

Comparison of Aphakic Refraction and Biometry-Based Formulae for Secondary In-The-Bag and Sulcus-Implanted Intraocular Lens Power Estimation in Children

Purpose: The aim of the study was to compare the accuracy of refractive outcomes in children undergoing secondary in-the-bag or cilliary sulcus intraocular lens (IOL) implantation, using aphakic refraction (AR)-based formulae (Hug and Khan) and biometry-based formulae (Holladay 1, Hoffer Q, SRK/T, and SRK II). Methods: In this retrospective study, a total of 65 eyes of 44 patients who underwent secondary in-the-bag or cilliary sulcus IOL implantation were included and divided into 2 groups: 39 eyes of the in-the-bag IOL group and the other 26 eyes of the sulcus-implanted IOL group. Holladay 1, Hoffer Q, SRK/T, and SRK II formulae were employed depending on the biometric data, while Hug and Khan formulae were used based on preoperative AR. The prediction error (PE) and the absolute value of predicted error (APE) were compared between the 2 groups and formulae. Results: In the in-the-bag IOL group, nonsignificant differences of APE were found among the 6 formulae, while the Holladay 1, Hoffer Q, SRK/T, and SRK II all demonstrated a significant hyperopic shift of median PE value compared to the Hug formula (p < 0.05, all), and Holladay 1 and SRK II also showed a significant hyperopic shift of PE compared to the Khan formula (p < 0.05, both). Higher percentages of eyes with PE <1 D were found using Hoffer Q and SRK/T. In the sulcus-implanted group, the Holladay 1, Hoffer Q, and SRK/T had a significantly smaller median value of APE than the Hug and Khan formulae (p < 0.05, all), and the SRK II had a significantly smaller median value of APE than the Hug formula (p < 0.05), while Holladay 1 had the lowest value of APE. Higher percentages of eyes within PE <1 D were found using Holladay 1, Hoffer Q, and SRK/T, while the highest one was SRK/T. Significantly larger hyperopic shifts of median PE value using all the 6 formulae were found in eyes with sulcus-implanted IOL than eyes with in-the-bag implanted IOL (p < 0.05, all). In the eyes of with in-the-bag implanted IOL, the Hug and Khan formulae had significantly smaller APE values when compared with the eyes with sulcus-implanted IOL (p < 0.05, both). Conclusions: Whether IOL was in the bag or implanted in the sulcus, almost all the formulae showed hyperopic shift, SRK/T showed the best accuracy. Biometry-based formulae were superior to AR-based formulae in accuracy of IOL power calculation, especially when IOL was implanted in the sulcus. In-the-bag IOL implantation should always be with higher priorities, especially when using AR-based formulae in IOL power calculation.

Marker-based crop model-assisted ideotype design to improve avoidance of abiotic stress in bread wheat.

Wheat phenology allows escape from seasonal abiotic stresses including frosts and high temperatures, the latter being forecast to increase with climate change. The use of marker-based crop models to identify ideotypes has been proposed to select genotypes adapted to specific weather and management conditions and anticipate climate change. In this study, a marker-based crop model for wheat phenology was calibrated and tested. Climate analysis of 30 years of historical weather data in 72 locations representing the main wheat production areas in France was performed. We carried out marker-based crop model simulations for 1019 wheat cultivars and three sowing dates, which allowed calculation of genotypic stress avoidance frequencies of frost and heat stress and identification of ideotypes. The phenology marker-based crop model allowed prediction of large genotypic variations for the beginning of stem elongation (GS30) and heading date (GS55). Prediction accuracy was assessed using untested genotypes and environments, and showed median genotype prediction errors of 8.5 and 4.2 days for GS30 and GS55, respectively. Climate analysis allowed the definition of a low risk period for each location based on the distribution of the last frost and first heat days. Clustering of locations showed three groups with contrasting levels of frost and heat risks. Marker-based crop model simulations showed the need to optimize the genotype depending on sowing date, particularly in high risk environments. An empirical validation of the approach showed that it holds good promises to improve frost and heat stress avoidance.

Changing practice patterns in European cataract surgery as reflected in the European Registry of Quality Outcomes for Cataract and Refractive Surgery 2008 to 2017.

To study practice patterns in European cataract surgery over a 10-year period. European clinics affiliated to the European Registry of Quality Outcomes for Cataract and Refractive Surgery (EUREQUO). Registry cohort study. The EUREQUO contains preoperative, intraoperative, and postoperative parameters reported by surgeons in many European clinics. All data reported to the registry are anonymized. Preoperative parameters included age, sex, visual acuity, target refraction, ocular comorbidity, and surgical difficulties. Surgical data included anesthesia, surgical technique, intraocular lens optic biomaterial, and complications. Postoperative parameters included visual acuity, refraction, and short-term complications. During the study period (January 1, 2008, to December 31, 2017), a total of 2 714 108 cataract extractions were reported to the EUREQUO. Preoperative data changed over time, with decreases in mean age (74.5-73.0 years), proportion of women from 60.6% (100 373/165 628) to 57.2% (174 908/305 845), and proportion of coexisting eye diseases from 30.0% (49 638/165 650) to 27.0% (82 704/305 846) and with improvements in preoperative visual acuity (mean logarithm of minimum angle of resolution [logMAR] 0.46 to 0.37). The use of topical anesthesia increased over time from 28.1% (26 238/93 320) to 71.7% (130 525/182 083). Surgical complications showed a significant decrease from 2.5% (4107/165 650) to 1.2% (3573/305 846). The visual outcome improved over time (mean logMAR 0.08 to 0.05), as did the absolute median prediction error (0.38 diopter [D] to 0.28 D). Trends in European cataract surgery practice patterns from 2008 to 2017 have moved toward younger patients with better preoperative visual acuity, fewer surgical complications, and better predicted refractions and visual outcomes.

Evaluation of Bayesian Forecasting Methods for Prediction of Tacrolimus Exposure Using Samples Taken on Two Occasions in Adult Kidney Transplant Recipients.

Bayesian forecasting-based limited sampling strategies (LSSs) for tacrolimus have not been evaluated for the prediction of subsequent tacrolimus exposure. This study examined the predictive performance of Bayesian forecasting programs/services for the estimation of future tacrolimus area under the curve (AUC) from 0 to 12 hours (AUC0-12) in kidney transplant recipients. Tacrolimus concentrations were measured in 20 adult kidney transplant recipients, 1 month post-transplant, on 2 occasions one week apart. Twelve samples were taken predose and 13 samples were taken postdose at the specified times on the first and second sampling occasions, respectively. The predicted AUC0-12 (AUCpredicted) was estimated using Bayesian forecasting programs/services and data from both sampling occasions for each patient and compared with the fully measured AUC0-12 (AUCmeasured) calculated using the linear trapezoidal rule on the second sampling occasion. The bias (median percentage prediction error [MPPE]) and imprecision (median absolute prediction error [MAPE]) were determined. Three programs/services were evaluated using different LSSs (C0; C0, C1, C3; C0, C1, C2, C4; and all available concentrations). MPPE and MAPE for the prediction of fully measured AUC0-12 were <15% for each program/service (with the exclusion of when only C0 was used), when using estimated AUC from data on the same (second) occasion. The MPPE and MAPE for the prediction of a future fully measured AUC0-12 were <15% for 2 programs/services (and for the third when participants who had a tacrolimus dose change between sampling days were excluded), when the occasion 1-AUCpredicted, using C0, C1, and C3, was compared with the occasion 2-AUCmeasured. All 3 Bayesian forecasting programs/services evaluated had acceptable bias and imprecision for predicting a future AUC0-12, using tacrolimus concentrations at C0, C1, and C3, and could be used for the accurate prediction of tacrolimus exposure in adult kidney transplant recipients.

Prediction of Individual Cochlear Implant Recipient Speech Perception With the Output Signal to Noise Ratio Metric.

A cochlear implant (CI) implements a variety of sound processing algorithms that seek to improve speech intelligibility. Typically, only a small number of parameter combinations are evaluated with recipients but the optimal configuration may differ for individuals. The present study evaluates a novel methodology which uses the output signal to noise ratio (OSNR) to predict complete psychometric functions that relate speech recognition to signal to noise ratio for individual CI recipients. Speech scores from sentence-in-noise tests in a "reference" condition were mapped to OSNR and a psychometric function was fitted. The reference variability was defined as the root mean square error between the reference scores and the fitted curve. To predict individual scores in a different condition, OSNRs in that condition were calculated and the corresponding scores were read from the reference psychometric function. In a retrospective experiment, scores were predicted for each condition and subject in three existing data sets of sentence scores. The prediction error was defined as the root mean square error between observed and predicted scores. In data set 1, sentences were mixed with 20 talker babble or speech weighted noise and presented at 65 dB sound pressure level (SPL). An adaptive test procedure was used. Sound processing was advanced combinatorial encoding (ACE, Cochlear Limited) and ACE with ideal binary mask processing, with five different threshold settings. In data set 2, sentences were mixed with speech weighted noise, street-side city noise or cocktail party noise and presented at 65 dB SPL. An adaptive test procedure was used. Sound processing was ACE and ACE with two different noise reduction schemes. In data set 3, sentences were mixed with four-talker babble at two input SNRs and presented at levels of 55-89 dB SPL. Sound processing utilised three different automatic gain control configurations. For data set 1, the median of individual prediction errors across all subjects, noise types and conditions, was 12% points, slightly better than the reference variability. The OSNR prediction method was inaccurate for the specific condition with a gain threshold of +10 dB. For data set 2, the median of individual prediction errors was 17% points and the reference variability was 11% points. For data set 3, the median prediction error was 9% points and the reference variability was 7% points. A Monte Carlo simulation found that the OSNR prediction method, which used reference scores and OSNR to predict individual scores in other conditions, was significantly more accurate (p < 0.01) than simply using reference scores as predictors. The results supported the hypothesis that the OSNR prediction method could accurately predict individual recipient scores for a range of algorithms and noise types, for all but one condition. The medians of the individual prediction errors for each data set were accurate within 6% points of the reference variability and compared favourably with prediction methodologies in other recent studies. Overall, the novel OSNR-based prediction method shows promise as a tool to assist researchers and clinicians in the development or fitting of CI sound processors.

Analysing the Epoch of Reionization with three-point correlation functions and machine learning techniques

ABSTRACT Three-point and high-order clustering statistics of the high-redshift 21 cm signal contain valuable information about the Epoch of Reionization (EoR). We present 3PCF-fast, an optimized code for estimating the three-point correlation function (3PCF) of 3D pixelized data such as the outputs from numerical and seminumerical simulations. After testing 3PCF-fast on data with known analytical 3PCF, we use machine learning techniques to recover the mean bubble size and global ionization fraction from correlations in the outputs of the publicly available 21cmfast code. We assume that foregrounds have been perfectly removed and negligible instrumental noise. Using ionization fraction data, our best multilayer perceptron (MLP) model recovers the mean bubble size with a median prediction error of around $10 {{\ \rm per\ cent}}$, or from the 21 cm differential brightness temperature with median prediction error of around $14 {{\ \rm per\ cent}}$. A further two MLP models recover the global ionization fraction with median prediction errors of around $4 {{\ \rm per\ cent}}$ (using ionization fraction data) or around $16 {{\ \rm per\ cent}}$ (using brightness temperature). Our results indicate that clustering in both the ionization fraction field and the brightness temperature field encode useful information about the progress of the EoR in a complementary way to other summary statistics. Using clustering would be particularly useful in regimes where high signal-to-noise ratio prevents direct measurement of bubble size statistics. We compare the quality of MLP models using the power spectrum, and find that using the 3PCF outperforms the power spectrum at predicting both global ionization fraction and mean bubble size.

Comparison of the accuracy of intraocular lens power calculation formulas for eyes after corneal refractive surgery.

In cataract surgery, calculating intraocular lens (IOL) power in patients who have previously received corneal refractive surgery on the same eye presents a clinical challenge. This study aims to compare the accuracy of the Haigis-L, Barrett True-K, and Shammas-PL formulas in predicting the IOL power in eyes following corneal refractive surgery. This study analyzed 32 eyes belonging to 28 patients who underwent cataract surgery and IOL implantation after previously undergoing myopic corneal refractive surgery. The IOL power was calculated using the Haigis-L, Barrett True-K, and Shammas-PL formulas, and the accuracy of the three formulas was compared. The Haigis-L, Barrett True-K, and Shammas-PL formulas had a mean arithmetic IOL prediction error of -0.65, -0.39, and -0.46, respectively. The mean numerical errors of the three formulas were significantly different from zero (P<0.001). The smallest median absolute refraction prediction error (median =0.40) belonged to the Barrett True-K formula, which was significantly smaller than that of the Haigis-L formula (median =0.57, P<0.05) but similar to that of the Shammas-PL formula (median =0.49, P>0.05). There was no significant difference in the percentage of eyes within either ±0.50 D or ±1.00 D of the predicted refraction error across the three formulas. The Barrett True-K formula can predict IOL power in eyes that have previously undergone myopic corneal refractive surgery better than the Haigis-L formula.