Absolute Prediction Error Research Articles

Recording of single-unit activities with flexible micro-electrocorticographic array in rats for decoding of whole-body navigation

Objective.Micro-electrocorticographic (μECoG) arrays are able to record neural activities from the cortical surface, without the need to penetrate the brain parenchyma. Owing in part to small electrode sizes, previous studies have demonstrated that single-unit spikes could be detected from the cortical surface, and likely from Layer I neurons of the neocortex. Here we tested the ability to useμECoG arrays to decode, in rats, body position during open field navigation, through isolated single-unit activities.Approach. μECoG arrays were chronically implanted onto primary motor cortex (M1) of Wistar rats, and neural recording was performed in awake, behaving rats in an open-field enclosure. The signals were band-pass filtered between 300-3000 Hz. Threshold-crossing spikes were identified and sorted into distinct units based on defined criteria including waveform morphology and refractory period. Body positions were derived from video recordings. We used gradient-boosting machine to predict body position based on previous 100 ms of spike data, and correlation analyses to elucidate the relationship between position and spike patterns.Main results.Single-unit spikes could be extracted during chronic recording fromμECoG, and spatial position could be decoded from these spikes with a mean absolute error of prediction of 0.135 and 0.090 in the x- and y- dimensions (of a normalized range from 0 to 1), and Pearson's r of 0.607 and 0.571, respectively.Significance. μECoG can detect single-unit activities that likely arise from superficial neurons in the cortex and is a promising alternative to intracortical arrays, with the added benefit of scalability to cover large cortical surface with minimal incremental risks. More studies should be performed in human related to its use as brain-machine interface.

Feature attention graph neural network for estimating brain age and identifying important neural connections in mouse models of genetic risk for Alzheimer’s disease

Abstract Alzheimer’s disease (AD), a widely studied neurodegenerative disorder, poses significant research challenges due to its high prevalence and complex etiology. Age, a critical risk factor for AD, is typically assessed by comparing physiological and estimated brain ages. This study utilizes mouse models expressing human alleles of APOE and human nitric oxide synthase 2 (hNOS2), replicating genetic risks for AD alongside a human-like immune response. We developed a multivariate model that incorporates brain structural connectomes, APOE genotypes, demographic traits (age and sex), environmental factors such as diet, and behavioral data to estimate brain age. Our methodology employs a Feature Attention Graph Neural Network (FAGNN) to integrate these diverse datasets. Behavioral data are processed using a 2D convolutional neural network (CNN), demographic traits via a 1D CNN, and brain connectomes through a graph neural network equipped with a quadrant attention module that accentuates critical neural connections. The FAGNN model demonstrated a mean absolute error in age prediction of 31.85 days and a root mean squared error of 41.84 days, significantly outperforming simpler models. Our analysis further focused on the brain age delta, which assesses accelerated or delayed aging by comparing brain age, predicted by FAGNN, to the chronological age. A high-fat diet and the presence of the human NOS2 gene were identified as significant accelerators of brain aging in the old age group. Key neural connections identified by FAGNN, such as those between the cingulum, corpus callosum, striatum, hippocampus, thalamus, hypothalamus, cerebellum, and piriform cortex, were found to be significant in the aging process. Validation using diffusion MRI-based metrics, including fractional anisotropy and return-to-origin probability measures across these connections, revealed significant age-related differences. These findings suggest that white matter degradation in the connections highlighted by FAGNN plays a key role in aging. Our findings suggest that the complex interplay of APOE genotype with sex, immunity, and environmental factors modulates brain aging and enhance our understanding of AD risk in mouse models of aging.

Developing Safety Performance Functions for Severe Distraction-Related Crashes along Kentucky’s Rural and Urban Two-Lane Roadways

This study develops safety performance functions (SPFs) for severe (“KA” or “fatal and suspected serious injury”) distraction-related crashes along Kentucky’s rural and urban two-lane undivided roadway segments using recent four-year (2018–2021) crash records. Additional efforts were made to meticulously scrutinize crash narratives categorized as non-distracted and correct those cases. To account for crash under-dispersion, the Conway–Maxwell–Poisson, heterogeneous Conway–Maxwell–Poisson, zero-inflated Conway–Maxwell–Poisson, and zero-inflated heterogeneous Conway–Maxwell–Poisson (ZI-HTCMP) models were fitted and compared. The ZI-HTCMP model outperformed the other models with respect to several goodness-of-fit measures (e.g., mean absolute deviance and mean square prediction error). From the developed SPFs for rural and urban two-lane roads, wider lanes and higher speed limits (55 mph) were associated with increased severe distraction-related crash frequencies. Furthermore, some variables were found to be significant in rural areas, but insignificant in urban areas, and vice versa. For example, major collector roads, minor collector/local roads, the presence of roadside guardrails, wider right-hand shoulders, the presence of horizontal curves, and the presence of vertical grades were associated with increased crash frequencies along rural two-lane roads. In addition, the proportion of heavy vehicles (>5%) and the existence of paved shoulders were associated with increased crashes along urban two-lane roads. The empirical Bayes method was then used to rank the top 10 distraction-related high crash locations (HCLs) for both rural and urban two-lane segments. In-depth investigation of HCLs highlighted single-vehicle distraction-related crashes as the most common collision type. Countermeasures were proposed to help reduce severe distraction-related crashes, for example, installing chevron signs along rural two-lane roads.

Intraocular Lens Power Calculation Formulas in Children-A Systematic Review.

Objectives: The selection of an appropriate formula for intraocular lens power calculation is crucial in phacoemulsification, particularly in pediatric patients. The most commonly used formulas are described and their accuracy evaluated in this study. Methods: This review includes papers evaluating the accuracy of intraocular lens power calculation formulas for children's eyes published from 2019-2024. The articles were identified by a literature search of medical and other databases (Pubmed/MEDLINE, Crossref, Google Scholar) using the combination of the following key words: "IOL power calculation formula", "pediatric cataract", "congenital cataract", "pediatric intraocular lens implantation", "lens power estimation", "IOL power selection", "phacoemulsification", "Hoffer Q", "Holladay 1", "SRK/T", "Barrett Universal II", "Hill-RBF", and "Kane". A total of 14 of the most recent peer-reviewed papers in English with the maximum sample sizes and the greatest number of compared formulas were considered. Results: The outcomes of mean absolute error and percentage of predictions within ±0.5 D and ±1.0 D were used to assess the accuracy of the formulas. In terms of MAE, Hoffer Q yielded the best result most often, just ahead of SRK/T and Barrett Universal II, which, together with Holladay 1, most often yielded the second-best outcomes. Considering patients with PE within ±1.0 D, Barrett Universal II most often gave the best results and Holladay 1 most often gave the second-best. Conclusions: Barrett Universal II seems to be the most accurate formula for intraocular lens calculation for children's eyes. Very good postoperative outcomes can also be achieved using the Holladay 1 formula. However, there is still no agreement in terms of formula choice.

Accuracy of 4 Different Methods for Estimation of Remaining Growth and Timing of Epiphysiodesis.

The calculation of remaining growth in children and the timing of epiphysiodesis in those with leg-length discrepancy (LLD) is most often done with 4 methods: the Green-Anderson, White-Menelaus, Moseley straight-line graph, and multiplier methods. The aims of this study were to identify the most accurate method with use of bone age or chronological age and to evaluate the influence of including inhibition in the calculations. One hundred and ninety-one children (10 to 17 years of age) with LLD who underwent surgical closure of the growth plate and were followed until skeletal maturity were identified from a local health register. Patients had at least 2 leg-length examinations with simultaneous bone-age assessments (according to the Greulich and Pyle method), with the last examination performed ≤6 months before surgery. The accuracy of each method was calculated as the mean absolute prediction error (predicted leg length - actual leg length at maturity) for the short leg, the long leg, and the LLD. Comparisons were made among the 4 methods and among calculations made with chronological age versus bone age and those made with versus those without incorporation of a reduced growth rate (inhibition) of the short leg compared with the long leg. The White-Menelaus method with use of bone age and a fixed inhibition rate was the most accurate method, with a prediction error of 1.5 ± 1.5 cm for the short leg, 1.0 ± 1.2 cm for the long leg, and 0.7 ± 0.7 cm for the LLD. Pairwise comparison of short-leg length and LLD according to the White-Menelaus and other methods showed that they were significantly different (p ≤ 0.002). The calculated inhibition rate did not increase accuracy. The White-Menelaus method used with bone age and constant inhibition should be the preferred method when predicting remaining growth and the timing of epiphysiodesis in children between 10 and 17 years of age. One examination is in most cases sufficient for the preoperative clinical investigation when chronological age and bone age are concordant. Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Short-term high-speed rail passenger flow prediction by integrating ensemble empirical mode decomposition with multivariate grey support vector machine

Short-term prediction of high-speed rail (HSR) passenger flow provides a daily ridership estimation for the near future, which is critical to HSR planning and operational decision making. This paper proposes a new methodology that integrates ensemble empirical mode decomposition with multivariate support vector machines (EEMD-MSVM). There are four steps in this hybrid forecasting approach: (i) explore the correlation of multivariate HSR passenger flows at various stations based on archived data; (ii) decompose empirical modes of historical passenger flows for each HSR station, using EEMD to generate a number of intrinsic mode functions (IMFs) and a trend term; (iii) predict the IMF for each correlated station pair using MSVM; and (iv) reconstruct the refined IMF components to predict daily multivariate HSR passenger flows. The proposed EEMD-MSVM approach is demonstrated with multiple OD pairs along the Wuhan-Guangzhou HSR in China. Results from various origin-destination pairs, show that the EEMD-MSVM approach outperforms the existing ensemble empirical mode decomposition with grey support vector machine approach (EEMD-GSVM). With the multivariate approach, the mean absolute percentage error in demand prediction is reduced by 13.9%, 1.2%, 1.0%, 2.0%, and 2.7% and the mean absolute deviation is reduced by 78.8, 38.0, 4.4, 4.6, and 3.9, between these OD pairs respectively. Such increase in short-term demand prediction accuracy can significantly improve HSR service planning, operations, and revenue management in the real world.

Brain age prediction using interpretable multi-feature-based convolutional neural network in mild traumatic brain injury

BackgroundConvolutional neural network (CNN) can capture the structural features changes of brain aging based on MRI, thus predict brain age in healthy individuals accurately. However, most studies use single feature to predict brain age in healthy individuals, ignoring adding information from multiple sources and the changes in brain aging patterns after mild traumatic brain injury (mTBI) were still unclear. MethodsHere, we leveraged the structural data from a large, heterogeneous dataset (N = 1464) to implement an interpretable 3D combined CNN model for brain-age prediction. In addition, we also built an atlas-based occlusion analysis scheme with a fine-grained human Brainnetome Atlas to reveal the age-sstratified contributed brain regions for brain-age prediction in healthy controls (HCs) and mTBI patients. The correlations between brain predicted age gaps (brain-PAG) following mTBI and individual's cognitive impairment, as well as the level of plasma neurofilament light were also examined. ResultsOur model utilized multiple 3D features derived from T1w data as inputs, and reduced the mean absolute error (MAE) of age prediction to 3.08 years and improved Pearson's r to 0.97 on 154 HCs. The strong generalizability of our model was also validated across different centers. Regions contributing the most significantly to brain age prediction were the caudate and thalamus for HCs and patients with mTBI, and the contributive regions were mostly located in the subcortical areas throughout the adult lifespan. The left hemisphere was confirmed to contribute more in brain age prediction throughout the adult lifespan. Our research showed that brain-PAG in mTBI patients was significantly higher than that in HCs in both acute and chronic phases. The increased brain-PAG in mTBI patients was also highly correlated with cognitive impairment and a higher level of plasma neurofilament light, a marker of neurodegeneration. The higher brain-PAG and its correlation with severe cognitive impairment showed a longitudinal and persistent nature in patients with follow-up examinations. ConclusionWe proposed an interpretable deep learning framework on a relatively large dataset to accurately predict brain age in both healthy individuals and mTBI patients. The interpretable analysis revealed that the caudate and thalamus became the most contributive role across the adult lifespan in both HCs and patients with mTBI. The left hemisphere contributed significantly to brain age prediction may enlighten us to be concerned about the lateralization of brain abnormality in neurological diseases in the future. The proposed interpretable deep learning framework might also provide hope for testing the performance of related drugs and treatments in the future.

Accuracy of intraocular lens power formulas in patients with average keratometry greater than 46 diopters

ObjectiveTo compare the accuracy of Kane, Barrett Universal II, Haigis, and SRK-T formulas in eyes with average keratometry greater than 46 diopters (D). MethodsA retrospective analysis was conducted on 101 eyes of 101 patients with average keratometry greater than 46 D. The absolute prediction error (EA) was obtained for each patient one month after surgery. The mean absolute prediction error (MEA), median absolute prediction error (MedEA) and the percentage of patients with absolute refractive error less than 0.25 D, 0.50 D, and 1.00 D were calculated for each formula analyzed. ResultsThe Kane formula achieved the lowest MEA (0.53 ± 0.43) and the lowest MedEA (0.41), followed by Barrett Universal II (MEA: 0.56 ± 0.42, MedEA: 0.49), SRK-T (MEA: 0.59 ± 0.44, MedEA: 0.54), and Haigis (MEA: 0.77 ± 0.47, MedEA: 0.69), showing a significant difference in the results. It was also observed that the Kane formula was the most accurate, with the highest percentage of patients, with EA less than 0.25 D, 0.50 D, and 1.00 D (30.7%, 54.4%, and 86.1%, respectively), while the Haigis formula was the least accurate (12.9%, 33.7%, and 69.3%, respectively). ConclusionIn eyes with corneas having average keratometry greater than 46 D, the Kane formula proves to be a useful tool in intraocular lens (IOL) power calculation and demonstrates higher precision compared to the Barrett Universal II, SRK-T, and Haigis formulas.

Accuracy of formulas for intraocular lens power for eyes undergoing descemet stripping automated endothelial keratoplasty and cataract surgery.

To compare accuracy in intraocular lens (IOL) power calculation in eyes undergoing combined cataract and Descemet stripping and automated endothelial keratoplasty (C-DSAEK) surgery of the following formulas: Barrett Universal II, EVO, Haigis, Hoffer Q, Holladay 2, Kane and SRK/T. 72 eyes from 72 patients (38 males, (53%)) with a mean age 68.08 ± 8.69 years (from 44 to 88 years old) underwent combined C-DSAEK were included. The IOL powers to implant were calculated with Barrett Universal II formula targeting -1 D refraction. Preoperative and postoperative data were used to obtain the median of absolute prediction errors (MAE) targeting emmetropia with every tested formula. Means of MAE calculated were +1.45 D for Barrett Universal II, +1.37 D for EVO, +1.48 D for Haigis, +1.38 D for Hoffer Q, +1.37 D for Holladay 2, +1.39 D for Kane and +1.31 D for SRK/T. SRK/T MAE showed major significant (p < 0.01) differences compared to the other formulas. Even if tested formulas are not able to accurately target emmetropia, SRK/T seems to be able to provide closer results in eyes undergoing C-DSAEK.

Fellow eye data for intraocular lens calculation in eyes undergoing combined phacovitrectomy.

To evaluate whether the intraocular lens (IOL) calculation of the fellow eye (FE) can be used in eyes undergoing combined phacovitrectomy. In this retrospective, consecutive case series, we enrolled patients who underwent combined phacovitrectomy with silicone oil removal and IOL implantation at the Goethe-University. Preoperative examinations included biometry (IOLMaster 700; Carl Zeiss). We used the IOL calculation of the FE (FE group) to calculate the prediction error compared with the IOL calculation using only the axial length (AL) of the FE (AL-FE group), as well as using the AL of the operated eye (OE group) in addition to the measurable biometric parameters. IOL calculation was performed using the Barrett Universal II formula. We compared the mean (MAE) and median absolute prediction error (MedAE) with each other. Furthermore, the number of eyes with ±0.50, ±1.00 and ±2.00 dioptres (D) deviation from the target refraction was compared. In total, 79 eyes of 79 patients were included. MedAE was lowest in the OE group (0.41 D), followed by FE group (1.00 D) and AL-FE group (1.02 D). Comparison between the AL-FE and FE groups showed no statistically significant difference (p = 0.712). Comparing eyes within ±0.50 D of the target refraction, the OE group (63.3%) performed best, followed by the AL-FE group (27.8%) and the FE group (26.6%). Our results indicate no clinically relevant difference between using the IOL calculation of the FE versus using only the AL of the FE in addition to the measurable parameters for the IOL calculation. A two-step procedure should always be strived for.

Intraocular lens tilt and decentration after primary and delayed implantation in phacovitrectomy for macula-off rhegmatogenous retinal detachment repair.

To evaluate the tilt and decentration of intraocular lenses (IOLs) in eyes that underwent phacovitrectomy for macula-off rhegmatogenous retinal detachment (RRD) repair. In this retrospective cohort study, 63 patients with macula-off RRD who underwent phacovitrectomy were enrolled and divided into a primary implantation group (group A) and a delayed implantation group (group B) according to the time of IOL implantation. Routine ocular examinations, including optometry, were performed, and the IOL tilt and decentration were measured. Postoperative refractive prediction error and IOL tilt and decentration were compared between the two groups, and the relevant factors of IOL positions were analyzed. The mean spherical refractive prediction error (MSE) was - 0.53 ± 0.39 D in group A and - 0.09 ± 0.55 in group B. The mean absolute spherical refractive prediction error (MASE) was 0.61 ± 0.23 D in group A and 0.50 ± 0.22 D in group B. The difference in MSE (t = -3.623, P < 0.05) but not in MASE (t = 1.866, P = 0.067) between the two groups was significant. The mean IOL tilt and mean decentration were 8.54° ±3.65° and 0.55 ± 0.18 mm in group A and 10.62° ±3.29° and 0.66 ± 0.14 mm, respectively, in Group B. Differences in tilt (t = -2.373, P < 0.05) and decentration (t = -2.698, P < 0.05) between the two groups were both significant. In group B, the mean time interval between phacovitrectomy and IOL implantation (TV-I) was 25.2 ± 6.9 weeks, and the absolute values of both IOL tilt and decentration were positively correlated with TV-I. In phacovitrectomy for macula-off RRD repair, primary IOL implantation leads to a postoperative myopic shift, while delayed IOL implantation may cause greater IOL tilt and decentration.

An evaluation of sample size requirements for developing risk prediction models with binary outcomes

BackgroundRisk prediction models are routinely used to assist in clinical decision making. A small sample size for model development can compromise model performance when the model is applied to new patients. For binary outcomes, the calibration slope (CS) and the mean absolute prediction error (MAPE) are two key measures on which sample size calculations for the development of risk models have been based. CS quantifies the degree of model overfitting while MAPE assesses the accuracy of individual predictions.MethodsRecently, two formulae were proposed to calculate the sample size required, given anticipated features of the development data such as the outcome prevalence and c-statistic, to ensure that the expectation of the CS and MAPE (over repeated samples) in models fitted using MLE will meet prespecified target values. In this article, we use a simulation study to evaluate the performance of these formulae.ResultsWe found that both formulae work reasonably well when the anticipated model strength is not too high (c-statistic < 0.8), regardless of the outcome prevalence. However, for higher model strengths the CS formula underestimates the sample size substantially. For example, for c-statistic = 0.85 and 0.9, the sample size needed to be increased by at least 50% and 100%, respectively, to meet the target expected CS. On the other hand, the MAPE formula tends to overestimate the sample size for high model strengths. These conclusions were more pronounced for higher prevalence than for lower prevalence. Similar results were drawn when the outcome was time to event with censoring. Given these findings, we propose a simulation-based approach, implemented in the new R package ‘samplesizedev’, to correctly estimate the sample size even for high model strengths. The software can also calculate the variability in CS and MAPE, thus allowing for assessment of model stability.ConclusionsThe calibration and MAPE formulae suggest sample sizes that are generally appropriate for use when the model strength is not too high. However, they tend to be biased for higher model strengths, which are not uncommon in clinical risk prediction studies. On those occasions, our proposed adjustments to the sample size calculations will be relevant.

IOL calculation using six formulas in children undergoing lens extraction and primary IOL implantation with and without posterior optic capture.

To evaluate formulas for intraocular lens (IOL) calculation in children undergoing lens extraction and IOL implantation. Retrospective, consecutive case series at the Department of Ophthalmology, Goethe University Frankfurt, Germany. We included eyes that received lens extraction and IOL implantation (SN60AT, Alcon, Fort Worth, TX) due to congenital or juvenile cataract. Preoperative assessments included biometry (IOLMaster 500/700, Carl Zeiss Meditec, Germany). To evaluate the measurements, we compared the mean prediction error (MPE), mean and median absolute prediction error (MAE, MedAE) of six different formulas, and number of eyes within ± 0.5, ± 1.0, ± 2.0D of target refraction. Postoperative spherical equivalent was measured by retinoscopy 4-12 weeks after surgery. 66 eyes matched our inclusion criteria with a mean age of 6.3 years ± 3.2. MedAE was lowest in SRK/T (0.55D ± 1.08) followed by Holladay I (0.75D ± 1.00), EVO 2.0 (0.80D ± 0.89), Barrett Universal II (BUII, 0.86D ± 1.00), Hoffer Q (0.97 D ± 0.94), and Haigis (1.10D ± 0.95). Regarding eyes within ± 0.5D SRK/T (45.5.%, 30 eyes) performed best, followed by Holladay I (36.4%, 24 eyes), EVO 2.0 and BUII (each 34.8%, 23 eyes). There was a myopic shift seen in all formulas (MPE: -0.21 to -0.90D). Using modern formulas, or even AI formulas, for IOL calculation in children's eyes does barely improve predictability of the postoperative refraction. A myopic shift can be found for all formulas. However, specific formulas like SRK/T seem to better anticipate this.

Comparative Accuracy of Barrett Integrated Keratometry Toric Calculator With Predicted Versus Measured Posterior Corneal Astigmatism.

To compare the prediction accuracy of the Barrett toric calculator using standard or integrated keratometry (IK) mode in combination with predicted or measured posterior corneal astigmatism (PCA) in a group of patients with cataract implanted with non-toric IOLs. In this retrospective clinical cohort study, the medical records of patients with age-related cataract who underwent phacoemulsification with the implantation of an aspheric monofocal IOL were reviewed. Four methods, including standard keratometry with predicted PCA (PPCA), IK combined with predicted PCA (IK-PPCA), and IK combined with measured PCA derived from IOLMaster 700 (Carl Zeiss Meditec AG) or CASIA2 (Tomey) (IK-MMPCA or IK-CMPCA), were applied to the Barrett toric calculator to calculate the predicted residual astigmatism. The mean absolute prediction error (MAPE), centroid of the prediction error, and proportion of eyes within the prediction error of ±0.50, ±0.75, and ±1.00 diopters (D) were all ciphered out from the four methods, respectively. Data from 129 eyes of 129 patients were included in this study. The MAPE of the IK-PPCA method (0.57 ± 0.36 D) was significantly smaller than that of the PPCA (0.62 ± 0.38 D) and IK-CMPCA (0.63 ± 0.46 D) methods (P = .048 and .014, respectively). There were no significant differences in the centroid vectors of prediction errors and predictability rates among the four methods (all P > .05). In the current version of the Barrett toric calculator, the predictive accuracy of the IK mode incorporating PPCA was slightly superior to using the standard keratometry mode or incorporating MPCA. [J Refract Surg. 2024;40(7):e453-e459.].

Machine Learning Approach in Dosage Individualization of Isoniazid for Tuberculosis.

Isoniazid is a first-line antituberculosis agent with high variability, which would profit from individualized dosing. Concentrations of isoniazid at 2 h (C2h), as an indicator of safety and efficacy, are important for optimizing therapy. The objective of this study was to establish machine learning (ML) models to predict the C2h, that can be used for establishing an individualized dosing regimen in clinical practice. Published population pharmacokinetic (PopPK) models for adults were searched based on PubMed and ultimately four reliable models were selected for simulating individual C2h datasets under different conditions (demographics, genotype, ethnicity, etc.). Machine learning models were trained on simulated C2h obtained from the four PopPK models. Five different algorithms were used for ML model building to predict C2h. Real-world data were used for predictive performance evaluations. Virtual trials were used to compare ML-optimized doses with PopPK model-optimized doses. Categorical boosting (CatBoost) exhibited the highest prediction ability. Target C2h can be predicted using the ML model combined with the dosing regimen and three covariates (N-acetyltransferase 2 [NAT2] genotypes, weight and race [Asians and Africans]). Real-world data validation results showed that the ML model can achieve an overall prediction accuracy of 93.4%. Using the final ML model, the mean absolute prediction error value decreased by 45.7% relative to the average of PopPK models. Using the ML-optimized dosing regimen, the probability of target attainment increased by 43.7% relative to the PopPK model-optimized dosing regimens. Machine learning models were developed with great predictive performance, which can be used to determine the individualized initial dose of isoniazid in adult patients.

Accuracy of intra ocular lens calculation formulae in patients with pseudoexfoliation syndrome

BackgroundThe purpose of this study was to investigate the visual and refractive outcomes in patients with pseudoexfoliation (PXF) undergoing routine cataract surgery and to compare the accuracy of intraocular lens (IOL) power calculation formulae.MethodsRetrospective case-series study from Shamir medical center, a public hospital, Israel. Medical records of patients who underwent routine cataract surgery between January 2019 and August 2021 were investigated. Postoperative visual acuity and manifest refraction were examined. The error in predicted refraction and IOL power calculation accuracy within a range of ± 0.50 to ± 1.00 diopters were compared between different IOL calculating formulae.Results151 eyes of 151 patients ages 73.9 ± 7.1 years were included in this study- 58 eyes in the PXF group and 93 eyes in the control group. The mean absolute error (MAE) for the BUII formula was 0.63D ± 0.87 for the PXF group and 0.36D ± 0.48 for the control group (p < 0.05). The MAE for the Hill-RBF 3.0 formula was 0.61D ± 0.84 for the PXF group and 0.42D ± 0.55 for the control group (p = 0.05). There were significant differences in MAE and MedAE between PXF group and control group measures (p < 0.05). In the PXF group there were no significant differences between the different formulae.ConclusionsThere were significant differences in accuracy of IOL power calculations in all formulae between PXF group and control group measures. PXF patients show hyperopic shift from predicted refraction. Barret universal II formula had the highest proportion of eyes with absolute error in prediction below or equal to 0.50 D in both PXF and control groups.

Development and validation of individualized tacrolimus dosing software for Chinese pediatric liver transplantation patients: a population pharmacokinetic approach.

We aim to describe the population pharmacokinetics (PPK) of tacrolimus in Chinese pediatric patients under 4years old after liver transplantation and to develop individualized tacrolimus dosing software. A total of 663 blood concentrations from 85 patients aged 4.57months to 3.97years were collected in this study. PPK analysis was performed using a nonlinear mixed effects modeling approach with the software, Phoenix. Using C#, an individualized tacrolimus dosing software was created. The software was then used to predict the concentrations of another ten pediatric liver transplantation patients to verify the accuracy of said software. The predictive error (PE) and the absolute predictive error (APE) for each predicted time point were computed. A one-compartment model with first-order elimination best fitted the data. The apparent volume of distribution (V/F) and apparent clearance (CL/F) were 198.65 L and 2.41 L/h. Postoperative days (POD), total bilirubin (TBIL), and the use of voriconazole significantly influenced tacrolimus apparent clearance. The incorporation of an increasing number of actual blood drug concentrations into the prediction resulted in a decrease in both PE (72%, 17%, 7%) and APE (87%, 53%, 26%). A qualified PPK model of tacrolimus was developed in Chinese pediatric patients. The individualized tacrolimus dosing software could be used as a suitable tool for the personalization of tacrolimus dosing for pediatric patients after liver transplantation.

Прогноз сильной жары в Красноярске с использованием региональной модели WRF-ARW

The reasons for the occurrence of a severe heatwave in Krasnoyarsk have been studied, and the accuracy of the forecasts of the WRF-ARW regional model has been assessed. Four episodes of hot weather in 2020, 2021, and 2023 have been analyzed. The heatwave in June 2023 was especially intense and met the criteria for a severe weather event. All examined cases were associated with the intense advection of warm tropical air in the leading edge of the upper-air trough. The formation of the urban heat island has been investigated, and its intensity in the nighttime was found to be 6.5°C. During daytime hours, the heat island was weaker than at nighttime. It has been determined that on the second day of forecasting, the WRF-ARW model accurately reproduces the urban heat island and the impact of the Yenisei River, although it underestimates the predicted amplitude of daily temperature variations. The absolute error in air temperature predictions for the time moments close to the time of daily maximum temperatures was 2.6°C (33 hours of model time).

Accuracy of using the axial length of the fellow eye for IOL calculation in retinal detachment eyes undergoing silicone oil removal

PurposeEvaluate whether the axial length of the fellow eye can be used to calculate the intraocular lens (IOL) in eyes with retinal detachment.DesignRetrospective, consecutive case series.MethodsOur study was conducted at...

Decision-making Support Platform and Security Design for Rural Leisure Tour Industry based on SEA Method and SVR Model

An accurate short-term passenger flow forecast of rural tourism can avoid accidents as much as possible. However, the short-term passenger flow of rural leisure tourism shows nonlinear, seasonal, random, and other complex characteristics. Meanwhile, the traditional forecasting methods are often difficult to achieve accurate forecasting. Therefore, this study now used the nonlinear mapping function in the support vector regression to convert the passenger traffic training sample into a high-dimensional feature space and established a linear decision function. Then the influence of periodicity on the prediction effect through seasonal index adjustment was reduced. Finally, event triggering combined with core embedding technology for tamper-proof detection was adopted to improve the security of the platform. The results showed that the minimum absolute error of prediction with improved SVM model was 0.27% compared with traditional model and autoregressive integrated moving average model. After the introduction of the Internet search factor, the traffic prediction result was more accurate, which was 0.0425 smaller than that without the introduction of the Internet search factor. When the concurrency was less than 100 times/s, the average response time difference before and after adding the core embedded program was small, indicating that the security of page tampering technology was high. This research method can effectively predict the passenger flow of rural leisure tourism industry and ensure the safety of the platform.