Category: Statistics

Ophthalmol Ther. 2023 Oct 9. doi: 10.1007/s40123-023-00814-w. Online ahead of print.

ABSTRACT

INTRODUCTION: During phacoemulsification, the infusion pressure can cause the liquefied vitreous fluid to escape through the ciliary fiber interspace in highly myopic eyes, leading to reduced vitreous cavity pressure similar to vitrectomized eyes. This study assessed the probability of low intraocular pressure (IOP) in high myopic eyes with different axial length (AL) group undergoing cataract surgery, as well as the impact of balanced salt solution (BSS) supplementation and the optimal IOP value for such supplementation.

METHODS: The control group consisted of cataract eyes with normal AL (group 1: 22 mm ≤ AL < 24.5 mm), while cataract eyes with high axial myopia were categorized into three groups (group 2: 26 mm ≤ AL < 28 mm, group 3: 28 mm ≤ AL < 30 mm, group 4: AL ≥ 30 mm). IOP was measured using the iCare pro tonometer intraoperatively. BSS supplementation was performed to raise IOP in cases of low IOP, before intraocular lens (IOL) implantation and before the end of surgery. The probability of low IOP was calculated, and the IOP before and after supplementation were compared.

RESULTS: Ninety-five eyes were included. The total probability of low IOP in groups 2, 3, and 4 was 56.52, 62.50, and 70.83%, respectively, significantly higher than that in group 1 (16.67%). Similarly, the probability of low IOP before IOL implantation was significantly higher in groups 2, 3, 4 (43.48, 41.67, and 62.50%) compared to group 1 (4.17%, P < 0.05). The IOP before and after the first BSS supplementation in three high myopia groups were statistically significant (P < 0.05), increasing from 12.10 mmHg (range, 6.0-24.9 mmHg) to 16.60 mmHg (range, 10.2-34.4 mmHg). After the second BSS supplementation before the end of surgery, the IOP of high myopia groups increased from 12.60 mmHg (range, 7.0-25.3 mmHg) to 14.60 mmHg (range, 9.8-25.3 mmHg).

CONCLUSIONS: The condition of highly myopic eyes seems more likely to develop low IOP during cataract surgery. There is an observed correlation: as AL increases, the total probability of low IOP rises. In patients with IOP < 9.5 mmHg intraoperatively, fluid supplementation via a side-port incision can effectively raise IOP to about 16 mmHg before IOL implantation and about 14 mmHg after incision sealing, facilitating smoother IOL implantation and reducing the risk of postoperative low IOP.

TRIAL REGISTRATION: NCT, NCT05201677. Registered 23 November 2021-Retrospectively registered, https://clinicaltrials.gov/ct2/show/NCT05201677 .

PMID:37812307 | DOI:10.1007/s40123-023-00814-w

Eur Radiol. 2023 Oct 9. doi: 10.1007/s00330-023-10254-6. Online ahead of print.

ABSTRACT

OBJECTIVES: To develop a multitask deep learning (DL) algorithm to automatically classify mammography imaging findings and predict the existence of extensive intraductal component (EIC) in invasive breast cancer.

METHODS: Mammograms with invasive breast cancers from 2010 to 2019 were downloaded for two radiologists performing image segmentation and imaging findings annotation. Images were randomly split into training, validation, and test datasets. A multitask approach was performed on the EfficientNet-B0 neural network mainly to predict EIC and classify imaging findings. Three more models were trained for comparison, including a single-task model (predicting EIC), a two-task model (predicting EIC and cell receptor status), and a three-task model (combining the abovementioned tasks). Additionally, these models were trained in a subgroup of invasive ductal carcinoma. The DeLong test was used to examine the difference in model performance.

RESULTS: This study enrolled 1459 breast cancers on 3076 images. The EIC-positive rate was 29.0%. The three-task model was the best DL model with an area under the curve (AUC) of EIC prediction of 0.758 and 0.775 at the image and breast (patient) levels, respectively. Mass was the most accurately classified imaging finding (AUC = 0.915), followed by calcifications and mass with calcifications (AUC = 0.878 and 0.824, respectively). Cell receptor status prediction was less accurate (AUC = 0.625-0.653). The multitask approach improves the model training compared to the single-task model, but without significant effects.

CONCLUSIONS: A mammography-based multitask DL model can perform simultaneous imaging finding classification and EIC prediction.

CLINICAL RELEVANCE STATEMENT: The study results demonstrated the potential of deep learning to extract more information from mammography for clinical decision-making.

KEY POINTS: • Extensive intraductal component (EIC) is an independent risk factor of local tumor recurrence after breast-conserving surgery. • A mammography-based deep learning model was trained to predict extensive intraductal component close to radiologists’ reading. • The developed multitask deep learning model could perform simultaneous imaging finding classification and extensive intraductal component prediction.

PMID:37812297 | DOI:10.1007/s00330-023-10254-6

Biochem Genet. 2023 Oct 9. doi: 10.1007/s10528-023-10507-5. Online ahead of print.

ABSTRACT

Breast cancer (BC) is the most abundant and aggressive cancer that impacts millions of women with poorly understood mechanisms. Here, we aimed to investigate the function of LINC01806 in BC development. Human BC tissues and nearby normal specimens were taken from diagnosed BC patients. The expression levels of LINC01806, miR-1286, ZEB1, and EMT-related markers were evaluated by qRT-PCR and western blotting. FISH was used to visualize the subcellular localization of LINC01806. The viability, proliferation, migration and invasion capacities of BC cells were assessed by MTT, colony formation, and transwell assays. Interactions among LINC01806, miR-1286 and ZEB1 were validated by dual luciferase assay. The unpaired Student t-test (for two groups) or one-way ANOVA following with Tukey post-hoc test (for more than three groups) was employed for statistical analysis. LINC01806 level was elevated in BC tissues. Knockdown of LINC01806 suppressed EMT process and BC cell proliferation, migration, and invasion. LINC01806 co-localized and directly bound with miR-1286 in the cytoplasm. MiR-1286 inhibitor blocked the effects of LINC01806 knockdown on BC cell EMT, proliferation and migration. MiR-1286 targeted ZEB1 and overexpression of ZEB1 blocked the regulatory functions of miR-1286 mimics in BC. LINC01806 facilitates EMT and accelerates BC cell proliferation, migration, and invasion via acting as miR-1286 sponge to disinhibit ZEB1 expression.

PMID:37812283 | DOI:10.1007/s10528-023-10507-5

Discov Nano. 2023 Oct 9;18(1):124. doi: 10.1186/s11671-023-03906-5.

ABSTRACT

Integration and scalability have posed significant problems in the advancement of brain-inspired intelligent systems. Here, we report a self-formed Ag device fabricated through a chemical dewetting process using an Ag organic precursor, which offers easy processing, scalability, and flexibility to address the above issues to a certain extent. The conditions of spin coating, precursor dilution, and use of solvents were varied to obtain different dewetted structures (broadly classified as bimodal and nearly unimodal). A microscopic study is performed to obtain insight into the dewetting mechanism. The electrical behavior of selected bimodal and nearly unimodal devices is related to the statistical analysis of their microscopic structures. A capacitance model is proposed to relate the threshold voltage (V_th) obtained electrically to the various microscopic parameters. Synaptic functionalities such as short-term potentiation (STP) and long-term potentiation (LTP) were emulated in a representative nearly unimodal and bimodal device, with the bimodal device showing a better performance. One of the cognitive behaviors, associative learning, was emulated in a bimodal device. Scalability is demonstrated by fabricating more than 1000 devices, with 96% exhibiting switching behavior. A flexible device is also fabricated, demonstrating synaptic functionalities (STP and LTP).

PMID:37812259 | DOI:10.1186/s11671-023-03906-5

Prev Med Rep. 2023 Sep 17;36:102404. doi: 10.1016/j.pmedr.2023.102404. eCollection 2023 Dec.

ABSTRACT

The World Health Organization stipulate children and adolescents should accumulate 60 min of physical activity (PA) daily; globally only 25% achieve this. Active travel to school (ATS) is a method of integrating PA into daily life with a documented health benefit accruing. Understanding factors associated with ATS is essential to inform a systems approach to increase ATS participation. This study described patterns of commuting to school and examined factors associated with ATS. Children’s Sport Participation & Physical Activity Study 2018 data was used, an all-Ireland cross-sectional study of 6,650 students. Logistic regression analysis was performed to determine factors independently associated with ATS. Most common commute to school methods were private car for primary (57%) and public transport for secondary (39%) students. The recommended 60 min of daily PA a week prior to the survey was achieved by 19.5% for primary and 12.6% for secondary students. Republic of Ireland (ROI) nationality (OR 1.09 95 %CI 1.02-1.16), meeting PA guidelines (OR 1.26 95 %CI 1.08-1.46), attending a ROI school (OR 2.27 95 %CI 2.02-2.57), attending a non-Delivering Equality of Opportunity in Schools (DEIS) school (OR 2.47 95 %CI 1.87-3.24), attending an urban school (OR 3.96 95 %CI 3.41-4.59) were each independently statistically significantly associated with ATS. Living in a family with a car (OR 0.27 95 %CI 0.19-0.39), attending secondary school (OR 0.69 95 %CI 0.62-0.78), attending a small sized (<33rd percentile) school (OR 0.68 95 %CI 0.60-0.77), living >5 km from school (OR 0.22 95 %CI 0.2-0.24) were each significantly negatively associated with ATS. ATS is a means of increasing youth PA and health. Factors associated with ATS can inform further research and intervention to increase ATS participation.

PMID:37810264 | PMC:PMC10558775 | DOI:10.1016/j.pmedr.2023.102404

iScience. 2023 Sep 22;26(10):108008. doi: 10.1016/j.isci.2023.108008. eCollection 2023 Oct 20.

ABSTRACT

Recent work suggests that serial dependence, where perceptual decisions are biased toward previous stimuli, arises from the prior that sensory input is temporally correlated. However, existing studies have mostly used random stimulus sequences that do not involve such temporal consistencies. Here, we manipulated the temporal statistics of visual stimuli to examine the role of true temporal correlations in serial dependence. In two experiments, observers reproduced the orientation of the last stimulus in a sequence, while we varied temporal correlations in the stimulus features at two timescales: stimulus history within the trial and decision history across trials. We found a clear dissociation: increasing temporal correlation in the stimulus history led to adaptation-like repulsive biases, whereas increasing temporal correlation in the decision history reduced attractive biases. Thus, we suggest that temporal correlation enhances the discriminative ability of the visual system, revealing the fundamental role of the broader temporal context.

PMID:37810242 | PMC:PMC10551895 | DOI:10.1016/j.isci.2023.108008

iScience. 2023 Sep 22;26(10):108007. doi: 10.1016/j.isci.2023.108007. eCollection 2023 Oct 20.

ABSTRACT

The vascular system is a multi-scale network whose functionality depends on its structure, and for which structural alterations can be linked to pathological shifts. Though biologists use multiple 3D imaging techniques to visualize vascular networks, the 3D image processing methodologies remain sources of biases, and the extraction of quantitative morphometric descriptors remains flawed. The article, first, reviews the current 3D image processing methodologies, and morphometric descriptors of vascular network images mainly obtained by light-sheet microscopy on optically cleared organs, found in the literature. Second, it proposes operator-independent segmentation and skeletonization methodologies using the freeware ImageJ. Third, it gives more extractable network-level (density, connectivity, fractal dimension) and segment-level (length, diameter, tortuosity) 3D morphometric descriptors and how to statistically analyze them. Thus, it can serve as a guideline for biologists using 3D imaging techniques of vascular networks, allowing the production of more comparable studies in the future.

PMID:37810224 | PMC:PMC10551892 | DOI:10.1016/j.isci.2023.108007

iScience. 2023 Sep 15;26(10):107934. doi: 10.1016/j.isci.2023.107934. eCollection 2023 Oct 20.

ABSTRACT

Ovarian cancer is the deadliest gynecologic malignancy in women, with a 46% five-year overall survival rate. The objective of the study was to investigate the effects of non-homologous end-joining (NHEJ) genes on clinical outcomes of ovarian cancer patients. To determine if these genes act as prognostic biomarkers of mortality and disease progression, the expression profiles of 48 NHEJ-associated genes were analyzed using an array of statistical and machine learning techniques: logistic regression models, decision trees, naive-Bayes, two sample t-tests, support vector machines, hierarchical clustering, principal component analysis, and neural networks. In this process, the correlation of genes with patient survival and disease progression and recurrence was noted. Also, multiple features from the gene set were found to have significant predictive capabilities. APTX, BRCA1, PAXX, LIG1, and TP53 were identified as most important out of all the candidate genes for predicting clinical outcomes of ovarian cancer patients.

PMID:37810216 | PMC:PMC10558711 | DOI:10.1016/j.isci.2023.107934

IEEE Trans Inf Theory. 2022 Oct;68(10):6631-6662. doi: 10.1109/tit.2022.3175691. Epub 2022 May 17.

ABSTRACT

The recent success of generative adversarial networks and variational learning suggests that training a classification network may work well in addressing the classical two-sample problem, which asks to differentiate two densities given finite samples from each one. Network-based methods have the computational advantage that the algorithm scales to large datasets. This paper considers using the classification logit function, which is provided by a trained classification neural network and evaluated on the testing set split of the two datasets, to compute a two-sample statistic. To analyze the approximation and estimation error of the logit function to differentiate near-manifold densities, we introduce a new result of near-manifold integral approximation by neural networks. We then show that the logit function provably differentiates two sub-exponential densities given that the network is sufficiently parametrized, and for on or near manifold densities, the needed network complexity is reduced to only scale with the intrinsic dimensionality. In experiments, the network logit test demonstrates better performance than previous network-based tests using classification accuracy, and also compares favorably to certain kernel maximum mean discrepancy tests on synthetic datasets and hand-written digit datasets.

PMID:37810208 | PMC:PMC10558099 | DOI:10.1109/tit.2022.3175691

J Comput Graph Stat. 2023;32(3):974-984. doi: 10.1080/10618600.2022.2119986. Epub 2022 Oct 14.

ABSTRACT

The reduced-rank regression model is a popular model to deal with multivariate response and multiple predictors, and is widely used in biology, chemometrics, econometrics, engineering, and other fields. In the reduced-rank regression modelling, a central objective is to estimate the rank of the coefficient matrix that represents the number of effective latent factors in predicting the multivariate response. Although theoretical results such as rank estimation consistency have been established for various methods, in practice rank determination still relies on information criterion based methods such as AIC and BIC or subsampling based methods such as cross validation. Unfortunately, the theoretical properties of these practical methods are largely unknown. In this paper, we present a novel method called StARS-RRR that selects the tuning parameter and then estimates the rank of the coefficient matrix for reduced-rank regression based on the stability approach. We prove that StARS-RRR achieves rank estimation consistency, i.e., the rank estimated with the tuning parameter selected by StARS-RRR is consistent to the true rank. Through a simulation study, we show that StARS-RRR outperforms other tuning parameter selection methods including AIC, BIC, and cross validation as it provides the most accurate estimated rank. In addition, when applied to a breast cancer dataset, StARS-RRR discovers a reasonable number of genetic pathways that affect the DNA copy number variations and results in a smaller prediction error than the other methods with a random-splitting process.

PMID:37810194 | PMC:PMC10554232 | DOI:10.1080/10618600.2022.2119986