Categories
Nevin Manimala Statistics

Pathological Spectra of the Fisher Information Metric and Its Variants in Deep Neural Networks

Neural Comput. 2021 Jul 26;33(8):2274-2307. doi: 10.1162/neco_a_01411.

ABSTRACT

The Fisher information matrix (FIM) plays an essential role in statistics and machine learning as a Riemannian metric tensor or a component of the Hessian matrix of loss functions. Focusing on the FIM and its variants in deep neural networks (DNNs), we reveal their characteristic scale dependence on the network width, depth, and sample size when the network has random weights and is sufficiently wide. This study covers two widely used FIMs for regression with linear output and for classification with softmax output. Both FIMs asymptotically show pathological eigenvalue spectra in the sense that a small number of eigenvalues become large outliers depending on the width or sample size, while the others are much smaller. It implies that the local shape of the parameter space or loss landscape is very sharp in a few specific directions while almost flat in the other directions. In particular, the softmax output disperses the outliers and makes a tail of the eigenvalue density spread from the bulk. We also show that pathological spectra appear in other variants of FIMs: one is the neural tangent kernel; another is a metric for the input signal and feature space that arises from feedforward signal propagation. Thus, we provide a unified perspective on the FIM and its variants that will lead to more quantitative understanding of learning in large-scale DNNs.

PMID:34310678 | DOI:10.1162/neco_a_01411

Categories
Nevin Manimala Statistics

Early alpha/beta oscillations reflect the formation of face-related expectations in the brain

PLoS One. 2021 Jul 26;16(7):e0255116. doi: 10.1371/journal.pone.0255116. eCollection 2021.

ABSTRACT

Although statistical regularities in the environment often go explicitly unnoticed, traces of implicit learning are evident in our neural activity. Recent perspectives have offered evidence that both pre-stimulus oscillations and peri-stimulus event-related potentials are reliable biomarkers of implicit expectations arising from statistical learning. What remains ambiguous, however, is the origination and development of these implicit expectations. To address this lack of knowledge and determine the temporal constraints of expectation formation, pre-stimulus increases in alpha/beta power were investigated alongside a reduction in the N170 and a suppression in peri-/post-stimulus gamma power. Electroencephalography was acquired from naive participants who engaged in a gender classification task. Participants were uninformed, that eight face images were sorted into four reoccurring pairs which were pseudorandomly hidden amongst randomly occurring face images. We found a reduced N170 for statistically expected images at left parietal and temporo-parietal electrodes. Furthermore, enhanced gamma power following the presentation of random images emphasized the bottom-up processing of these arbitrary occurrences. In contrast, enhanced alpha/beta power was evident pre-stimulus for expected relative to random faces. A particularly interesting finding was the early onset of alpha/beta power enhancement which peaked immediately after the depiction of the predictive face. Hence, our findings propose an approximate timeframe throughout which consistent traces of enhanced alpha/beta power illustrate the early prioritisation of top-down processes to facilitate the development of implicitly cued face-related expectations.

PMID:34310657 | DOI:10.1371/journal.pone.0255116

Categories
Nevin Manimala Statistics

Expression and role of lumican in acute aortic dissection: A human and mouse study

PLoS One. 2021 Jul 26;16(7):e0255238. doi: 10.1371/journal.pone.0255238. eCollection 2021.

ABSTRACT

INTRODUCTION: Aortic dissection (AD) is a life-threatening emergency, and lumican (LUM) is a potential Biomarker for AD diagnosis. We investigated LUM expression patterns in patients with AD and explored the molecular functions of Lum in AD mice model.

METHODS: LUM expression patterns were analyzed using aortic tissues of AD patients, and serum soluble LUM (s-LUM) levels were compared between patients with acute AD (AAD) and chronic AD (CAD). Lum-knockout (Lum-/-) mice were challenged with β-aminopropionitrile (BAPN) and angiotensin II (Ang II) to induce AD. The survival rate, AD incidence, and aortic aneurysm (AA) in these mice were compared with those in BAPN-Ang II-challenged wildtype (WT) mice. Tgf-β/Smad2, Mmps, Lum, and Nox expression patterns were examined.

RESULTS: LUM expression was detected in the intima and media of the ascending aorta in patients with AAD. Serum s-LUM levels were significantly higher in patients with AAD than CAD. Furthermore, AD-associated mortality and thoracic aortic rupture incidence were significantly higher in the Lum-/- AD mice than in the WT AD mice. However, no significant pathologic changes in AA were observed in the Lum-/- AD mice compared with the WT AD mice. The BAPN-Ang II-challenged WT and Lum-/- AD mice had higher Tgf-β, p-Smad2, Mmp2, Mmp9, and Nox4 levels than those of non-AD mice. We also found that Lum expression was significantly higher in the BAPN-Ang II-challenged WT in comparison to the unchallenged WT mice.

CONCLUSION: LUM expression was altered in patients with AD display increased s-LUM in blood, and Lum-/- mice exhibited augmented AD pathogenesis. These findings support the notion that LUM is a biomarker signifying the pathogenesis of injured aorta seen in AAD. The presence of LUM is essential for maintenance of connective tissue integrity. Future studies should elucidate the mechanisms underlying LUM association in aortic changes.

PMID:34310653 | DOI:10.1371/journal.pone.0255238

Categories
Nevin Manimala Statistics

Urinary actin, as a potential marker of sepsis-related acute kidney injury: A pilot study

PLoS One. 2021 Jul 26;16(7):e0255266. doi: 10.1371/journal.pone.0255266. eCollection 2021.

ABSTRACT

INTRODUCTION: A major complication of sepsis is the development of acute kidney injury (AKI). Recently, it was shown that intracellular actin released from damaged tissues appears in the urine of patients with multiple organ dysfunction syndrome. Our aims were to measure urinary actin (u-actin) concentrations of septic and control patients and to test if u-actin levels could predict AKI and mortality.

METHODS: Blood and urine samples were collected from septic and sepsis-related AKI patients at three time points (T1-3): T1: within 24 hours after admission; T2: second day morning; T3: third day morning of follow-up. Patients with malignancies needing palliative care, end-stage renal disease or kidney transplantation were excluded. Serum and u-actin levels were determined by quantitative Western blot. Patients were categorized by the Sepsis-3 and KDIGO AKI classifications.

RESULTS: In our study, 17 septic, 43 sepsis-induced AKI and 24 control patients were enrolled. U-actin levels were higher in septic patients compared with controls during follow-up (p<0.001). At T1, the septic and sepsis-related AKI groups also showed differences (p<0.001), yet this increase was not statistically significant at T2 and T3. We also detected significantly elevated u-actin concentrations in AKI-2 and AKI-3 septic patients compared with AKI-1 septic patients (p<0.05) at T1 and T3, along with a significant increase in AKI-2 septic patients compared with AKI-1 septic patients at T2 (p<0.01). This tendency remained the same when referring u-actin to urine creatinine. Parameters of first-day septic patient samples could discriminate AKI from non-AKI state (AUC ROC, p<0.001): u-actin: 0.876; se-creatinine: 0.875. Derived cut-off value for u-actin was 2.63 μg/L (sensitivity: 86.0%, specificity: 82.4%).

CONCLUSION: U-actin may be a complementary diagnostic biomarker to se-creatinine in sepsis-related AKI while higher u-actin levels also seem to reflect the severity of AKI. Further investigations may elucidate the importance of u-actin release in sepsis-related AKI.

PMID:34310652 | DOI:10.1371/journal.pone.0255266

Categories
Nevin Manimala Statistics

University students’ strategies of coping with stress during the coronavirus pandemic: Data from Poland

PLoS One. 2021 Jul 26;16(7):e0255041. doi: 10.1371/journal.pone.0255041. eCollection 2021.

ABSTRACT

The COVID-19 pandemic has changed the functioning of universities worldwide. In Poland, the transfer to online teaching was announced without prior warning, which radically changed students’ daily functioning. This situation clearly showed the students’ helplessness and difficulties with coping with this new, stressful situation, highlighted in many previous studies. A sudden and far-reaching change in daily functioning caused anxiety, depression, and stress in this group. Thus, from a pedagogical and psychological point of view, it is pertinent to examine the students’ strategies of coping with stress caused by the COVID-19 pandemic. To this end, in 2020, a sample of Polish students was anonymously measured using the Mini-COPE questionnaire. Data was gathered from 577 students from 17 universities. The statistical analysis showed that during the coronavirus pandemic, Polish students most often used the coping strategies of: acceptance, planning, and seeking emotional support. Such factors as age, gender, and place of residence influenced the choice of specific strategies of coping with stress during the COVID-19 pandemic. The results also showed that the youngest students had the lowest coping skills. The results allow for concluding that the students’ maladaptive strategies of coping with stress, especially during the pandemic, may result in long-term consequences for their psychophysiological health and academic achievements. Thus, based on the current results and on the participatory model of intervention, a support program for students is proposed which would involve psychological, organizational, and instrumental support.

PMID:34310649 | DOI:10.1371/journal.pone.0255041

Categories
Nevin Manimala Statistics

Modeling the survival times of the COVID-19 patients with a new statistical model: A case study from China

PLoS One. 2021 Jul 26;16(7):e0254999. doi: 10.1371/journal.pone.0254999. eCollection 2021.

ABSTRACT

Over the past few months, the spread of the current COVID-19 epidemic has caused tremendous damage worldwide, and unstable many countries economically. Detailed scientific analysis of this event is currently underway to come. However, it is very important to have the right facts and figures to take all possible actions that are needed to avoid COVID-19. In the practice and application of big data sciences, it is always of interest to provide the best description of the data under consideration. The recent studies have shown the potential of statistical distributions in modeling data in applied sciences, especially in medical science. In this article, we continue to carry this area of research, and introduce a new statistical model called the arcsine modified Weibull distribution. The proposed model is introduced using the modified Weibull distribution with the arcsine-X approach which is based on the trigonometric strategy. The maximum likelihood estimators of the parameters of the new model are obtained and the performance these estimators are assessed by conducting a Monte Carlo simulation study. Finally, the effectiveness and utility of the arcsine modified Weibull distribution are demonstrated by modeling COVID-19 patients data. The data set represents the survival times of fifty-three patients taken from a hospital in China. The practical application shows that the proposed model out-classed the competitive models and can be chosen as a good candidate distribution for modeling COVID-19, and other related data sets.

PMID:34310646 | DOI:10.1371/journal.pone.0254999

Categories
Nevin Manimala Statistics

Helicobacter pylori eradication with a clarithromycin-based triple therapy in elderly patients: a case-control study

Panminerva Med. 2021 Jul 26. doi: 10.23736/S0031-0808.21.04500-6. Online ahead of print.

ABSTRACT

BACKGROUND: Helicobacter pylori (H. pylori), main agents of several gastroduodenal diseases, represents a therapeutic challenge. Since the influence of age on the success of bacterial treatment remains uncertain, in this case-control study we assessed the efficacy of a standard H. pylori eradication therapy among elderly patients.

METHODS: In this retrospective study, a total of 361 naïve patients (194 males, mean age 79.8± 3.4 years) aged more than 65 years and treated with a triple therapy regimen comprising a standard dose of omeprazole twice daily, amoxicillin 1g twice daily and clarithromycin 500 mg twice daily, for 7, 10 or 14 days, were included. They were compared with naïve patients, younger than 65 years (mean age 43± 2.7 years). Since in the year 2017, we began to use the three-in-one single capsule bismuth-containing quadruple therapy, the search was ended at 31 December 2016.

RESULTS: Overall, H. pylori eradication rate in the intention-to-treat (ITT) analysis, was 70.9% (256/361) among elderly patients versus 70.9% (256/361) among young patients. Dividing by treatment duration, among elderly patients, eradication was obtained in 78.1% (50/64), 69.1% (139/201) and 69.7% (67/96) elderly patients within 7-day, 10-day and 14-day regimens, respectively, without statistical difference. Out of 361 elderly patients, 11 were excluded from the per protocol (PP) analysis because of discontinuations (7 for adverse events). One subject discontinued treatment among young patients. Also the PP analysis showed no statistical difference, with an eradication rate of 73.1% (256/350) among elderly patients versus 71.1% (256/360) among young patients.

CONCLUSIONS: Elderly does not affect efficacy or safety of a clarithromycin-based triple therapy for H. pylori eradication.

PMID:34309333 | DOI:10.23736/S0031-0808.21.04500-6

Categories
Nevin Manimala Statistics

Are electrocardiograms correctly interpreted by pediatric residents? Experience before and after an educational intervention in a teaching hospital

Arch Argent Pediatr. 2021 Aug;119(4):273-278. doi: 10.5546/aap.2021.eng.273.

ABSTRACT

The incidence of pediatric cardiac arrest is unknown; its main etiologies are congenital heart diseases, cardiomyopathies, and ventricular arrhythmias. An electrocardiogram (ECG) is a diagnostic method that may allow to detect them in an early manner and reduce morbidity and mortality. The objective of this study was to describe pediatric residents’ skills to determine if an ECG was normal or abnormal and make an accurate electrocardiographic diagnosis before and after an educational intervention. First-year pediatric residents participated in this study. An assessment including 12 ECG tracings was done before and after an educational module, and scores were compared using the t-test for paired data. No differences were observed between both assessments regarding the interpretation of ECG as normal or abnormal (p: 0.42). However, a statistically significant difference was observed in definitive diagnoses (p < 0.002). Definitive electrocardiographic diagnoses improved after the implementation of an educational module.

PMID:34309305 | DOI:10.5546/aap.2021.eng.273

Categories
Nevin Manimala Statistics

Improvement in communication during patient handoff between areas from a children’s hospital

Arch Argent Pediatr. 2021 Aug;119(4):259-265. doi: 10.5546/aap.2021.eng.259.

ABSTRACT

INTRODUCTION: Patient handoff is an interactive process including data communication and responsible transfer in order to safely maintain the continuity of care. Failure in this process may result in inadequate care and favor the occurrence of errors.

OBJECTIVE: To implement a standardized instrument for patient handoff from the intensive care unit (ICU) to the intermediate-medium care unit (IMCU), and compare communication between health care providers before and after the intervention.

POPULATION AND METHODS: Before-and-after study conducted at Hospital de Pediatría “Prof. Dr. Juan P. Garrahan.” The intervention consisted in a written handoff form. The pre-intervention sample included patients transferred from ICUs to IMCUs between October 1st and October 31st, 2015. The post-intervention sample included patients transferred between March 1st and March 31st, 2016. A total of 4 IMCUs and 3 ICUs participated in the study. The main study variable was the written part of the handoff; in particular, whether it was timely and complete.

RESULTS: A total of 50 handoffs were analyzed for each stage. With the written handoff, there was an increase in the communication of clinical data in 88 % of variables (oral communication between physicians, treating physician, therapeutic adequacy, diagnosis, course, etc.); the difference was statistically significant.

CONCLUSION: After implementing the tool, there was an improvement in the transfer of patient clinical data relevant to the safe continuity of care.

PMID:34309302 | DOI:10.5546/aap.2021.eng.259

Categories
Nevin Manimala Statistics

Concordance of testicular measurement in male adolescents with three methods of orchidometry

Arch Argent Pediatr. 2021 Aug;119(4):251-258. doi: 10.5546/aap.2021.eng.251.

ABSTRACT

INTRODUCTION: The Prader orchidometer is the standard method used to measure testicular volume (TV) in children and adolescents.

OBJECTIVE: To assess the concordance in the estimation of TV and puberty onset with the Prader, Chipkevitch, and Sotos orchidometric techniques.

METHODS: Cross-sectional descriptive study conducted among male children and adolescents aged 9-20 years. For each adolescent, TV was measured with the methods by Prader (gold standard), Chipkevitch (graphic model), and Sotos (measurement of testicular width with a plastic ruler and use of a formula equivalent to the ellipsoid equation). Male children and adolescents with urogenital conditions and disorders affecting testicular growth were excluded. Kappa statistics was used to determine concordance among methods for puberty onset, and intraclass correlation coefficient (ICC) and Bland-Altman (B&A) plots for TV.

RESULTS: In total, 377 healthy males were included. Regarding the concordance for TV (mL), the Prader-Chipkevitch comparison obtained an ICC of 0.994 and a p< 0.001; while the Prader- Soto comparison obtained an ICC of 0.312 and a p< 0.001. With the B&A plots, mean differences were close to 0 mL in the Prader-Chipkevitch comparison and close to 8 mL in the Prader- Sotos comparison. Concordance for puberty onset obtained a kappa value of 0.93 and 0.75 in the Prader-Chipkevitch and Prader-Sotos comparisons, respectively.

CONCLUSION: The Prader and Chipkevitch orchidometers show an excellent concordance in estimating TV and puberty onset; therefore, both methods could be used interchangeably in the daily care of male adolescents. The Sotos method showed a high concordance in estimating pubertal onset, but low in measuring TV.

PMID:34309301 | DOI:10.5546/aap.2021.eng.251