Privacy-Preserving Statistical Analysis by Exact Logistic Regression

2015 ◽  
Author(s):  
David A. Duverle ◽  
Shohei Kawasaki ◽  
Yoshiji Yamada ◽  
Jun Sakuma ◽  
Koji Tsuda
Keyword(s):  
Logistic Regression ◽  
Statistical Analysis ◽  
Privacy Preserving

ASYMMETRIES IN UNIVERSAL GRAMMAR The Role of Method and Statistics

2000 ◽  
Vol 22 (2) ◽  
pp. 209-228 ◽  
Author(s):  
John C. Paolillo
Keyword(s):  
Logistic Regression ◽  
Statistical Analysis ◽  
Statistical Models ◽  
Logistic Regression Model ◽  
Universal Grammar ◽  
Nonnative Speakers ◽  
Potential Explanation ◽  
Differential Learning ◽  
Grammaticality Judgments

Felix (1988) claimed to demonstrate that UG-based knowledge of grammaticality causes nonnative speakers (NNSs) to have more accurate grammaticality judgments on sentences that are ungrammatical according to UG than on those that are grammatical. Birdsong (1994) criticized the methodology employed, noting that it ignores “response bias” (a propensity to judge sentences as ungrammatical) as a potential explanation. Felix and Zobl (1994) dismissed this criticism as merely methodological. In this paper, Birdsong's criticism is upheld by considering a statistical model of the data. At the same time, a more complete logistic regression model allows a fuller statistical analysis, revealing tentative support for the asymmetry claim, as well as differential learning states for different constructions and a tendency toward transfer avoidance. These theoretically significant effects were unnoticed in the earlier discussion of this research. For SLA research on grammaticality judgments to proceed fruitfully, appropriate statistical models need to be considered in designing the research.

scholarly journals PRESEPSIN AND PROCALCITONIN VALUES TO DETERMINE THE PROGNOSIS OF UROSEPSIS

2018 ◽  
Vol 25 (2) ◽  
Author(s):  
Sofyan Rais Addin ◽  
Tarmono Djojodimedjo ◽  
Fikri Rizaldi
Keyword(s):  
Logistic Regression ◽  
Statistical Analysis ◽  
Prospective Study ◽  
Prognostic Value ◽  
Observational Prospective Study ◽  
Regression Test ◽  
The Difference ◽  
Logistic Regression Test ◽  
First Admission

Objective: To evaluate the prognostic value of presepsin and procalcitonin in patient with urosepsis. Material & Method: This is an observational prospective study. Patients who fulfilled the criteria for urosepsis at Soetomo Hospital Surabaya were enrolled. Presepsin and procalcitonin were measured at first admission. All patients were managed according to standard urosepsis therapy. At the 28th day of treatment, patients were evaluated and classified into survivor and non-survivor. The statistical analysis were tested with logistic regression test using software SPSS 23. Results: 30 urosepsis patients were included in this study with average age was 48.3 years (range 21-77 years). There were 23 survivor and 7 non-survivor. Mean presepsin values were higher in non-survivor than in survivor but the difference was not significant (4405 ± 4664 vs 4042 ± 2643, p=0.259). Mean procalcitonin value was significantly higher in non-survivor than in survivor (7.68 ± 6.81 vs 3.27 ± 2.74, p=0.013). Using the cut off value ≥2.24 ng/ml, procalcitonin can predict mortality in 28 days with sensitivity 71.4% and specificity 47.8%. Conclusion: Presepsin can not be used to determine the prognosis of urosepsis patients. Procalcitonin showed a significant correlation with outcome of urosepsis patients so it can be use to determine the prognosis of urosepsis.

scholarly journals Hybrid Neuro-Fractal Analysis of ECG Signals to Predict Ischemia

2021 ◽  
Author(s):  
Hedieh Montazeri
Keyword(s):  
Neural Network ◽  
Logistic Regression ◽  
Statistical Analysis ◽  
Linear Regression ◽  
Fractal Analysis ◽  
Prediction Accuracy ◽  
Clinical Characteristics ◽  
Hybrid Approach ◽  
Binary Logistic Regression ◽  
Ecg Signals

In this thesis, we propose and implement a new hybrid approach using fractal analysis, statistical analysis and neural network computation to build a model for prediction the number of ischemia occurrence based on ECG recordings. The main advantage of the proposed approach over similar earlier related works is that first useful parameters from fractal analysis of the signal are extracted to build a model that includes both clinical characteristics and signal attributes. Statistical analysis such as binary logistic regression and multivariate linear regression are then used to further explore the relation of parameters in order to obtain a more accurate model. We show that the results compare well with those of earlier work and clearly indicate that the augmentation of the above mentioned approaches improves the prediction accuracy.

Female Authorship in Radiology: Trends in the Past Decade in CARJ

2020 ◽  
pp. 084653712094664
Author(s):  
Nicole Li ◽  
Mostafa Alabousi ◽  
Michael N. Patlas
Keyword(s):  
Logistic Regression ◽  
Statistical Analysis ◽  
P Value ◽  
Inclusion Criteria ◽  
Odds Ratios ◽  
The Past ◽  
Single Author ◽  
Canadian Association ◽  
Female Authorship ◽  
Over Time

Purpose: To identify trends in female authorship in the Canadian Association of Radiologists Journal (CARJ) from 2010 to 2019. Methods: We retrieved papers published in the CARJ over a 10-year period, and retrospectively reviewed 602 articles. All articles except editorials and advertisements were included. We categorized the names of the first and last position authors as female or male and excluded articles that had at least one author of which gender was not known. We compared the trends in the first and last position authors of the articles from 2010 to 2019. For statistical analysis, logistic regression was performed with reported odds ratios (ORs), and a P value of <.05 was defined as statistically significant. Results: Five hundred thirteen articles met inclusion criteria. Among them, 23 articles with a single author were classified as having only a first author. 39.8% (204/513) of first authors were female and 26.9% (132/490) of last authors were female. There has been an overall temporal increase in the odds of both the first and last author being female in CARJ publications (OR: 1.11, P = .034). Similarly, the odds a CARJ publication’s first author being female increased over time (OR: 1.07, P = .033). Female last author did not predict female first author (OR: 1.48, P = .056). There was no association identified between female last author and year of publication (OR: 1.04, P = .225). Conclusion: There has been an overall increase in engagement of female authorship in CARJ.

Statistical Analysis of High-Level Features from State of the Union Addresses

2017 ◽  
Vol 8 (2) ◽  
pp. 50-73 ◽  
Author(s):  
Trevor J. Bihl ◽  
Kenneth W. Bauer Jr.
Keyword(s):  
Logistic Regression ◽  
Feature Selection ◽  
Statistical Analysis ◽  
Classification Methods ◽  
State Of The Union ◽  
Recent Interest ◽  
Science Approach ◽  
High Level ◽  
Statistical Clustering ◽  
Delivery Medium

A computational political science approach is taken to analyze the State of the Union Addresses (SUA) from 1790 to 2015. While low-level features, e.g. linguistic characteristics, are commonly used for lexical analysis, the authors herein illustrate the utility of high-level features, e.g. Flesch-Kincaid readability, for knowledge discovery and discrimination between types of speeches. A process is developed and employed to exploit high-level features which employs 1) statistical clustering (k-means) and a literature review to define types of speeches (e.g. written or oral), 2) classification methods via logistic regression to examine the validity of the defined classes, and 3) classifier-based feature selection to determine salient features. Recent interest in the SUA has posited that changes in readability in the SUA are due to declining audience capabilities; however, the authors' results show that changes in readability are a reflection of changes in the SUA delivery medium.

A game-theoretic model for protest in the context of post-communism

2011 ◽  
Vol 44 (1) ◽  
pp. 1-15 ◽  
Author(s):  
Konstantin Ash
Keyword(s):  
Logistic Regression ◽  
Statistical Analysis ◽  
Large Scale ◽  
Theoretic Model ◽  
Regression Analyses ◽  
Game Theoretic ◽  
Game Theoretic Model ◽  
Post Communism ◽  
Eventual Outcome

In recent years, large-scale protests have forced several incumbent governments in former Soviet countries from power. Scholarly examinations of these events have lacked a cohesive explanation of the reasons for the success of certain movements and the failure of others. This study uses prior research on the dynamics of protest to formulate a game-theoretic model for why protest takes place and how its eventual outcome comes about. The model is tested through logistic regression analyses of monthly protest data. The statistical analysis shows that elections, prior protests and government transgressions increase the likelihood of anti-government protests.

scholarly journals User-centric privacy-preserving statistical analysis of ubiquitous health monitoring data

2014 ◽  
Vol 11 (2) ◽  
pp. 525-548 ◽  
Author(s):  
George Drosatos ◽  
Pavlos Efraimidis
Keyword(s):  
Statistical Analysis ◽  
Health Monitoring ◽  
Wearable Sensors ◽  
Privacy Preserving ◽  
Monitoring Data ◽  
Biomedical Data ◽  
Enormous Amount ◽  
User Centric ◽  
Personal Agents ◽  
To Receive

In this paper, we propose a user-centric software architecture for managing Ubiquitous Health Monitoring Data (UHMD) generated from wearable sensors in a Ubiquitous Health Monitoring System (UHMS), and examine how these data can be used within privacy-preserving distributed statistical analysis. Two are the main goals of our approach. First, to enhance the privacy of patients. Second, to decongest the Health Monitoring Center (HMC) from the enormous amount of biomedical data generated by the users? wearable sensors. In our solution personal software agents are used to receive and manage the personal medical data of their owners. Moreover, the personal agents can support privacy-preserving distributed statistical analysis of the health data. To this end, we present a cryptographic protocol based on secure multi-party computations that accept as input current or archived values of users? wearable sensors. We describe a prototype implementation that performs a statistical analysis on a community of independent personal agents. Finally, experiments with up to several hundred agents confirm the viability and the effectiveness of our approach.

Sign in / Sign up

Export Citation Format

Share Document