Determining the Most Valuable Clinical Variables : a Stepwise Multiple Logistic Regression Program

1980 ◽  
Vol 19 (01) ◽  
pp. 42-49 ◽  
Author(s):  
B. W. Brown ◽  
C. Engelhard ◽  
J. Haipern ◽  
J. F. Fries ◽  
L. S. Coles
Keyword(s):  
Logistic Regression ◽  
Predictor Variable ◽  
Clinical Problem ◽  
Relevant Information ◽  
Stepwise Logistic Regression ◽  
Multiple Logistic Regression ◽  
Binary Outcome ◽  
Medical Problems ◽  
Valuable Predictor ◽  
Data Banks

In solving a clinical problem of diagnosis, prognosis, or treatment choice, a physician must select from among a large group of possible tests. In general, an ordering exists specifying which tests are most valuable in providing relevant information concerning the problem on hand. The computer program package to be described (MW) extracts appropriate data from the ARAMIS data banks and then analyzes the data by stepwise logistic regression. A binary outcome (diagnosis, prognostic event, or treatment response) is sequentially associated with possible tests, and the most powerful combination of tests is identified. For example, the most valuable predictor variable of early mortality in SLE is proteinuria, followed sequentially by anemia and absence of arthritis. Experience with these techniques suggests : 1. optimal certainty is usually reached after only three or four tests; 2. several different test sequences may lead to the same level of certainty; 3. diagnosis may usually be ascertained with greater certainty than prognosis; 4. many medical problems contain considerable non-reducible uncertainty; 5. a relatively small group of tests are typically found among the most powerful; 6. results are consistent across several patient populations; 7. results are largely independent of the particular statistic employed. These observations suggest strategies for maximizing information while minimizing risk and expense.

Selection of Predictor Variables for Pneumonia Using Neural Networks and Genetic Algorithms

2005 ◽  
Vol 44 (01) ◽  
pp. 89-97 ◽  
Author(s):  
B. S. Gerber ◽  
T. G. Tape ◽  
R. S. Wigton ◽  
P. S. Heckerling
Keyword(s):  
Genetic Algorithm ◽  
Neural Networks ◽  
Logistic Regression ◽  
Fitness Function ◽  
Predictor Variable ◽  
Community Acquired Pneumonia ◽  
Stepwise Logistic Regression ◽  
Training Cohort ◽  
Input Layer ◽  
Roc Area

Summary Background: Artificial neural networks (ANN) can be used to select sets of predictor variable that incorporate nonlinear interactions between variables. We used a genetic algorithm, with selection based on maximizing network accuracy and minimizing network input-layer cardinality, to evolve parsimonious sets of variables for predicting community-acquired pneumonia among patients with respiratory complaints. Methods: ANN were trained on data from 1044 patients in a training cohort, and were applied to 116 patients in a testing cohort. Chromosomes with binary genes representing input-layer variables were operated on by crossover recombination, mutation, and probabilistic selection based on a fitness function incorporating both network accuracy and input-layer cardinality. Results: The genetic algorithm evolved best 10-variable sets that discriminated pneumonia in the training cohort (ROC areas, 0.838 for selection based on average cross entropy (ENT); 0.954 for selection based on ROC area (ROC)), and in the testing cohort (ROC areas, 0.847 for ENT selection; 0.963 for ROC selection), with no significant differences between cohorts. Best variable sets based on the genetic algorithm using ROC selection discriminated pneumonia more accurately than variable sets based on stepwise neural networks (ROC areas, 0.954 versus 0.879, p = 0.030), or stepwise logistic regression (ROC areas, 0.954 versus 0.830, p = 0.000). Variable sets of lower cardinalities were also evolved, which also accurately discriminated pneumonia. Conclusion: Variable sets derived using a genetic algorithm for neural networks accurately discriminated pneumonia from other respiratory conditions, and did so with greater accuracy than variables derived using stepwise neural networks or logistic regression in some cases.

scholarly journals Prediction of Endometrial Hyperplasia and Cancer among Premenopausal Women with Abnormal Uterine Bleeding

2019 ◽  
Vol 2019 ◽  
pp. 1-6 ◽  
Author(s):  
Luca Giannella ◽  
Lillo Bruno Cerami ◽  
Tiziano Setti ◽  
Ezio Bergamini ◽  
Fausto Boselli
Keyword(s):  
Risk Factors ◽  
Logistic Regression ◽  
Prediction Model ◽  
Endometrial Hyperplasia ◽  
Premenopausal Women ◽  
Abnormal Uterine Bleeding ◽  
Uterine Bleeding ◽  
Stepwise Logistic Regression ◽  
Simultaneous Presence ◽  
Endometrial Pathology

Objective. To create a prediction model including clinical variables for the prediction of premalignant/malignant endometrial pathology in premenopausal women with abnormal uterine bleeding (AUB). Methods. This is an observational retrospective study including 240 premenopausal women with AUB referred to diagnostic hysteroscopy. Based on the presence of endometrial hyperplasia (EH) or cancer (EC), the women were divided into cases (EH/EC) and controls (no EH/EC). Univariate, stepwise logistic regression and ROC curve analysis were performed. Results. 12 women had EH/EC (5%). Stepwise logistic regression analysis showed that EH/EC associated significantly with BMI ≥ 30 (OR=7.70, 95% CI 1.90 to 31.17), diabetes (OR=9.71, 95% CI 1.63 to 57.81), and a thickened endometrium (OR=1.20, 95% CI 1.08 to 1.34, criterion > 11 mm). The AUC was 0.854 (95% confidence intervals 0.803 to 0.896, p<0.0001). Considering the pretest probability for EH/EC of 5%, the prediction model with a positive likelihood ratio of 8.14 showed a posttest probability of 30%. The simultaneous presence of two or three risk factors was significantly more common in women with EH/EC than controls (50% vs. 6.6 and 25% vs. 0%, respectively, p<0.0001). Conclusion. When premenopausal vaginal bleeding occurs in diabetic obese women with ET > 11 mm, the percentage of premalignant/malignant endometrial pathology increases by 25%. It is likely that the simultaneous presence of several risk factors is necessary to significantly increase the probability of endometrial pathology.

Experimental study on the factors affecting squeak noise occurrence in automotive suspension bushings

2021 ◽  
pp. 095440702110241
Author(s):  
Byunghyun Kang ◽  
Cheol Choi ◽  
Daeun Sung ◽  
Seongho Yoon ◽  
Byoung-Ho Choi
Keyword(s):  
Neural Network ◽  
Logistic Regression ◽  
Influencing Factors ◽  
Frictional Force ◽  
Prediction Models ◽  
Statistical Tests ◽  
Network Models ◽  
Multiple Logistic Regression ◽  
Neural Network Models ◽  
Automotive Suspension

In this study, friction tests are performed, via a custom-built friction tester, on specimens of natural rubber used in automotive suspension bushings. By analyzing the problematic suspension bushings, the eleven candidate factors that influence squeak noise are selected: surface lubrication, hardness, vulcanization condition, surface texture, additive content, sample thickness, thermal aging, temperature, surface moisture, friction speed, and normal force. Through friction tests, the changes are investigated in frictional force and squeak noise occurrence according to various levels of the influencing factors. The degree of correlation between frictional force and squeak noise occurrence with the factors is determined through statistical tests, and the relationship between frictional force and squeak noise occurrence based on the test results is discussed. Squeak noise prediction models are constructed by considering the interactions among the influencing factors through both multiple logistic regression and neural network analysis. The accuracies of the two prediction models are evaluated by comparing predicted and measured results. The accuracies of the multiple logistic regression and neural network models in predicting the occurrence of squeak noise are 88.2% and 87.2%, respectively.

Association between Procalcitonin and Acute Kidney Injury in Patients with Bacterial Septic Shock

2021 ◽  
pp. 1-10
Author(s):  
Guang Fu ◽  
Hai-chao Zhan ◽  
Hao-li Li ◽  
Jun-fu Lu ◽  
Yan-hong Chen ◽  
...  
Keyword(s):  
Septic Shock ◽  
Acute Kidney Injury ◽  
Logistic Regression ◽  
Piecewise Linear ◽  
Kidney Injury ◽  
Multiple Logistic Regression ◽  
Primary Analysis ◽  
Female Patients ◽  
Potential Biomarker ◽  
The Relationship

Objective: The objective of this study was to assess the relationship between serum procalcitonin (PCT) and acute kidney injury (AKI) induced by bacterial septic shock. Methods: A retrospective study was designed which included patients who were admitted to the ICU from January 2015 to October 2018. Multiple logistic regression and receiver operating characteristic (ROC) as well as smooth curve fitting analysis were used to assess the relationship between the PCT level and AKI. Results: Of the 1,631 patients screened, 157 patients were included in the primary analysis in which 84 (53.5%) patients were with AKI. Multiple logistic regression results showed that PCT (odds ratio [OR] = 1.017, 95% confidence interval [CI] 1.009–1.025, p < 0.001) was associated with AKI induced by septic shock. The ROC analysis showed that the cutoff point for PCT to predict AKI development was 14 ng/mL, with a sensitivity of 63% and specificity 67%. Specifically, in multivariate piecewise linear regression, the occurrence of AKI decreased with the elevation of PCT when PCT was between 25 ng/mL and 120 ng/mL (OR 0.963, 95% CI 0.929–0.999; p = 0.042). The AKI increased with the elevation of PCT when PCT was either <25 ng/mL (OR 1.077, 95% CI 1.022–1.136; p = 0.006) or >120 ng/mL (OR 1.042, 95% CI 1.009–1.076; p = 0.013). Moreover, the PCT level was significantly higher in the AKI group only in female patients aged ≤75 years (p = 0.001). Conclusions: Our data revealed a nonlinear relationship between PCT and AKI in septic shock patients, and PCT could be used as a potential biomarker of AKI in female patients younger than 75 years with bacterial septic shock.

scholarly journals Adjacent digit fingerprint white line count differences: a pointer to sexual dimorphism for forensic application

2019 ◽  
Vol 9 (1) ◽  
Author(s):  
Magaji Garba Taura ◽  
Lawan Hassan Adamu ◽  
Abdullahi Yusuf Asuku ◽  
Kabiru Bilkisu Umar ◽  
Musa Abubakar
Keyword(s):  
Logistic Regression ◽  
Sexual Dimorphism ◽  
Binary Logistic Regression ◽  
Stepwise Logistic Regression ◽  
Regression Analyses ◽  
Forensic Application ◽  
White Line ◽  
Study Population ◽  
Right And Left Hands

Abstract Background Sex determination is one of the leading criterion in identification and verification of an individual. However, the potential roles of differences in adjacent fingerprint white line count (FWLC) in sex inference are not well elucidated in the literature especially among Hausa population. The study was conducted to determine sexual dimorphism and predict sex using adjacent digit FWLC difference (adj. DFWLCD) among Hausa population of Kano state, Nigeria. Methods The study population involved 300 participants. FWLC was determined from a plain fingerprint captured using live scanner. The formula for adj. DFWLCD of thumb and fifth digit is dR15 for right hand. The same applied for possible combination in cephalocaudal direction. Mann-Whitney and t tests were used for comparison of variables between sexes. Binary logistic regression analyses were employed for determination of sex. Results We observed a significantly larger adj. DFWLCD in males compared with females in most of the digit combination. A significant sexual dimorphism was observed in most of the adj. DFWLCD involving ring digit in both right (dR14, dR24, and dR34) and left (dL14, dL24, and dL34). The best discrimination was observed in adjacent FWLC difference of second and fourth digits in both right and left digits (dR24 and dL24). This was further supported by stepwise logistic regression analyses. Conclusion The adj. DFWLCD exhibits sexual dimorphism. The best prediction potentials were found to be dR24 and dL24 for right and left hands respectively.

scholarly journals Factors Affecting Salt Reduction Measure Adoption among Chinese Residents

2021 ◽  
Vol 18 (2) ◽  
pp. 445
Author(s):  
Zeying Huang ◽  
Di Zeng
Keyword(s):  
Logistic Regression ◽  
Food Intake ◽  
Nutrition Education ◽  
Online Survey ◽  
Adaptive Lasso ◽  
Stepwise Logistic Regression ◽  
Salt Reduction ◽  
Salt Restriction ◽  
Salt Diet ◽  
Factors Affecting

China has the highest mortality rate caused by diseases and conditions associated with its high-salt diet. Since 2016, China has initiated a national salt reduction campaign that aims at promoting the usage of salt information on food labels and salt-restriction spoons and reducing condiment and pickled food intake. However, factors affecting individuals’ decisions to adopt these salt reduction measures remain largely unknown. By comparing the performances of logistic regression, stepwise logistic regression, lasso logistic regression and adaptive lasso logistic regression, this study aims to fill this gap by analyzing the adoption behaviour of 1610 individuals from a nationally representative online survey. It was found that the practices were far from adopted and only 26.40%, 22.98%, 33.54% and 37.20% reported the adoption of labelled salt information, salt-restriction spoons, reduced condiment use in home cooking and reduced pickled food intake, respectively. Knowledge on salt, the perceived benefits of salt reduction, participation in nutrition education and training programs on sodium reduction were positively associated with using salt information labels. Adoption of the other measures was largely explained by people’s awareness of hypertension risks and taste preferences. It is therefore recommended that policy interventions should enhance Chinese individuals’ knowledge of salt, raise the awareness of the benefits associated with a low-salt diet and the risks associated with consuming excessive salt and reshape their taste choices.

scholarly journals Multidisciplinary Approach for the Management of Complex Hepatic Injuries AAST-OIS Grades IV—V: A Prospective Study

2007 ◽  
Vol 96 (3) ◽  
pp. 214-220 ◽  
Author(s):  
J. A. Asensio ◽  
P. Petrone ◽  
L. Garcí-Núñez ◽  
B. Kimbrell ◽  
E. Kuncir
Keyword(s):  
Emergency Department ◽  
Logistic Regression ◽  
Blood Loss ◽  
Blood Cells ◽  
Multidisciplinary Approach ◽  
Stepwise Logistic Regression ◽  
Packed Red Blood Cells ◽  
Emergency Department Thoracotomy ◽  
Estimated Blood Loss ◽  
Hepatic Injuries

Background: Complex hepatic injuries grades IV—V are highly lethal. The objective of this study is to assess the multidisciplinary approach for their management and to evaluate if survival could be improved with this approach. Study Design: Prospective 54-month study of all patients sustaining hepatic injuries grades IV—V managed operatively at a Level I Trauma Center. Main outcome measure: survival. Statistical analysis: univariate and stepwise logistic regression. Results: Seventy-five patients sustained penetrating (47/63%) and blunt (28/37%) injuries. Seven (9%) patients underwent emergency department thoracotomy with a mortality of 100%. Out of the 75 patients, 52 (69%) sustained grade IV, and 23 (31%) grade V. The estimated blood loss was 3,539±-3,040 ml. The overall survival was 69%, adjusted survival excluding patients requiring emergency department thoracotomy was 76%. Survival stratified to injury grade: grade IV 42/52–81%, grade V 10/23–43%. Mortality grade IV versus V injuries (p <0.002; RR 2.94; 95% CI 1.52–5.70). Risk factors for mortality: packed red blood cells transfused in operating room (p=0.024), estimated blood loss (p<0.001), dysryhthmia (p<0.0001), acidosis (p=0.051), hypothermia (p=0.04). The benefit of angiography and angioembolization indicated: 12% mortality (2/17) among those that received it versus a 36% mortality (21/58) among those that did not (p=0.074; RR 0.32; 95% CI 0.08–1.25). Stepwise logistic regression identified as significant independent predictors of outcome: estimated blood loss (p=0.0017; RR 1.24; 95% CI 1.08–1.41) and number of packed red blood cells transfused in the operating room (p=0.0358; RR 1.16; 95% CI 1.01–1.34). Conclusions: The multidisciplinary approach to the management of these severe grades of injuries appears to improve survival in these highly lethal injuries. A prospective multi-institutional study is needed to validate this approach.

Sign in / Sign up

Export Citation Format

Share Document