scholarly journals Handwriting identification using random forests and score‐based likelihood ratios

2021 ◽  
Author(s):  
Madeline Quinn Johnson ◽  
Danica M. Ommen
Keyword(s):  
Random Forests ◽  
Likelihood Ratios

Sentence Repetition: A Clinical Marker for Developmental Language Disorder in Danish

2019 ◽  
Vol 62 (12) ◽  
pp. 4450-4463
Author(s):  
Rikke Vang Christensen
Keyword(s):  
Language Disorder ◽  
Future Research ◽  
Typically Developing ◽  
Developmental Language Disorder ◽  
Older Children ◽  
Likelihood Ratios ◽  
Clinical Marker ◽  
Sentence Repetition ◽  
Good Identification ◽  
Syntactic Properties

Purpose The aim of the study was to explore the potential of performance on a Danish sentence repetition (SR) task—including specific morphological and syntactic properties—to identify difficulties in children with developmental language disorder (DLD) relative to typically developing (TD) children. Furthermore, the potential of the task as a clinical marker for Danish DLD was explored. Method SR performance of children with DLD aged 5;10–14;1 (years;months; n = 27) and TD children aged 5;3–13;4 ( n = 87) was investigated. Results Compared to TD same-age peers, children with DLD were less likely to repeat the sentences accurately but more likely to make ungrammatical errors with respect to verb inflection and use of determiners and personal pronouns. Younger children with DLD also produced more word order errors that their TD peers. Furthermore, older children with DLD performed less accurately than younger TD peers, indicating that the SR task taps into morphosyntactic areas of particular difficulty for Danish children with DLD. The classification accuracy associated with SR performance showed high levels of sensitivity and specificity (> 90%) and likelihood ratios indicating good identification potential for clinical and future research purposes. Conclusion SR performance has a strong potential for identifying children with DLD, also in Danish, and with a carefully designed SR task, performance has potential for revealing morphosyntactic difficulties. Supplemental Material https://doi.org/10.23641/asha.10314437

Clausal Density Between Ages 4 and 9 Years for the Edmonton Narrative Norms Instrument: Reference Data and Psychometric Properties

2020 ◽  
pp. 1-15
Author(s):  
Ling-Yu Guo ◽  
Phyllis Schneider ◽  
William Harrison
Keyword(s):  
Diagnostic Accuracy ◽  
Psychometric Properties ◽  
Language Impairment ◽  
Reference Data ◽  
Reliability And Validity ◽  
Correlation Coefficients ◽  
Criterion Validity ◽  
Likelihood Ratios ◽  
Generation Task ◽  
Sensitivity Specificity

Purpose This study provided reference data and examined psychometric properties for clausal density (CD; i.e., number of clauses per utterance) in children between ages 4 and 9 years from the database of the Edmonton Narrative Norms Instrument (ENNI). Method Participants in the ENNI database included 300 children with typical language (TL) and 77 children with language impairment (LI) between the ages of 4;0 (years;months) and 9;11. Narrative samples were collected using a story generation task, in which children were asked to tell stories based on six picture sequences. CD was computed from the narrative samples. The split-half reliability, concurrent criterion validity, and diagnostic accuracy were evaluated for CD by age. Results CD scores increased significantly between ages 4 and 9 years in children with TL and those with LI. Children with TL produced higher CD scores than those with LI at each age level. In addition, the correlation coefficients for the split-half reliability and concurrent criterion validity of CD scores were all significant at each age level, with the magnitude ranging from small to large. The diagnostic accuracy of CD scores, as revealed by sensitivity, specificity, and likelihood ratios, was poor. Conclusions The finding on diagnostic accuracy did not support the use of CD for identifying children with LI between ages 4 and 9 years. However, given the attested reliability and validity for CD, reference data of CD from the ENNI database can be used for evaluating children's difficulties with complex syntax and monitoring their change over time. Supplemental Material https://doi.org/10.23641/asha.13172129

Diagnostic accuracy of magnetic resonance angiography for acute pulmonary embolism - a systematic review and meta-analysis

VASA ◽  
2016 ◽  
Vol 45 (2) ◽  
pp. 149-154 ◽  
Author(s):  
Jie Li ◽  
Lei Feng ◽  
Jiangbo Li ◽  
Jian Tang
Keyword(s):  
Pulmonary Embolism ◽  
Magnetic Resonance ◽  
Magnetic Resonance Angiography ◽  
Diagnostic Accuracy ◽  
Acute Pulmonary Embolism ◽  
Meta Analysis ◽  
Likelihood Ratios ◽  
Summary Receiver Operating Characteristic ◽  
Acute Pe ◽  
Limited Sensitivity

Abstract. Background: The aim of this meta-analysis was to evaluate the diagnostic accuracy of magnetic resonance angiography (MRA) for acute pulmonary embolism (PE). Methods: A systematic literature search was conducted that included studies from January 2000 to August 2015 using the electronic databases PubMed, Embase and Springer link. The summary receiver operating characteristic (SROC) curve, sensitivity, specificity, positive likelihood ratios (PLR), negative likelihood ratios (NLR), and diagnostic odds ratio (DOR) as well as the 95 % confidence intervals (CIs) were calculated to evaluate the diagnostic accuracy of MRA for acute PE. Meta-disc software version 1.4 was used to analyze the data. Results: Five studies were included in this meta-analysis. The pooled sensitivity (86 %, 95 % CI: 81 % to 90 %) and specificity (99 %, 95 % CI: 98 % to 100 %) demonstrated that MRA diagnosis had limited sensitivity and high specificity in the detection of acute PE. The pooled estimate of PLR (41.64, 95 % CI: 17.97 to 96.48) and NLR (0.17, 95 % CI: 0.11 to 0.27) provided evidence for the low missed diagnosis and misdiagnosis rates of MRA for acute PE. The high diagnostic accuracy of MRA for acute PE was demonstrated by the overall DOR (456.51, 95 % CI: 178.38 - 1168.31) and SROC curves (AUC = 0.9902 ± 0.0061). Conclusions: MRA can be used for the diagnosis of acute PE. However, due to limited sensitivity, MRA cannot be used as a stand-alone test to exclude acute PE.

Comparison of Rheumatological Diagnoses by a Bayesian Program and by Physicians

1991 ◽  
Vol 30 (03) ◽  
pp. 187-193 ◽  
Author(s):  
H. J. Moens ◽  
J. K. van der Korst
Keyword(s):  
Receiver Operating Characteristics ◽  
Bayesian Decision ◽  
Operating Characteristics ◽  
Likelihood Ratios ◽  
Improve Performance ◽  
Mathematical Basis ◽  
Human Expert ◽  
Test Population ◽  
Sensitivity Specificity ◽  
Diagnostic Outcome

AbstractA Bayesian decision support system was developed for the diagnosis of rheumatic disorders. Knowledge in this system is represented as evidential weights of findings. Simple weights were calculated as the logarithm of likelihood ratios on the basis of 1,000 consecutive patients from a rheumatological clinic. The effect of various methods to improve performance of the system by modification of the weights was studied. Three methods had a mathematical basis; a fourth consisted of weights adapted by a human expert, which allowed inclusion of diagnostic rules such as defined in widely accepted criteria sets. The system’s performance was measured in a test population of 570 different cases from the same clinic and compared with predictions of diagnostic outcome made by rheumatologists. The weights from a human expert gave optimal results (sensitivity 65% and specificity 96%), that were close to the physicians’ predictions (sensitivity 64% and specificity 98%). The methods to measure the performance of the various models used in this study emphasize sensitivity, specificity and the use of receiver operating characteristics.

Proposal for Objective Evaluation of the Performance of Various Functional APC-resistance Tests in Genotyped Patients

1997 ◽  
Vol 78 (05) ◽  
pp. 1360-1365 ◽  
Author(s):  
G Freyburger ◽  
S Javorschi ◽  
S Labrouche ◽  
P Bernard
Keyword(s):  
Protein S ◽  
Objective Evaluation ◽  
Original Method ◽  
Screening Methods ◽  
Factor Xii ◽  
Factor V ◽  
Likelihood Ratios ◽  
Modified Method ◽  
Apc Resistance ◽  
Two Parameters

SummaryThe aim of the present study was to evaluate the relative performance of five screening methods for APC resistance caused by the factor V:Q506 mutation: the original method Coatest® APC™ Resistance Chromogenix, a modified method using the same reagents but a predilution 1+4 of the plasma in a factor V deficient plasma from Stago (Stago deficient V) or from Chromogenix (V-DEF Plasma), the Coatest® APC™ Resistance V (Chromogenix), and Accelerimat™ from bioMérieux. Normalization was done against a pool of normal plasmas for the methods from Chromogenix. The study included 350 subjects, 219 were genotyped (174 FV:R506R, 42 FV:Q506R, 3 FV:Q506Q) and most of them were assessed by more than one method. Uncertainty in predicting the FV genotype was evaluated by statistical analysis, which provided a way to quantitate the performance of the different diagnostic approaches. Performance of each test was evaluated by its sensitivity, specificity, R.O.C. curves, positive and negative likelihood ratios (LR), and the overall performance was determined by two parameters derived from the LR curves : the maximum LR value obtained at the crossover of the two curves, and the distance between the two curves for LR = 10. Coatest® APC™ Resistance V and Accelerimat™ were proven to be the methods most able to discriminate for factor V:Q506, while normalization was not shown to improve the screening performance. The original method from Chromogenix was confirmed to undergo many influences (factor XII, PAI-1, thrombin- antithrombin complexes, antithrombin III, hematocrit). Although a very good improvement was provided by the newest methods, they were shown to be influenced by protein S and/or factor V levels in the sample plasma.

DIAGNOSTIC ACCURACY OF ONE SAMPLE OR TWO SAMPLES QUANTITATIVE FECAL IMMUNOCHEMICAL TESTS FOR INTESTINAL NEOPLASIA DETECTION

2020 ◽  
Vol 57 (3) ◽  
pp. 316-322
Author(s):  
Rejane MATTAR ◽  
Sergio Barbosa MARQUES ◽  
Maurício Kazuyoshi MINATA ◽  
Joyce Matie Kinoshita da SILVA-ETTO ◽  
Paulo SAKAI ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Negative Likelihood Ratio ◽  
Occult Blood ◽  
Test Methods ◽  
Advanced Adenoma ◽  
Sample Collection ◽  
Likelihood Ratios ◽  
Intestinal Neoplasia ◽  
Two Samples ◽  
Eiken Chemical

ABSTRACT BACKGROUND: Rectal bleeding is the most important symptom of intestinal neoplasia; thus, tests of occult blood detection in stools are widely used for pre neoplastic lesions and colorectal cancer (CRC) screening. OBJECTIVE: Evaluate the accuracy of OC-Sensor quantitative test (Eiken Chemical, Tokyo, Japan) at cut-off 10 µg Hb/g feces (50 ng/mL) in a cohort of subjects that had to undergo diagnostic colonoscopy, and if more than one sample collected in consecutive days would improve the diagnostic accuracy of the test. METHODS: Patients (mean age 56.3±9.7 years) that underwent colonoscopy prospectively randomly received one (1-sample FIT, FIT 1) or two (2-sample FIT, FIT 2) collection tubes. They collected the stool sample before starting colonoscopy preparation. Samples were analyzed by the OC-Auto Micro 80 (Eiken Chemical, Tokyo, Japan). The performance of FIT 1 and FIT 2 were compared to the colonoscopy findings. RESULTS: Among 289 patients, CRC was diagnosed in 14 (4.8%), advanced adenoma in 37 (12.8%), early adenoma in 71 (24.6%) and no abnormalities in 141 (48.8%). For FIT 1, the sensitivity for CRC was 83.3% (95%CI 36.5-99.1%), for advanced adenoma was 24% (95%CI 10.1-45.5%), with specificity of 86.9% (95%CI 77.3-92.9%). For FIT 2, the sensitivity for CRC was 75% (95%CI 35.6-95.5%), for advanced adenoma was 50% (95%CI 22.3-77.7%), with specificity of 92.9% (95%CI 82.2-97.7%). The positive likelihood ratios were 1.8 (95%CI 0.7-4.4 for FIT 1) and 7.1 (95%CI 2.4-21.4 for FIT 2) for advanced adenoma, and 6.4 (95%CI 3.3-12.3, for FIT 1) and 10.7 (95%CI 3.8-29.8, for FIT 2) for CRC. The negative likelihood ratio were 0.9 (95%CI 0.7-1, for FIT 1) and 0.5 (95%CI 0.3-0.9, for FIT 2) for advanced adenoma, and 0.2 (0.03-1.1, for FIT 1) and 0.3 (0.08-0.9, for FIT 2) for CRC. The differences between FIT 1 and FIT 2 performances were not significant. However, the comparison of the levels of hemoglobin in feces of patients of FIT 1 and FIT 2 showed that the differences between no polyp group and advanced adenoma and CRC were significant. CONCLUSION: The accuracy of OCR Sensor with 10 µg Hb/g feces cut-off was comparable to other reports and two-sample collection improved the detection rate of advanced adenoma, a pre neoplastic condition to prevent CRC incidence.

Predictors of remission from body dysmorphic disorder after internet-delivered cognitive behavior therapy: a machine learning approach

2019 ◽  
Author(s):  
Oskar Flygare ◽  
Jesper Enander ◽  
Erik Andersson ◽  
Brjánn Ljótsson ◽  
Volen Z Ivanov ◽  
...  
Keyword(s):  
Machine Learning ◽  
Logistic Regression ◽  
Random Forests ◽  
Clinical Utility ◽  
Body Dysmorphic Disorder ◽  
Prediction Models ◽  
Behavioral Therapy ◽  
Learning Approach ◽  
Learning Approaches ◽  
Machine Learning Approach

**Background:** Previous attempts to identify predictors of treatment outcomes in body dysmorphic disorder (BDD) have yielded inconsistent findings. One way to increase precision and clinical utility could be to use machine learning methods, which can incorporate multiple non-linear associations in prediction models. **Methods:** This study used a random forests machine learning approach to test if it is possible to reliably predict remission from BDD in a sample of 88 individuals that had received internet-delivered cognitive behavioral therapy for BDD. The random forest models were compared to traditional logistic regression analyses. **Results:** Random forests correctly identified 78% of participants as remitters or non-remitters at post-treatment. The accuracy of prediction was lower in subsequent follow-ups (68%, 66% and 61% correctly classified at 3-, 12- and 24-month follow-ups, respectively). Depressive symptoms, treatment credibility, working alliance, and initial severity of BDD were among the most important predictors at the beginning of treatment. By contrast, the logistic regression models did not identify consistent and strong predictors of remission from BDD. **Conclusions:** The results provide initial support for the clinical utility of machine learning approaches in the prediction of outcomes of patients with BDD. **Trial registration:** ClinicalTrials.gov ID: NCT02010619.

Improved Decoding for LDPC Coded Modulation Systems Using Averaged Log-Likelihood Ratios

2011 ◽  
Vol 30 (8) ◽  
pp. 1845-1848
Author(s):  
Ping Huang ◽  
Ming Jiang ◽  
Chun-ming Zhao
Keyword(s):  
Coded Modulation ◽  
Likelihood Ratios ◽  
Log Likelihood
Sign in / Sign up

Export Citation Format

Share Document