Associated Area Under The Curve Research Articles

Machine learning for screening and predicting the availability of medications for children: a cross-sectional survey study.

The aim of the study was to explore the factors influencing the availability of medications for children, and establish a machine learning model to provide an empirical basis for the subsequent formulation and improvement of relevant policies. Design: Cross-sectional survey. Setting: 12 provinces, China. Medical doctors from 25 public hospitals were enrolled. All data were randomly divided into a training set and a validation set at a ratio of 7:3. Three prediction models, namely random forest (RF), logistic regression (LR), and extreme gradient boosting (XGBoost), were developed and compared. The receiver operating characteristic curve (ROC) and the associated area under the curve (AUC) were used to evaluate the three models. A nomogram and clinical impact curve (CIC) for availability of medication were developed. Fifteen of 29 factors in the database that were most likely to be selected were considered to establish the prediction model. The XGBoost model (AUC = 0.915) demonstrated better performance than the RF model (AUC = 0.902) and the LR model (AUC = 0.890). According to the Shapley additive explanation values, the five factors that most significantly affected the availability of medications for children in the XGboost model were as follows: the relatively small number of specialized dosage forms for children; unaffordable medications for children; public education on the accessibility and safety of medication for children; uneven distribution of medical resources, leading to insufficient access to medication for children; and years of service as a doctor. The CIC was used to assess the practical applicability of the factor prediction nomogram. The XGBoost model can be used to establish a prediction model to screen the factors associated with the availability of medications for children. The most important contributing factors to the models were the following: the relatively small number of specialized dosage forms for children; unaffordable medications for children; public education on the accessibility and safety of medication for children; uneven distribution of medical resources, leading to insufficient access to medication for children; and years of service as a doctor.

Automatic Detection of Perilunate and Lunate Dislocations on Wrist Radiographs Using Deep Learning.

Delayed or missed diagnosis of perilunate or lunate dislocations can lead to significant morbidity. Advances in computer vision provide an opportunity to improve diagnostic performance. In this study, a deep learning algorithm was used for detection of perilunate and lunate dislocations on lateral wrist radiographs. A total of 435 lateral wrist radiographs were labeled as normal or pathologic (perilunate or lunate dislocation). The lunate in each radiograph was segmented with a rectangular bounding box. Images were partitioned into training and test sets. Two neural networks, consisting of an object detector followed by an image classifier, were applied in series. First, the object detection module was used to localize the lunate. Next, the image classifier performed a binary classification for normal or pathologic. The accuracy, sensitivity, and specificity of the overall system were evaluated. A receiver operating characteristic curve and the associated area under the curve were used to demonstrate the overall performance of the computer vision algorithm. The lunate object detector was 97.0% accurate at identifying the lunate. Accuracy was 98.7% among the subgroup of normal wrist radiographs and 91.3% among the subgroup of wrist radiographs with perilunate/lunate dislocations. The perilunate/lunate dislocation classifier had a sensitivity (recall) of 93.8%, a specificity of 93.3%, and an accuracy of 93.4%. The area under the curve was 0.986. The authors have developed a proof-of-concept computer vision system for diagnosis of perilunate/lunate dislocations on lateral wrist radiographs. This novel deep learning algorithm has potential to improve clinical sensitivity to ultimately prevent delayed or missed diagnosis of these injuries.

Application of amide proton transfer imaging to pretreatment risk stratification of childhood neuroblastoma: comparison with neuron-specific enolase.

The diagnosis and treatment of childhood neuroblastoma (NB) varies with different risk groups, thus requiring accurate preoperative risk assessment. This study aimed to verify the feasibility of amide proton transfer (APT) imaging in risk stratification of abdominal NB in children, and compare it with the serum neuron-specific enolase (NSE). This prospective study enrolled 86 consecutive pediatric volunteers with suspected NB, and all subjects underwent abdominal APT imaging on a 3T magnetic resonance imaging scanner. A 4-pool Lorentzian fitting model was used to mitigate motion artifacts and separate the APT signal from the contaminating ones. The APT values were measured from tumor regions delineated by two experienced radiologists. The one-way analysis of variance, independent-sample t-test, Mann-Whitney U-test, and receiver operating characteristic analysis were performed to evaluate and compare the risk stratification performance of the APT value and serum NSE index-a routine biomarker of NB in clinics. Thirty-four cases (mean age, 38.6±32.4 months; 5 very-low-risk, 5 low-risk, 8 intermediate-risk and 16 high-risk ones) were included in the final analysis. The APT values were significantly higher in high-risk NB (5.80%±1.27%) than in the non-high-risk group (3.88%±1.01%) composed of the other three risk groups (P<0.001). However, there was no significant difference (P=0.18) in NSE levels between the high-risk (93.05±97.14 ng/mL) and non-high-risk groups (41.45±30.99 ng/mL). The associated area under the curve (AUC) of the APT parameter (AUC =0.89) in differentiating high-risk NB from non-high-risk NB was significantly higher (P=0.03) than that of NSE (AUC =0.64). As an emerging non-invasive magnetic resonance imaging technique, APT imaging has a promising prospect for distinguishing high-risk NB from non-high-risk NB in routine clinical applications.

Expression analysis of circulating miR-22, miR-122, miR-217 and miR-367 as promising biomarkers of acute lymphoblastic leukemia.

The role of serum-based biomarkers such as microRNAs in cancer diagnosis has been extensively established. This study aimed to determine the expression levels of bioinformatically selected miRNAs and whether they can be used as biomarkers or a new therapeutic target in patients with acute lymphoblastic leukemia (ALL). The expression levels of serum miR-22, miR-122, miR-217, and miR-367 in 21 ALL patients and 21 healthy controls were measured using quantitative real-time PCR. The receiver operating characteristic (ROC) curve and the associated area under the curve (AUC) was used to assess candidate miRNAs' diagnostic value as a biomarker. The results showed that miR-217 was markedly decreased in patients with ALL compared to controls. Moreover, miR-22, miR-122, and miR-367 were found to be upregulated. Furthermore, ROC analysis showed that serum miR-217 and miR-367 could differentiate ALL patients from healthy individuals, while miR-22 has approximate discriminatory power that requires further investigation. These results provide promising preliminary evidence that circulating miR-217 and miR-367 could be considered potent diagnostic biomarkers and therapeutic goals in this disease.

German translation and validation of the brief Epilepsy Anxiety Survey Instrument (brEASI)

BackgroundAnxiety disorders remain undiagnosed in routine clinical practice in up to two thirds of affected patients with epilepsy despite their significant impact on medical and psychosocial outcomes. The study objective was to translate and validate the German 8-item “brief Epilepsy Anxiety Survey Instrument” (brEASI) to facilitate effective screening for the presence of anxiety disorders in German-speaking patients. MethodsAfter expert translation into German, the brEASI was completed by consecutive adult inpatients with epilepsy hospitalized for seizures at an academic reference epilepsy center. Patients also completed the Neurological Disorders Depression Inventory for Epilepsy (NDDI-E), the Generalized Anxiety Disorder scale (GAD-7) for external validity, and underwent a standardized interview (Mini-DIPS-OA) as a gold standard to determine the presence of an ICD-10 anxiety disorder (generalized anxiety disorder (GAD), panic disorder, agoraphobia, and social phobia). Receiver operating characteristics (ROC) were calculated to determine the diagnostic accuracy of the brEASI, including the associated area under the curve (AUC) statistics to determine the potential of the brEASI to identify ICD-10 anxiety disorders diagnosed by interview. For comparative purposes, these analyses were also conducted for the GAD‐7. ResultsOf 80 recruited adult inpatients with epilepsy, 18 (23 %) were found to have a current anxiety disorder through standardized interview. In this study, both brEASI and GAD-7 showed a better diagnostic performance at a cutoff of >5 than at the previously reported cutoff values of >6 and >9, respectively. The AUC of the German brEASI was outstanding (AUC = 0.90, 95 % confidence interval (CI) = 0.82–0.96) for detecting all anxiety disorders and excellent for detecting non-GAD disorders (AUC = 0.85, CI = 0.76–0.92) at a cutoff of >5. At this optimal cutoff of >5 the brEASI demonstrated better sensitivity and specificity (89 % and 84 %) for identifying anxiety disorders than the GAD-7 (83 % and 74 %). The final German version of the brEASI is free to download at https://www.v-neuro.de/veroeffentlichungen/. ConclusionThe German version of the brEASI represents a valid and reliable epilepsy‐specific anxiety screening instrument. A positive screening result should be followed by further diagnostic procedures. Appropriate therapeutic steps should be initiated if the presence of an anxiety disorder or other psychiatric disorders is confirmed.

Clinical risk index of babies-II versus score for neonatal acute physiology-II in predicting mortality and morbidity in preterm babies

Background: The measurement of severity of illness using scoring systems is an important aspect in predicting mortality and morbidity in intensive care units which in turn can help in optimizing the limited healthcare resources in developing countries. The primary objective was to determine the correlation between clinical risk index of babies-II (CRIB-II) and score for neonatal acute physiology-II (SNAP-II) scores while the secondary objective was to identify which among them is superior in predicting mortality and morbidity in preterm neonates.Methods: The components of CRIB-II and SNAP-II scores were recorded prospectively over a period of 1 year in preterm very low birth weight (VLBW) babies and receiver-operating-characteristics (ROCs) were plotted for comparison. Correlation between CRIB-II and SNAP-II was examined by Pearson technique. The ability of CRIB-II and SNAP-II scores to correctly predict mortality, was assessed by calculating ROCs and their associated area under the curve (AUC).Results: Thirty nine neonates with a mean birth weight of 994.10 grams (SD±273.45 grams) and mean gestational age of 28.07 weeks (SD±2.29 weeks) were included in this study. The mean value of CRIB-II score and SNAP-II score was 8.54 (SD±4.67) and 9.82 (SD±8.93) respectively with a Pearson coefficient of 0.483 showing a modest correlation. CRIB-II (AUC 0.909) showed greater discrimination than SNAP-II (0.869) as a predictor of mortality. However, both the scores have poor discrimination when it comes to predicting neonatal morbidity.Conclusions: CRIB-II with its simplicity, need for uncomplicated variables and minimal time to generate a score for prediction of mortality and morbidity could be a useful tool in a busy neonatal intensive care unit (NICU).

Accuracy Assessment in Convolutional Neural Network-Based Deep Learning Remote Sensing Studies—Part 1: Literature Review

Convolutional neural network (CNN)-based deep learning (DL) is a powerful, recently developed image classification approach. With origins in the computer vision and image processing communities, the accuracy assessment methods developed for CNN-based DL use a wide range of metrics that may be unfamiliar to the remote sensing (RS) community. To explore the differences between traditional RS and DL RS methods, we surveyed a random selection of 100 papers from the RS DL literature. The results show that RS DL studies have largely abandoned traditional RS accuracy assessment terminology, though some of the accuracy measures typically used in DL papers, most notably precision and recall, have direct equivalents in traditional RS terminology. Some of the DL accuracy terms have multiple names, or are equivalent to another measure. In our sample, DL studies only rarely reported a complete confusion matrix, and when they did so, it was even more rare that the confusion matrix estimated population properties. On the other hand, some DL studies are increasingly paying attention to the role of class prevalence in designing accuracy assessment approaches. DL studies that evaluate the decision boundary threshold over a range of values tend to use the precision-recall (P-R) curve, the associated area under the curve (AUC) measures of average precision (AP) and mean average precision (mAP), rather than the traditional receiver operating characteristic (ROC) curve and its AUC. DL studies are also notable for testing the generalization of their models on entirely new datasets, including data from new areas, new acquisition times, or even new sensors.

From Talent Identification to Novo Basquete Brasil (NBB): Multifactorial Analysis of the Career Progression in Youth Brazilian Elite Basketball.

This study examined individual, task, and environmental constraints that influence the career progression of youth Brazilian elite basketball players and the probability of reaching Novo Basquete Brasil (NBB) and to determine if the association of the relative age effect (RAE) is a key factor in the career progression. The sample consisted of 4,692 male players who were registered to participate in at least one U15, U17, or U22 youth Brazilian basketball championship between 2004 and 2018. Athletes who reached a high-performance level were coded like NBB players (9.6%). The birthdates, height, body mass, playing position, geographic region, club, competition category, and team performance were retrieved from the official data archive of the Brazilian Basketball Confederation and the National Basketball League. The maturity status was estimated using the predicted age at peak height velocity. A binary logistic regression examined the influence of each characteristic on the probability of a youth Brazilian basketball player to reach the NBB. The receiver operating characteristic (ROC) curves and the associated area under the curve (AUC) were used to assess the discriminant ability of the model. The taller and younger players not selected early into national teams, without specialization by playing position, who participated in U22 national championship, migrated to the southeast region, and remained in the formation process over time have a greater chance to reach the NBB. The ROC curve demonstrated an AUC of 93%. A combination of individual, task, and environmental characteristics influences the sport career of a young Brazilian basketball player in reaching the NBB. Further, early-maturing athletes have a greater chance to reach higher performances. RAE influences lower-level categories, but not a “NBB player's” career progression. The coaches, stakeholders, and practitioners should perform a holistic evaluation of sport talent in terms of a constraint-based theoretical model with the aim of avoiding bias produced by the maturational status and RAE in the youth Brazilian elite basketball.

Using a national surgical database to predict complications following posterior lumbar surgery and comparing the area under the curve and F1-score for the assessment of prognostic capability

BACKGROUNDWith spinal surgery rates increasing in North America, models that are able to accurately predict which patients are at greater risk of developing complications are highly warranted. However, the previously published methods which have used large, multi-centre databases to develop their prediction models have relied on the receiver operator characteristics curve with the associated area under the curve (AUC) to assess their model's performance. Recently, it has been found that a precision-recall curve with the associated F1-score could provide a more realistic analysis for these models. PURPOSETo develop a logistic regression (LR) model for the prediction of complications following posterior lumbar spine surgery and to then assess for any difference in performance of the model when using the AUC versus the F1-score. STUDY DESIGNRetrospective review of a prospective cohort. PATIENT SAMPLEThe American College of Surgeons National Surgical Quality Improvement Program (NSQIP) registry was used. All patients that underwent posterior lumbar spine surgery between 2005 to 2016 with appropriate data were included. OUTCOME MEASURESBoth the AUC and F1-score were utilized to assess the prognostic performance of the prediction model. METHODSIn order to develop the LR model used to predict a complication during or following spine surgery, 19 variables were selected by three orthopedic spine surgeons from the NSQIP registry. Two datasets were developed for this analysis: (1) an imbalanced dataset, which was taken directly from the NSQIP registry, and (2) a down-sampled set. The purpose of the down-sampled set was to balance the data in order to evaluate whether balancing the data had an effect on model performance. The AUC and F1-score were applied to both of these datasets. RESULTSWithin the NSQIP database, 52,787 spine surgery cases were identified of which only 10% of these cases had complications during surgery. Applying the LR model showed a large difference between the AUC (0.69) and the F1 score (0.075) on the imbalanced dataset. However, no major differences existed between the AUC and F1-score when the data was balanced and the LR model was reapplied (0.69 and 0.62, AUC and F1-score, respectively). CONCLUSIONSThe F1-score detected a drastically lower performance for the prediction of complications when using the imbalanced data, but detected a performance similar to the AUC level when balancing techniques were utilized for the dataset. This difference is due to a low precision score when many false positive classifications are present, which is not identified when using the AUC value. This lowers the utility of the AUC score, as many of the datasets used in medicine are imbalanced. Therefore, we recommend using the F1-score on large, prospective databases when the data is imbalanced with a large amount of true negative classifications.

Comparison of Biomarker Modalities for Predicting Response to PD-1/PD-L1 Checkpoint Blockade

PD-L1 (programmed cell death ligand 1) immunohistochemistry (IHC), tumor mutational burden (TMB), gene expression profiling (GEP), and multiplex immunohistochemistry/immunofluorescence (mIHC/IF) assays have been used to assess pretreatment tumor tissue to predict response to anti-PD-1/PD-L1 therapies. However, the relative diagnostic performance of these modalities has yet to be established. To compare studies that assessed the diagnostic accuracy of PD-L1 IHC, TMB, GEP, and mIHC/IF in predicting response to anti-PD-1/PD-L1 therapy. A search of PubMed (from inception to June 2018) and 2013 to 2018 annual meeting abstracts from the American Association for Cancer Research, American Society of Clinical Oncology, European Society for Medical Oncology, and Society for Immunotherapy of Cancer was conducted to identify studies that examined the use of PD-L1 IHC, TMB, GEP, and mIHC/IF assays to determine objective response to anti-PD-1/PD-L1 therapy. For PD-L1 IHC, only clinical trials that resulted in US Food and Drug Administration approval of indications for anti-PD-1/PD-L1 were included. Studies combining more than 1 modality were also included. Preferred Reporting Items for Systematic Reviews and Meta-analysis guidelines were followed. Two reviewers independently extracted the clinical outcomes and test results for each individual study. Summary receiver operating characteristic (sROC) curves; their associated area under the curve (AUC); and pooled sensitivity, specificity, positive and negative predictive values (PPV, NPV), and positive and negative likelihood ratios (LR+ and LR-) for each assay modality. Tumor specimens representing over 10 different solid tumor types in 8135 patients were assayed, and the results were correlated with anti-PD-1/PD-L1 response. When each modality was evaluated with sROC curves, mIHC/IF had a significantly higher AUC (0.79) compared with PD-L1 IHC (AUC, 0.65, P < .001), GEP (AUC, 0.65, P = .003), and TMB (AUC, 0.69, P = .049). When multiple different modalities were combined such as PD-L1 IHC and/or GEP + TMB, the AUC drew nearer to that of mIHC/IF (0.74). All modalities demonstrated comparable NPV and LR-, whereas mIHC/IF demonstrated higher PPV (0.63) and LR+ (2.86) than the other approaches. In this meta-analysis, tumor mutational burden, PD-L1 IHC, and GEP demonstrated comparable AUCs in predicting response to anti-PD-1/PD-L1 treatment. Multiplex immunohistochemistry/IF and multimodality biomarker strategies appear to be associated with improved performance over PD-L1 IHC, TMB, or GEP alone. Further studies with mIHC/IF and composite approaches with a larger number of patients will be required to confirm these findings. Additional study is also required to determine the most predictive analyte combinations and to determine whether biomarker modality performance varies by tumor type.

Estimation of lung cancer risk using homology-based emphysema quantification in patients with lung nodules.

The purpose of this study was to assess whether homology-based emphysema quantification (HEQ) is significantly associated with lung cancer risk. This retrospective study was approved by our institutional review board. We included 576 patients with lung nodules (317 men and 259 women; age, 66.8 ± 12.3 years), who were selected from a database previously generated for computer-aided diagnosis. Of these, 283 were diagnosed with lung cancer, whereas the remaining 293 showed benign lung nodules. HEQ was performed and percentage of low-attenuation lung area (LAA%) was calculated on the basis of computed tomography scans. Statistical models were constructed to estimate lung cancer risk using logistic regression; sex, age, smoking history (Brinkman index), LAA%, and HEQ were considered independent variables. The following three models were evaluated: the base model (sex, age, and smoking history); the LAA% model (the base model + LAA%); and the HEQ model (the base model + HEQ). Model performance was assessed using receiver operating characteristic analysis and the associated area under the curve (AUC). Differences in AUCs among the models were evaluated using Delong’s test. AUCs of the base, LAA%, and HEQ models were 0.585, 0.593, and 0.622, respectively. HEQ coefficient was statistically significant in the HEQ model (P = 0.00487), but LAA% coefficient was not significant in the LAA% model (P = 0.199). Delong’s test revealed significant difference in AUCs between the LAA% and HEQ models (P = 0.0455). In conclusion, after adjusting for age, sex, and smoking history (Brinkman index), HEQ was significantly associated with lung cancer risk.

Predicting mortality in extremely low birth weight infants: Comparison between gestational age, birth weight, Apgar score, CRIB II score, initial and lowest serum albumin levels.

We explored GA, BW, Apgar score, CRIB II score, and serum albumin levels as univariate predictors of mortality in extremely low birth weight infants. Medical records of 564 extremely low birth weight infants were reviewed retrospectively. The infants were grouped as survivors (group I), expired ≤ 7th postnatal day (group II), and expired > 7th postnatal day (group III). The predictive value for mortality of gestational age, birth weight, Apgar scores at 1 and 5 min, clinical risk index for babies II score, and first and lowest serum albumin levels was assessed by calculating the associated area under the curve (AUC) in receiver operating characteristic (ROC) curves. The overall survival and mortality rates of groups I, II, and III were 81.0% (457/564), 7.6% (43/564), and 11.4% (64/564), respectively. Birth weight, Apgar scores at 1 and 5 min, and first serum albumin levels were significantly higher, while the clinical risk index for babies II score was significantly lower in group I when compared to groups II and III. Gestational age and lowest serum albumin level in group I were significantly higher than group III, but not group II. However, gestational age, birth weight, and clinical risk index for babies II score showed gestational age dependent variations regardless of survival or mortality. Apgar score at 5 min (0.756) and lowest serum albumin level (0.771) demonstrated the highest AUC of the ROC curve in predicting mortality in group II and III, respectively. In conclusion, Apgar score at 5 min and lowest serum albumin level were the most effective predictors for mortality in extremely low birth weight infants during ≤ 7th and > 7th postnatal days, respectively.

A novel thermoregulatory role for PDE10A in mouse and human adipocytes.

Phosphodiesterase type 10A (PDE10A) is highly enriched in striatum and is under evaluation as a drug target for several psychiatric/neurodegenerative diseases. Preclinical studies implicate PDE10A in the regulation of energy homeostasis, but the mechanisms remain unclear. By utilizing small‐animal PET/MRI and the novel radioligand [18F]‐AQ28A, we found marked levels of PDE10A in interscapular brown adipose tissue (BAT) of mice. Pharmacological inactivation of PDE10A with the highly selective inhibitor MP‐10 recruited BAT and potentiated thermogenesis in vivo. In diet‐induced obese mice, chronic administration of MP‐10 caused weight loss associated with increased energy expenditure, browning of white adipose tissue, and improved insulin sensitivity. Analysis of human PET data further revealed marked levels of PDE10A in the supraclavicular region where brown/beige adipocytes are clustered in adults. Finally, the inhibition of PDE10A with MP‐10 stimulated thermogenic gene expression in human brown adipocytes and induced browning of human white adipocytes. Collectively, our findings highlight a novel thermoregulatory role for PDE10A in mouse and human adipocytes and promote PDE10A inhibitors as promising candidates for the treatment of obesity and diabetes.

CRIB II score versus gestational age and birth weight in preterm infant mortality prediction: who will win the bet?

Introduction. In neonatology, various illness severity scores have been developed to predict mortality and morbidity risk in neonates. The aim of our study was to validate the ability of the ‘Clinical Risk Index for Babies’ (CRIB) II score to predict mortality in neonates born before 32 weeks’ gestation in a level 3 neonatal intensive care unit (NICU), setting. Materials and Methods. Prospective birth cohort study including all live-born neonates of 32 weeks’ gestation or less. . CRIB II score was calculated and the predicted mortality was compared with the observed mortality. Discrimination (the ability of the score to correctly predict survival or death) was assessed by calculating the receiver operating characteristic curve (ROC curve) and its associated area under the curve (AUC). Results. The ROC curve analysis in our study showed that the AUC was 0.9008 suggesting that mortality prediction was 90% accurate for all infants. Sensitivity and specificity were 77% and 88% respectively. In our study population, the CRIB II score appears to be more accurate than gestational age and birth weight in predicting mortality. Conclusions. The CRIB II scoring system is a useful tool for predicting mortality and morbidity in NICUs, and also a useful tool for evaluating the variations in mortality and other outcomes seen between different NICUs.

A Novel Data-Driven Approach to Preoperative Mapping of Functional Cortex Using Resting-State Functional Magnetic Resonance Imaging

Recent findings associated with resting-state cortical networks have provided insight into the brain's organizational structure. In addition to their neuroscientific implications, the networks identified by resting-state functional magnetic resonance imaging (rs-fMRI) may prove useful for clinical brain mapping. To demonstrate that a data-driven approach to analyze resting-state networks (RSNs) is useful in identifying regions classically understood to be eloquent cortex as well as other functional networks. This study included 6 patients undergoing surgical treatment for intractable epilepsy and 7 patients undergoing tumor resection. rs-fMRI data were obtained before surgery and 7 canonical RSNs were identified by an artificial neural network algorithm. Of these 7, the motor and language networks were then compared with electrocortical stimulation (ECS) as the gold standard in the epilepsy patients. The sensitivity and specificity for identifying these eloquent sites were calculated at varying thresholds, which yielded receiver-operating characteristic (ROC) curves and their associated area under the curve (AUC). RSNs were plotted in the tumor patients to observe RSN distortions in altered anatomy. The algorithm robustly identified all networks in all patients, including those with distorted anatomy. When all ECS-positive sites were considered for motor and language, rs-fMRI had AUCs of 0.80 and 0.64, respectively. When the ECS-positive sites were analyzed pairwise, rs-fMRI had AUCs of 0.89 and 0.76 for motor and language, respectively. A data-driven approach to rs-fMRI may be a new and efficient method for preoperative localization of numerous functional brain regions.

Total Plasma Protein in Very Preterm Babies: Prognostic Value and Comparison with Illness Severity Scores

ObjectiveWe aimed to investigate the predictive value for severe adverse outcome of plasma protein measurements on day one of life in very preterm infants and to compare total plasma protein levels with the validated illness severity scores CRIB, CRIB-II, SNAP-II and SNAPPE-II, regarding their predictive ability for severe adverse outcome.MethodsWe analyzed a cohort of infants born at 24–31 weeks gestation, admitted to the tertiary intensive care unit of a university hospital over 10.5 years. The outcome measure was “severe adverse outcome” defined as death before discharge or severe neurological injury on cranial ultrasound. The adjusted odd ratio (aOR) and 95% confidence interval (95% CI) of severe adverse outcome for hypoproteinemia (total plasma protein level <40 g/L) was calculated by univariate and multivariate analyses. Calibration (Hosmer-Lemeshow goodness-of-fit) was performed and the predictive ability for severe adverse outcome was assessed for total plasma protein and compared with CRIB, CRIB-II, SNAP-II and SNAPPE-II, by calculating receiver operating characteristic (ROC) curves and their associated area under the curve (AUC).Results761 infants were studied: 14.4% died and 4.1% survived with severe cerebral ultrasound findings. The aOR of severe adverse outcome for hypoproteinemia was 6.1 (95% CI 3.8–9.9). The rank order for variables, as assessed by AUCs and 95% CIs, in predicting outcome was: total plasma protein [0.849 (0.821–0.873)], SNAPPE-II [0.822 (0.792–0.848)], CRIB [0.821 (0.792–0.848)], SNAP-II [0.810 (0.780–0.837)] and CRIB-II [0.803 (0.772–0.830)]. Total plasma protein predicted severe adverse outcome significantly better than CRIB-II and SNAP-II (both p<0.05). Calibration for total plasma protein was very good.ConclusionsEarly hypoproteinemia has prognostic value for severe adverse outcome in very preterm, sick infants. Total plasma protein has a predictive performance comparable with CRIB and SNAPPE-II and greater than other validated severity scores.

Limited Predictive Accuracy of the Booklet Category Test in a Criminal Forensic Sample

A known-groups design was used to evaluate the clinical decision rules of the validity indicators of the Booklet Category Test for the detection of malingered neurocognitive dysfunction (MND) in an incarcerated male sample. Participants were 128 inmates who were classified into MND and presumed valid groups using the Slick, Sherman, and Iverson (1999) criteria. While errors on Subtests I and II had the best classification accuracy at 71.1% with 95.1% specificity and 49.3% sensitivity, the associated area under the curve (AUC) indicated only “acceptable” discrimination due to poor sensitivity. Logistic regression confirmed that errors on Subtests I and II were the only statistically significant validity indicator. The remaining validity indicators performed poorly with AUCs at < .70, indicating no discriminative power. The effectiveness of the clinical decision rules of the validity indicators in predicting the presence of MND evidenced limited support. A positive finding on Subtest I and II errors was clinically significant in this sample, but negative results revealed no utility. Validity indices composed from different items may yield more effective results among male prisoners referred for neuropsychological evaluation.

Exploring Polypharmacology Using a ROCS-Based Target Fishing Approach

Polypharmacology has emerged as a new theme in drug discovery. In this paper, we studied polypharmacology using a ligand-based target fishing (LBTF) protocol. To implement the protocol, we first generated a chemogenomic database that links individual protein targets with a specified set of drugs or target representatives. Target profiles were then generated for a given query molecule by computing maximal shape/chemistry overlap between the query molecule and the drug sets assigned to each protein target. The overlap was computed using the program ROCS (Rapid Overlay of Chemical Structures). We validated this approach using the Directory of Useful Decoys (DUD). DUD contains 2950 active compounds, each with 36 property-matched decoys, against 40 protein targets. We chose a set of known drugs to represent each DUD target, and we carried out ligand-based virtual screens using data sets of DUD actives seeded into DUD decoys for each target. We computed Receiver Operator Characteristic (ROC) curves and associated area under the curve (AUC) values. For the majority of targets studied, the AUC values were significantly better than for the case of a random selection of compounds. In a second test, the method successfully identified off-targets for drugs such as rimantadine, propranolol, and domperidone that were consistent with those identified by recent experiments. The results from our ROCS-based target fishing approach are promising and have potential application in drug repurposing for single and multiple targets, identifying targets for orphan compounds, and adverse effect prediction.

Automated vector selection of SIVQ and parallel computing integration MATLABTM: Innovations supporting large-scale and high-throughput image analysis studies

Introduction: Spatially invariant vector quantization (SIVQ) is a texture and color-based image matching algorithm that queries the image space through the use of ring vectors. In prior studies, the selection of one or more optimal vectors for a particular feature of interest required a manual process, with the user initially stochastically selecting candidate vectors and subsequently testing them upon other regions of the image to verify the vector’s sensitivity and specificity properties (typically by reviewing a resultant heat map). In carrying out the prior efforts, the SIVQ algorithm was noted to exhibit highly scalable computational properties, where each region of analysis can take place independently of others, making a compelling case for the exploration of its deployment on high-throughput computing platforms, with the hypothesis that such an exercise will result in performance gains that scale linearly with increasing processor count. Methods: An automated process was developed for the selection of optimal ring vectors to serve as the predicate matching operator in defining histopathological features of interest. Briefly, candidate vectors were generated from every possible coordinate origin within a user-defined vector selection area (VSA) and subsequently compared against user-identified positive and negative “ground truth” regions on the same image. Each vector from the VSA was assessed for its goodness-of-fit to both the positive and negative areas via the use of the receiver operating characteristic (ROC) transfer function, with each assessment resulting in an associated area-under-the-curve (AUC) figure of merit. Results: Use of the above-mentioned automated vector selection process was demonstrated in two cases of use: First, to identify malignant colonic epithelium, and second, to identify soft tissue sarcoma. For both examples, a very satisfactory optimized vector was identified, as defined by the AUC metric. Finally, as an additional effort directed towards attaining high-throughput capability for the SIVQ algorithm, we demonstrated the successful incorporation of it with the MATrix LABoratory (MATLABTM) application interface. Conclusion: The SIVQ algorithm is suitable for automated vector selection settings and high throughput computation.

Bayesian estimation of the receiver operating characteristic curve for a diagnostic test with a limit of detection in the absence of a gold standard

The receiver operating characteristic (ROC) curve is commonly used for evaluating the discriminatory ability of a biomarker. Measurements for a diagnostic test may be subject to an analytic limit of detection leading to immeasurable or unreportable test results. Ignoring the scores that are beyond the limit of detection of a test leads to a biased assessment of its discriminatory ability, as reflected by indices such as the associated area under the curve (AUC). We propose a Bayesian approach for the estimation of the ROC curve and its AUC for a test with a limit of detection in the absence of gold standard based on assumptions of normally and gamma-distributed data. The methods are evaluated in simulation studies, and a truncated gamma model with a point mass is used to evaluate quantitative real-time polymerase chain reaction data for bovine Johne's disease (paratuberculosis). Simulations indicated that estimates of diagnostic accuracy and AUC were good even for relatively small sample sizes (n=200). Exceptions were when there was a high per cent of unquantifiable results (60 per cent) or when AUC was < or =0.6, which indicated a marked overlap between the outcomes in infected and non-infected populations.