Proactive and reactive accumulation-to-bound processes compete during perceptual decisions

Standard models of perceptual decision-making postulate that a response is triggered in reaction to stimulus presentation when the accumulated stimulus evidence reaches a decision threshold. This framework excludes however the possibility that informed responses are generated proactively at a time independent of stimulus. Here, we find that, in a free reaction time auditory task in rats, reactive and proactive responses coexist, suggesting that choice selection and motor initiation, commonly viewed as serial processes, are decoupled in general. We capture this behavior by a novel model in which proactive and reactive responses are triggered whenever either of two competing processes, respectively Action Initiation or Evidence Accumulation, reaches a bound. In both types of response, the choice is ultimately informed by the Evidence Accumulation process. The Action Initiation process readily explains premature responses, contributes to urgency effects at long reaction times and mediates the slowing of the responses as animals get satiated and tired during sessions. Moreover, it successfully predicts reaction time distributions when the stimulus was either delayed, advanced or omitted. Overall, these results fundamentally extend standard models of evidence accumulation in decision making by showing that proactive and reactive processes compete for the generation of responses.

Opal: an implementation science tool for machine learning clinical decision support in anesthesia

Opal is the first published example of a full-stack platform infrastructure for an implementation science designed for ML in anesthesia that solves the problem of leveraging ML for clinical decision support. Users interact with a secure online Opal web application to select a desired operating room (OR) case cohort for data extraction, visualize datasets with built-in graphing techniques, and run in-client ML or extract data for external use. Opal was used to obtain data from 29,004 unique OR cases from a single academic institution for pre-operative prediction of post-operative acute kidney injury (AKI) based on creatinine KDIGO criteria using predictors which included pre-operative demographic, past medical history, medications, and flowsheet information. To demonstrate utility with unsupervised learning, Opal was also used to extract intra-operative flowsheet data from 2995 unique OR cases and patients were clustered using PCA analysis and k-means clustering. A gradient boosting machine model was developed using an 80/20 train to test ratio and yielded an area under the receiver operating curve (ROC-AUC) of 0.85 with 95% CI [0.80–0.90]. At the default probability decision threshold of 0.5, the model sensitivity was 0.9 and the specificity was 0.8. K-means clustering was performed to partition the cases into two clusters and for hypothesis generation of potential groups of outcomes related to intraoperative vitals. Opal’s design has created streamlined ML functionality for researchers and clinicians in the perioperative setting and opens the door for many future clinical applications, including data mining, clinical simulation, high-frequency prediction, and quality improvement.

Computational modeling of the N-Back task in the ABCD study: associations of drift diffusion model parameters to polygenic scores of mental disorders and cardiometabolic diseases

Background. Cognitive dysfunction is common in mental disorders and represents a potential risk factor in childhood. The nature and extent of associations between childhood cognitive function and polygenic risk for mental disorders is unclear. We applied computational modeling to gain insight into mechanistic processes underlying decision making and working memory in childhood and their associations with PRS for mental disorders and comorbid cardiometabolic diseases. Methods. We used the drift diffusion model to infer latent computational processes underlying decision-making and working memory during the N-back task in 3707 children aged 9-10 from the ABCD Study. SNP-based heritability was estimated for cognitive phenotypes, including computational parameters, aggregated N-back task performance and neurocognitive assessments. PRS was calculated for Alzheimer’s disease (AD), bipolar disorder, coronary artery disease (CAD), major depressive disorder, obsessive-compulsive disorder, schizophrenia and type 2 diabetes. Results. Heritability estimates of cognitive phenotypes ranged from 12 to 39%. Bayesian mixed models revealed that slower accumulation of evidence was associated with higher PRS for CAD and schizophrenia. Longer non-decision time was associated with higher PRS for AD and lower PRS for CAD. Narrower decision threshold was associated with higher PRS for CAD. Load-dependent effects on non-decision time and decision threshold were associated with PRS for AD and CAD, respectively. Aggregated neurocognitive test scores were not associated with PRS for any of the mental or cardiometabolic phenotypes.Conclusions. We identified distinct associations between computational cognitive processes to genetic risk for mental illness and cardiometabolic disease, which could represent childhood cognitive risk factors.

sFlt-1 and CA 15.3 are indicators of endothelial damage and pulmonary fibrosis in SARS-CoV-2 infection

COVID-19 pandemic led to a worldwide increase of hospitalizations for interstitial pneumonia with thrombosis complications, endothelial injury and multiorgan disease. Common CT findings include lung bilateral infiltrates, bilateral ground-glass opacities and/or consolidation whilst no current laboratory parameter consents rapidly evaluation of COVID-19 risk and disease severity. In the present work we investigated the association of sFLT-1 and CA 15.3 with endothelial damage and pulmonary fibrosis. Serum sFlt-1 has been associated with endothelial injury and sepsis severity, CA 15.3 seems an alternative marker for KL-6 for fibrotic lung diseases and pulmonary interstitial damage. We analysed 262 SARS-CoV-2 patients with differing levels of clinical severity; we found an association of serum sFlt-1 (ROC AUC 0.902, decision threshold > 90.3 pg/mL, p < 0.001 Sens. 83.9% and Spec. 86.7%) with presence, extent and severity of the disease. Moreover, CA 15.3 appeared significantly increased in COVID-19 severe lung fibrosis (ICU vs NON-ICU patients 42.6 ± 3.3 vs 25.7 ± 1.5 U/mL, p < 0.0001) and was associated with lung damage severity grade (ROC AUC 0.958, decision threshold > 24.8 U/mL, p < 0.0001, Sens. 88.4% and Spec. 91.8%). In conclusion, serum levels of sFlt-1 and CA 15.3 appeared useful tools for categorizing COVID-19 clinical stage and may represent a valid aid for clinicians to better personalise treatment.

Decomposing response times in Williams syndrome in separate cognitive processes

Williams Syndrome (WS) is a rare neurodevelopmental disorder of genetic origin. The syndrome is characterised by a selective set of deficits in a number of cognitive domains. In spite of a wealth of studies, response times (RTs) of WS have attracted little attention. In the present study, we fill this gap by analysing data from a receptive vocabulary task using the Diffusion Decision Model (DDM). Our results show that the speed of accumulation, decision threshold and non-decision time parameters of WS individuals are similar to these of typically developing 5-year-old preschoolers. In addition, WS verbal intelligence scores were associated with the speed of accumulation of lexical information. Finally, the performance of WS and preschooler individuals was correlated across the vocabulary task and an additional orientation discrimination task only at the group but not at the individual level; therefore, pointing to domain-specific lexical and perceptual processing in WS.

A computational experimental of noise suppressing technique stand on hard decision threshold dissimilarity

Due to the extreme insistence for digital image processing, plentiful modern noise suppressing techniques are embodied of dissimilarity process and suppressing process. One of the extreme capability dissimilarity is hard decision threshold (HDT) dissimilarity, which has been recently declared in 2012, for suppressing the impulsive noisy photographs thus the computer experimental statement attempts to investigate the capability of the noise suppressing technique that is stand on HDT dissimilarity for the processed photographs, which are corrupted by fixed-intensity impulse noise (FIIN). This paper proposes the noise suppressing technique stand on HDT dissimilarity for FIIN. There are 3 primary contributions of this paper. The first contribution is the statistical average of the HDT dissimilarity of noise-free elements, which are computed from plentiful ground-truth photographs by varying window size for the best HDT window size. The second contribution is the statistical average of the HDT dissimilarity of corrupted elements, which are computed from plentiful corrupted photographs by varying outlier density for the best HDT window size. The final contribution is the statistical interrelation of the capability of the noise suppressing technique and hard consistent of HDT dissimilarity are investigated by varying the outlier denseness for the best HDT hard consistence.

Developing Strategies for Onchocerciasis Elimination Mapping and Surveillance Through The Diagnostic Network Optimization Approach

BackgroundOnchocerciasis (river blindness) is a filarial disease targeted for elimination of transmission. However, challenges exist to the implementation of effective diagnostic and surveillance strategies at various stages of elimination programs. To address these challenges, we used a network data analytics approach to identify optimal diagnostic scenarios for onchocerciasis elimination mapping (OEM).MethodsThe diagnostic network optimization (DNO) method was used to model the implementation of the old Ov16 rapid diagnostic test (RDT) and of new RDTs in development for OEM under different testing strategy scenarios with varying testing locations, test performance and disease prevalence. Environmental suitability scores (ESS) based on machine learning algorithms were developed to identify areas at risk of transmission and used to select sites for OEM in Bandundu region in the Democratic Republic of Congo (DRC) and Uige province in Angola. Test sensitivity and specificity ranges were obtained from the literature for the existing RDT, and from characteristics defined in the target product profile for the new RDTs. Sourcing and transportation policies were defined, and costing information was obtained from onchocerciasis programs. Various scenarios were created to test various state configurations. The actual demand scenarios represented the disease prevalence at IUs according to the ESS, while the counterfactual scenarios (conducted only in the DRC) are based on adapted prevalence estimates to generate prevalence close to the statistical decision thresholds (5% and 2%), to account for variability in field observations. The number of correctly classified implementation units (IUs) per scenario were estimated and key cost drivers were identified.ResultsIn both Bandundu and Uige, the sites selected based on ESS had high predicted onchocerciasis prevalence &gt;10%. Thus, in the actual demand scenarios in both Bandundu and Uige, the old Ov16 RDT correctly classified all 13 and 11 IUs, respectively, as requiring CDTi. In the counterfactual scenarios in Bandundu, the new RDTs with higher specificity correctly classified IUs more cost effectively. The new RDT with highest specificity (99.8%) correctly classified all 13 IUs. However, very high specificity (e.g., 99.8%) when coupled with imperfect sensitivity, can result in many false negative results (missing decisions to start MDA) at the 5% statistical decision threshold (the decision rule to start MDA). This effect can be negated by reducing the statistical decision threshold to 2%. Across all scenarios, the need for second stage sampling significantly drove program costs upwards. The best performing testing strategies with new RDTs were more expensive than testing with existing tests due to need for second stage sampling, but this was offset by the cost of incorrect classification of IUs.ConclusionThe new RDTs modelled added most value in areas with variable disease prevalence, with most benefit in IUs that are near the statistical decision thresholds. Based on the evaluations in this study, DNO could be used to guide the development of new RDTs based on defined sensitivities and specificities. While test sensitivity is a minor driver of whether an IU is identified as positive, higher specificities are essential. Further, these models could be used to explore the development and optimization of new tools for other neglected tropical diseases.

Automatic Recognition of Colon and Esophagogastric Cancer with Machine Learning and Hyperspectral Imaging

There are approximately 1.8 million diagnoses of colorectal cancer, 1 million diagnoses of stomach cancer, and 0.6 million diagnoses of esophageal cancer each year globally. An automatic computer-assisted diagnostic (CAD) tool to rapidly detect colorectal and esophagogastric cancer tissue in optical images would be hugely valuable to a surgeon during an intervention. Based on a colon dataset with 12 patients and an esophagogastric dataset of 10 patients, several state-of-the-art machine learning methods have been trained to detect cancer tissue using hyperspectral imaging (HSI), including Support Vector Machines (SVM) with radial basis function kernels, Multi-Layer Perceptrons (MLP) and 3D Convolutional Neural Networks (3DCNN). A leave-one-patient-out cross-validation (LOPOCV) with and without combining these sets was performed. The ROC-AUC score of the 3DCNN was slightly higher than the MLP and SVM with a difference of 0.04 AUC. The best performance was achieved with the 3DCNN for colon cancer and esophagogastric cancer detection with a high ROC-AUC of 0.93. The 3DCNN also achieved the best DICE scores of 0.49 and 0.41 on the colon and esophagogastric datasets, respectively. These scores were significantly improved using a patient-specific decision threshold to 0.58 and 0.51, respectively. This indicates that, in practical use, an HSI-based CAD system using an interactive decision threshold is likely to be valuable. Experiments were also performed to measure the benefits of combining the colorectal and esophagogastric datasets (22 patients), and this yielded significantly better results with the MLP and SVM models.

Justification and Application of Sufficiency Criterion for Measuring the Activity of Uranium Radioisotopes in Biological Samples

Purpose: Ensuring the fulfillment of the sufficiency n criterion when measuring the activity of uranium radionuclides in biological samples carried out within the individual monitoring programme by calculation of the uncertainty and characteristics limits for measurements. Material and methods: The sufficiency criterion definition is given, which determines the maximum value of the decision threshold for measurements carried out for the individual monitoring of workers at which the fact of non-exceeding of the annual dose limit or permissible level takes place, taking into account the uncertainty of the dose assessment. A model approach is used to calculate the sufficiency criterion and characteristics limits when measuring the radioactive material excrected by individual workers. The model approach consisted in the development of a calculation model based on the functional dependence of measured input values on the process of radiochemical preparation and subsequent spectrometric measurement of the sample. Results: A model has been developed for calculating the activity of uranium radionuclides 234U, 235U and 238U in a biological sample based on the description of the procedure for spectrometric measurement, which consisted in the deposition by the electrolytic method on the target after chromatographic extraction of uranium from the urine sample. The reference radioactive solution of the 232U radionuclide added to the sample as a reference point for determining the efficiency of uranium radionuclide separation (chemical yield). Equations are obtained for calculating the values of the decision threshold and the detection limit for the total activity of the above alpha-emitting uranium radionuclides. Using these equations, the dependence of the decision threshold and the detection limit on measurement time is determined for the given input data. This allows planning the measurement time at which the activity of uranium radionuclides in the sample can be determined reliably or at which the sufficiency criterion of the measurement method will be provided (necessary in the case when the activity is not detected, that is, the measurement result is less than the decision threshold). The values of the activity of uranium radionuclides 234U, 235U and 238U and the corresponding characteristics limits for the measurement were calculated on the basis of a real example of spectrometric measurement of the activity of uranium radionuclides in a sample. Conclusion: Ensuring the fulfillment of the sufficiency criterion when measuring the activity of uranium radionuclides in biological samples is achieved by the correct determination of the sample measurement time. This is determined by time dependence analysis of the characteristics limits (the decision threshold and the detection limit ) for the measurement of the total activity of the above alpha-emitting uranium radionuclides 234U, 235U and 238U.