Confidence Error Research Articles

Permutation confidence region for multiple regression and fidelity to asymptotic approximation

The permutation inference distribution (PID) introduced in Wu and Vos (2019), The Journal of Nonparametric Statistics 31(3), 722-742) is extended to the joint PID for multiple regression. The joint PID is used for both hypothesis testing and the construction of confidence regions and its computational burden is similar to that of conducting a single hypothesis test. Asymptotic normality results show that PID confidence regions are asymptotically ellipsoidal and exact. In finite samples, the PID confidence regions can be used to check the fidelity of normal approximations. In some cases, normal theory-based confidence regions may be adjusted to approach nominal confidence errors. Simulation studies and real data applications are used to evaluate inferences obtained from the PID.

Probing for the host galaxies of the fast X-ray transients XRT 000519 and XRT 110103

ABSTRACT Over the past few years, ∼30 extragalactic fast X-ray transients (FXRTs) have been discovered, mainly in Chandra and XMM-Newton data. Their nature remains unclear, with proposed origins, including a double neutron star merger, a tidal disruption event involving an intermediate-mass black hole and a white dwarf, or a supernova shock breakout. A decisive differentiation between these three promising mechanisms for their origin requires an understanding of the FXRT energetics, environments, and/or host properties. We present optical observations obtained with the Very Large Telescope for the FXRTs XRT 000519 and XRT 110103 and Gran Telescopio Canarias observations for XRT 000519 designed to search for host galaxies of these FXRTs. In the gs, rs, and R-band images, we detect an extended source on the north-west side of the $\sim \, 1^{\prime \prime }$ (68 per cent confidence) error circle of the X-ray position of XRT 000519 with a Kron magnitude of gs = 26.29 ± 0.09 (AB magnitude). We discuss the XRT 000519 association with the probable host candidate for various possible distances, and we conclude that if XRT 000519 is associated with the host candidate a supernova shock breakout scenario is likely excluded. No host galaxy is found near XRT 110103 down to a limiting magnitude of R > 25.8.

Inter Rater Reliability Test of Digitalized Telescoping Goniometer to Measure Active Range of Motion of Selected Large Joints Among Healthy Individuals

Aim: This study aimed to create an alternative tool for joint measurements with higher level of confidence and minimal error. Methods: This was a quantitative exploratory research study that determines the inter-rater reliability of the Digitalized telescoping Goniometer (DTG). 25 healthy individuals were selected using Slovin’s Formula. SPSS version 24 was used in analyzing the reliability and validity. The inter- rater reliability is described by Intra class coefficient (ICC) and Limits of Agreement (LOA). The references for the interpretation of the ICC in the present study were poor reliability < 0.5, moderate reliability 0.5-0.75 good reliability 0.75-0.90, and excellent reliability for > 0.90. The Bland-Altman plots was used to investigate systematic differences and assess 95% limits of agreement (LOA) visual indication of agreement. In LOA is calculated as LOA = mean difference + 1.96 (SD). Results: The ICC result of the DGT had moderate to good reliability for shoulder, elbow and hip joint (0.8835, 0.7395, 0.66525) respectively, while poor reliability on the knee joint (0.48175). The SEM taken using DGT ranged from 1.40- 7.27. The result showed moderate reliability on shoulder (0.7055), elbow (0.5215) and knee joint (0.576) while the hip joint (0.4825) showed poor reliability. As for the LOA, DTG has a wider limit than universal goniometer for all ROM. Conclusion and Implication: DGT showed as a reliable tool in measuring the ROM. DTG is a device can be use in goniometry.

On the computation of confidence regions and error ellipses: a critical appraisal

Customary confidence regions do not truly reflect in the majority of our geodetic applications the confidence one can have in one’s produced estimators. As it is common practice in our daily data analyses to combine methods of parameter estimation and hypothesis testing before the final estimator is produced, it is their combined uncertainty that has to be taken into account when constructing confidence regions. Ignoring the impact of testing on estimation will produce faulty confidence regions and therefore provide an incorrect description of estimator’s quality. In this contribution, we address the interplay between estimation and testing and show how their combined non-normal distribution can be used to construct truthful confidence regions. In doing so, our focus is on the designing phase prior to when the actual measurements are collected, where it is assumed that the working (null) hypothesis is true. We discuss two different approaches for constructing confidence regions: Approach I in which the region’s shape is user-fixed and only its size is determined by the distribution, and Approach II in which both the size and shape are simultaneously determined by the estimator’s non-normal distribution. We also prove and demonstrate that the estimation-only confidence regions have a poor coverage in the sense that they provide an optimistic picture. Next to the provided theory, we provide computational procedures, for both Approach I and Approach II, on how to compute confidence regions and confidence levels that truthfully reflect the combined uncertainty of estimation and testing.

Exploring the Key Factors of Shared Decision-Making Through an Influential Network Relation Map: The Orthopedic Nurse's Perspective.

BackgroundFew studies have used quantitative methods to explore the key factors affecting shared decision-making (SDM) in nursing decision-making from the perspective of orthopedic nurses.PurposeTo understand the intercorrelations among shared decision-making questionnaire–nurse (SDM-Q-NUR) factors and identify key factors for clinical nursing care decisions in orthopedics.MethodsIn May 2021, this study investigated the interdependence of the SDM-Q-NUR scale and developed an influential network-relation map (INRM) from the clinical experience of 13 trained orthopedic nurses using the Decision-making Trial and Evaluation Laboratory method.ResultsThe INRM results showed that the nine criteria corresponded to three stages: preparation, discussion, and decision. “I helped my patient or patient's family understand all the information” (C5) and “I wanted to know from my patient or patient's family how they want to be involved in making the nursing care decision” (C2) are the main key factors for the beginning of nursing decision. In the discussion and decision stages, the corresponding key factors are “I made it clear to my patient or patient's family that a nursing care decision needs to be made” (C1) and “I asked my patient or patient's family which nursing care option they prefer” (C6). The result's statistical significance confidence and gap error were 98.106% and 1.894%, respectively.ConclusionsWhen making nursing decisions with patients, orthopedic nurses need to have detailed information about how patients are involved in SDM and all relevant information. Nurses should also inform patients and their families regarding the purpose of the discussion, namely, to help one understand the content, advantages, and disadvantages of the nursing care options, and finally, make a decision.

SegVeg: Segmenting RGB Images into Green and Senescent Vegetation by Combining Deep and Shallow Methods.

Pixel segmentation of high-resolution RGB images into chlorophyll-active or nonactive vegetation classes is a first step often required before estimating key traits of interest. We have developed the SegVeg approach for semantic segmentation of RGB images into three classes (background, green, and senescent vegetation). This is achieved in two steps: A U-net model is first trained on a very large dataset to separate whole vegetation from background. The green and senescent vegetation pixels are then separated using SVM, a shallow machine learning technique, trained over a selection of pixels extracted from images. The performances of the SegVeg approach is then compared to a 3-class U-net model trained using weak supervision over RGB images segmented with SegVeg as groundtruth masks. Results show that the SegVeg approach allows to segment accurately the three classes. However, some confusion is observed mainly between the background and senescent vegetation, particularly over the dark and bright regions of the images. The U-net model achieves similar performances, with slight degradation over the green vegetation: the SVM pixel-based approach provides more precise delineation of the green and senescent patches as compared to the convolutional nature of U-net. The use of the components of several color spaces allows to better classify the vegetation pixels into green and senescent. Finally, the models are used to predict the fraction of three classes over whole images or regularly spaced grid-pixels. Results show that green fraction is very well estimated (R 2 = 0.94) by the SegVeg model, while the senescent and background fractions show slightly degraded performances (R 2 = 0.70 and 0.73, respectively) with a mean 95% confidence error interval of 2.7% and 2.1% for the senescent vegetation and background, versus 1% for green vegetation. We have made SegVeg publicly available as a ready-to-use script and model, along with the entire annotated grid-pixels dataset. We thus hope to render segmentation accessible to a broad audience by requiring neither manual annotation nor knowledge or, at least, offering a pretrained model for more specific use.

Confident memory errors and disrupted reality testing in early psychosis

Schizophrenia is a psychiatric disorder characterized by a disruption in reality testing most often manifest in the form of delusions and hallucinations. Because determining the reality-basis of prior experiences is dependent on episodic and associative memory, deficits in mnemonic processes could be involved in the genesis of impaired reality testing. In the present study, we used an associative memory paradigm incorporating confidence ratings to examine whether patients with a recent onset of schizophrenia (n = 48) show a greater propensity for confident, yet incorrect responses during retrieval testing than healthy controls (n = 26) and whether such confident incorrect responses, specifically, are more strongly associated with positive symptoms than with negative symptoms. Using an analysis of variance design, we found that first-episode schizophrenia patients made a significantly greater number of confident errors than controls (i.e. they expressed high confidence in having seen pairs of items that were not paired at encoding and high confidence in having not seen pairs of items that were paired at encoding). We also found that the number of confident errors was specifically and differentially associated with positive symptom severity, to a significantly greater degree than with negative symptom severity and psychosocial functioning, and this association was not found between positive symptoms and uncertain responses, nor positive symptoms and overall task performance. These findings suggest that the propensity for incorrect memory judgements with high confidence, specifically, may be uniquely associated with disrupted reality testing and that this type of cognitive impairment is distinct from general deficits in memory and cognition in this respect.

Neural correlates of metacognition across the adult lifespan

Metacognitive accuracy describes the degree of overlap between the subjective perception of one's decision accuracy (i.e. confidence) and objectively observed performance. With older age, the need for accurate metacognitive evaluation increases; however, error detection rates typically decrease. We investigated the effect of ageing on metacognitive accuracy using event-related potentials (ERPs) reflecting error detection and confidence: the error/correct negativity (Ne/c) and the error/correct positivity (Pe/c). Sixty-five healthy adults (20 to 76 years) completed a complex Flanker task and provided confidence ratings. We found that metacognitive accuracy declined with age beyond the expected decline in task performance, while the adaptive adjustment of behaviour was well preserved. Pe amplitudes following errors varied by confidence rating, but they did not mirror the reduction in metacognitive accuracy. Ne amplitudes decreased with age for low confidence errors. The results suggest that age-related difficulties in metacognitive evaluation could be related to an impaired integration of decision accuracy and confidence information processing. Ultimately, training the metacognitive evaluation of fundamental decisions in older adults might constitute a promising endeavour.

A new Bayesian approach toward improved regression of low-count U[sbnd]Pb geochronology data generated by LA-ICPMS

UPb dating of hydrogenic and fossil material can be a significant challenge because samples often contain low U levels and variable amounts of non-radiogenic Pb. Line scans using LA-ICPMS provide the widest range of ratios available for defining the radiogenic to common Pb mixing line, hence the best age estimate, but extracting this information requires processing many low-count data. Conventional least-squares regression of ratios based on such data can give inconsistent results. A new approach is introduced where data are regressed as count rates in a 3D space rather than ratios on a 2D plot. This is a maximum likelihood approach in that mixing planes covering a range of ages and common 207Pb/206Pb ratios (PbC) are compared to the data set and used to determine a relative probability density surface whose maximum should fix the best-fit age and common 207Pb/206Pb ratio. 95% confidence errors are determined from T and PbC values that cut off 2.5% of the area to the left and right of 2D plots generated by integration of the 3D plot over PbC and age, respectively. Results from a rugose coral fossil show that this approach gives age results similar to regressions of concordia data and more accurate than Tera-Wasserburg regressions on data from natural samples. Tests on randomly generated synthetic data do not show significant discrepancy between Isoplot ages but they do between values of PbC. The 3D approach is the only one that gives accurate results for both age and PbC.

Revisiting noncentrality-based confidence intervals, error probabilities and estimation-based effect sizes

The paper focuses on several misconceptions in frequentist inference that undermine the current discussions on replicability and the trustworthiness of empirical evidence, including, (a) the use/abuse of estimation-based effect sizes, (b) (mis)interpreting error probabilities as conditional on hypotheses giving rise to ostensible confusions between P(D|H0) and P(H0|D), and (c) the questionable nature of noncentrality-based Confidence Intervals (CIs). It is argued that when (a)–(c) are viewed in the context of model-based [Mθ(x)] frequentist inference can be shown to be grounded in serious misapprehensions. In particular, it is argued that: (i) estimation-based effect sizes are based on the unwarranted claim that a point estimate θ̂n(x0) approximates the ‘true’ value of the unknown parameter θ, (ii) frequentist error probabilities are not conditional on H0 and H1 since the parameters θ, in terms of which they are framed, are unknown constants, rendering both P(D|H0) and P(H0|D) meaningless, and (iii) the cogency of noncentrality CIs, such as Cohen’s d, is questionable because they conflate hypothetical (testing) and factual (estimation) reasoning, and as a result they ignore the duality between CIs and hypothesis testing, forfeiting optimality. The paper discusses several additional misconceptions relating to frequentist inference and proposes [i] optimal CIs for Cohen’s d, and [ii] cogent testing-based effect sizes originating in the post-data severity evaluation.

Analysis of Muslim Customer Loyalty Shopping for Fashion in the Traditional Inpres I Market in Kisaran

Customer loyalty is the customer's commitment to a brand, store, or supplier and is reflected in consistent repeat purchases. This customer loyalty can be influenced by product, price, service, and purchasing decisions. The purpose of this study was to determine the direct and indirect influence between product variables, prices, services, and purchasing decisions on the loyalty of Muslim customers shopping for fashion in the traditional Inpres I Market in Kisaran. Method This study uses a quantitative approach with path analysis and a sample of 100 respondents. The results of this study indicate that product, price, and service variables influence purchasing decisions by 66.6% and the remaining 33.4% are influenced by other variables. Whereas product, price, service, and purchasing decision variables affect Muslim customer loyalty in fashion shopping at Traditional Inpres I Market in Kisaran with AA contribution of 70.9% and the remaining 29.1% is influenced by other variables. From the results of the significance test of the indirect effect it was found that there was a significant indirect effect of product, price, and service variables on the loyalty of Muslim customers shopping for fashion through purchasing decisions as an intervening variable in the traditional Inpes I Market in the Range with a 95% confidence level or error rate 5%.

Multiple comparison test by Tukey’s honestly significant difference (HSD): Do the confident level control type I error

Examining a huge amount of data is a typical issue in any research process. However, different statistical processes and techniques play essential role to derive a meaningful conclusion from the presented enormous data. Control of type I error is highly essential for a researcher or statistician while dealing with comparisons test with more than two variables. Multiple testing statistical tests provides a structural system and minimizes the error rate by helping to derive meaningful accurate conclusions. Among the different multiple test procedures Tukey's honestly significant difference test (Tukey's HSD) is most common and popular techniques. The main objective of this study was to explore how significantly selection of confidence level or error rate can affect the rate of committing type I error while drawing conclusion. The effect of committing type I error with selection of confidence level or error rate was explored with citing suitable case study in a special education setting. The case study focuses cognitive effect of music on growth and enhancement of the motor behavior (running, jumping and sliding) of children with mild intellectual disability enrolled in the special school setting. ANOVA test was performed and the significance of selection of individual confidence level and simultaneous confidence level) in Tukey’s HSD test was described.

Ο έλεγχος μηδενικών υποθέσεων: διαδικασία, παρανοήσεις και μερικές προτάσεις για καλύτερες πρακτικές

The rationale of Null Hypothesis Significance Testing (NHST) is described, and the consequences of its hybridism are discussed. The paper presents examples published in “PSYCHOLOGY: The Journal of the HPS” refer to NHST and interpret its outcomes. We examined the 445 articles published between 1992 and 2010. We noted misuses of NHST and searched for any use of confidence intervals or error bars or use of these to support interpretation. Part of the paper focuses on the statistical-reform debate and provides detailed guidance about good statistical practices in the analysis of research data and the interpretation of findings. The proposed guide does not fall into the trap of mandating the use of particular procedures; it rather aims to support readers’ understanding of research results.

Aptitud médica ocupacional y perfil clínico en conductores profesionales de un centro de salud ocupacional en Lima Metropolitana, 2014 - 2018

Objetivos: Describir las características médicas ocupacionales de los conductores profesionales y los resultados toxicológicos presentes en accidentes de tránsito en un centro de salud ocupacional de Lima Metropolitana. 2014-2018. Material y métodos: Estudio descriptivo retrospectivo de análisis de datos de fuente secundaria. El tamaño de muestra analizado fue 11520 conductores profesionales de la ciudad de Lima metropolitana durante 2014 al 2018, seleccionado de base de datos innominada mediante un muestreo probabilístico de aleatorización simple. Se utilizó estadística descriptiva univariadas usando valores de frecuencias absolutas y relativas, cálculo de intervalos de confianza al 95% y diagramas de error para cada una de las variables, identificando las diferencias existentes en cada una de las categorías o niveles de la variable. El procesamiento de la información fue realizado en el programa estadístico R versión 3.6. Resultados: El 64% de conductores sufren de sobrepeso u obesidad, el 58% de los conductores sufren alguna enfermedad de visión, el 67% siempre estuvieron en vigilia en cada conducción, el 73% de los conductores estuvieron actos para la conducción automovilística y un 13.3% de los conductores que sufrieron un accidente de tránsito estuvieron bajo efecto de las drogas. Conclusiones: Los conductores profesionales atendidos en salud ocupacional presentan con frecuencia obesidad y presbicia como defecto visual; mientras que el 27% fueron calificados como aptos para sus funciones, pero con restricciones. Las drogas detectadas con mayor frecuencia en conductores con accidentes de tránsito fue el alcohol y la marihuana.

El catéter vesical Foley BIP es superior al Foley convencional en la reducción de las infecciones urinarias secundarias a catéter en pacientes hospitalizados

Introducción: La infección de tracto urinario asociada a catéter (ITUCV) es la causa más frecuente de infección asociada a los cuidados de la salud. Existen diversas estrategias para prevenir dichas infecciones. Se deben considerar nuevas estrategias como los catéteres vesicales BIP (catéteres con recubrimiento de plata, oro, paladio), como una alternativa para reducir las mismas.
 Objetivos: Comparar la incidencia de las ITUCV en pacientes con catéter vesical Foley BIP versus Foley convencional.
 Metodología: Se realizó un ensayo clínico con aleatorización simple, siguiendo el sistema de doble ciego. Se estableció un nivel de confianza del 95%, un error no mayor del 5%. Obteniendo un total de 140 casos, alcanzando así una muestra representativa.
 Resultados: 140 casos con una inclusión de 125 casos y una exclusión de 15 casos (11%). La media de edad fue 50 años ± 18, con un 52 % hombres y 48% mujeres. El tiempo observado promedio fue de 10 días ± 7. Se presentó una tasa de incidencia global de 25 ITUCV x 1000 días.
 Discusión: El uso del catéter urinario Foley es una práctica cotidiana. El uso de catéteres Foley BIP disminuyó la tasa de incidencia de infección de vías urinarias asociada a catéter.
 Conclusiones: Las ITUCV son la infección más frecuente asociado a cuidados de la salud. Las infecciones nosocomiales representan un indicador de calidad de atención de los pacientes, repercuten en la morbimortalidad y economía de los pacientes. Los catéteres con recubrimiento fueron una medida eficaz, disminuyendo de manera considerable la incidencia de ITUCV.

The BIP Foley catheter is superior to the conventional Foley catheter in reducing catheter-associated urinary tract infections in hospitalized patients

Background: Catheter-associated urinary tract infections (CAUTIs) are the most frequent cause of infection related to healthcare. There are different strategies for preventing them and new ones, such as BIP Foley catheters (catheters coated in silver, gold, and palladium), should be considered as an alternative in reducing said infections.
 Aims: To compare the incidence of CAUTIs in patients with BIP Foley catheters versus conventional Foley catheters.
 Materials and methods: A double-blind clinical trial with simple randomization was conducted, obtaining a representative sample of 140 cases. A 95% confidence interval and error not greater than 5% were established.
 Results: Of the 140 cases, 15 were excluded (11%), leaving a total of 125 cases. Mean patient age was 50 years ± 18; 52% were men and 48% were women. Mean time observed was 10 days ± 7. Overall incidence was 25 CAUTIs x 1000 days.
 Discussion: Foley catheter use is an everyday practice. BIP Foley catheters reduce the incidence of CAUTIs.
 Conclusions: CAUTIs are the most frequent infections associated with healthcare. Nosocomial infections are an indicator of the quality of patient care, negatively impacting morbidity and mortality, as well as patient economy. The use of coated catheters was an efficacious measure that considerably reduced the incidence of CAUTIs.

An Electrophysiological Dissociation of Encoding vs. Maintenance Failures in Visual-Spatial Working Memory.

Working memory (WM) performance varies substantially among individuals but the precise contribution of different WM component processes to these functional limits remains unclear. By analyzing different types of responses in a spatial WM task, we recently demonstrated a functional dissociation between confident and not-confident errors reflecting failures of WM encoding and maintenance, respectively. Here, we use event-related brain potentials to further explore this dissociation. Healthy participants performed a delayed orientation-discrimination task and rated their response confidence for each trial. The encoding-related N2pc component was significantly reduced for confident errors compared to confident correct responses, which is indicative of an encoding failure. In contrast, the maintenance-related contra-lateral delay activity was similar for these response types indicating that in confident error trials, WM representations – potentially the wrong ones – were maintained accurately and with stability throughout the delay interval. However, contra-lateral delay activity measured during the early part of the delay period was decreased for not-confident errors, potentially reflecting compromised maintenance processes. These electrophysiological findings contribute to a refined understanding of the encoding and maintenance processes that contribute to limitations in WM performance and capacity.

Information Theoretic Characterization of Uncertainty Distinguishes Surprise From Accuracy Signals in the Brain.

Uncertainty presents a problem for both human and machine decision-making. While utility maximization has traditionally been viewed as the motive force behind choice behavior, it has been theorized that uncertainty minimization may supersede reward motivation. Beyond reward, decisions are guided by belief, i.e., confidence-weighted expectations. Evidence challenging a belief evokes surprise, which signals a deviation from expectation (stimulus-bound surprise) but also provides an information gain. To support the theory that uncertainty minimization is an essential drive for the brain, we probe the neural trace of uncertainty-related decision variables, namely confidence, surprise, and information gain, in a discrete decision with a deterministic outcome. Confidence and surprise were elicited with a gambling task administered in a functional magnetic resonance imaging experiment, where agents start with a uniform probability distribution, transition to a non-uniform probabilistic state, and end in a fully certain state. After controlling for reward expectation, we find confidence, taken as the negative entropy of a trial, correlates with a response in the hippocampus and temporal lobe. Stimulus-bound surprise, taken as Shannon information, correlates with responses in the insula and striatum. In addition, we also find a neural response to a measure of information gain captured by a confidence error, a quantity we dub accuracy. BOLD responses to accuracy were found in the cerebellum and precuneus, after controlling for reward prediction errors and stimulus-bound surprise at the same time point. Our results suggest that, even absent an overt need for learning, the human brain expends energy on information gain and uncertainty minimization.

Neglected X-ray discovered polars

We report on the X-ray observations of the eclipsing polar HY Eri (RX J0501–0359), along with its photometric, spectrophotometric, and spectropolarimetric optical variations, collected over 30 years. With an orbital period of 2.855 h, HY Eri falls near the upper edge of the 2–3 h period gap. After 2011, the system went into a prolonged low state, continuing to accrete at a low level. We present an accurate alias-free long-term orbital ephemeris and report a highly significant period change by 10 ms that took place over the time interval from 2011 to 2018. We acquired a high-quality eclipse spectrum that shows the secondary star as a dM5–6 dwarf at a distance d = 1050 ± 110 pc. Based on phase-resolved cyclotron and Zeeman spectroscopy, we identify the white dwarf (WD) in HY Eri as a two-pole accretor with nearly opposite accretion spots of 28 and 30 MG. The Zeeman analysis of the low state spectrum reveals a complex magnetic field structure, which we fit by a multipole model. We detected narrow emission lines from the irradiated face of the secondary star, of which Mg Iλ5170 with a radial velocity amplitude of K′2 = 139 ± 10 km s−1 (90% confidence) tracks the secondary more reliably than the narrow Hα line. Based on the combined dynamical analysis and spectroscopic measurement of the angular radius of the WD, we obtain a primary mass of M1 = 0.42 ± 0.05 M⊙ (90% confidence errors), identifying it as a probable He WD or hybrid HeCO WD. The secondary is a main sequence star of M2 = 0.24 ± 0.04 M⊙ that seems to be slightly inflated. The large distance of HY Eri and the lack of similar systems suggest a very low space density of polars with low-mass primary. According to current theory, these systems are destroyed by induced runaway mass transfer, suggesting that HY Eri may be doomed to destruction. Over the last 30 years, HY Eri experienced high and low states with mass transfer rates that differed by three orders of magnitude, varying between Ṁ ≃ 10−9 M⊙ yr−1 and 10−12 M⊙ yr−1. At a galactic latitude of −26.1°, it is located about 500 pc below the galactic plane.

A Convolutional Neural Network for Real Time Classification, Identification, and Labelling of Vocal Cord and Tracheal Using Laryngoscopy and Bronchoscopy Video.

The use of artificial intelligence, including machine learning, is increasing in medicine. Use of machine learning is rising in the prediction of patient outcomes. Machine learning may also be able to enhance and augment anesthesia clinical procedures such as airway management. In this study, we sought to develop a machine learning algorithm that could classify vocal cords and tracheal airway anatomy real-time during video laryngoscopy or bronchoscopy as well as compare the performance of three novel convolutional networks for detecting vocal cords and tracheal rings. Following institutional approval, a clinical dataset of 775 video laryngoscopy and bronchoscopy videos was used. The dataset was divided into two categories for use for training and testing. We used three convolutional neural networks (CNNs): ResNet, Inception and MobileNet. Backpropagation and a mean squared error loss function were used to assess accuracy as well as minimize bias and variance. Following training, we assessed transferability using the generalization error of the CNN, sensitivity and specificity, average confidence error, outliers, overall confidence percentage, and frames per second for live video feeds. After the training was complete, 22 models using 0 to 25,000 steps were generated and compared. The overall confidence of classification for the vocal cords and tracheal rings for ResNet, Inception and MobileNet CNNs were as follows: 0.84, 0.78, and 0.64 for vocal cords, respectively, and 0.69, 0.72, 0.54 for tracheal rings, respectively. Transfer learning following additional training resulted in improved accuracy of ResNet and Inception for identifying the vocal cords (with a confidence of 0.96 and 0.93 respectively). The two best performing CNNs, ResNet and Inception, achieved a specificity of 0.985 and 0.971, respectively, and a sensitivity of 0.865 and 0.892, respectively. Inception was able to process the live video feeds at 10 FPS while ResNet processed at 5 FPS. Both were able to pass a feasibility test of identifying vocal cords and tracheal rings in a video feed. We report the development and evaluation of a CNN that can identify and classify airway anatomy in real time. This neural network demonstrates high performance. The availability of artificial intelligence may improve airway management and bronchoscopy by helping to identify key anatomy real time. Thus, potentially improving performance and outcomes during these procedures. Further, this technology may theoretically be extended to the settings of airway pathology or airway management in the hands of experienced providers. The researchers in this study are exploring the performance of this neural network in clinical trials.