Accuracy In Such Applications Research Articles

Deep learning in the marking of medical student short answer question examinations: Student perceptions and pilot accuracy assessment

Introduction: Machine learning has previously been applied to text analysis. There is limited data regarding the acceptability or accuracy of such applications in medical education. This project examined medical student opinion regarding computer-based marking and evaluated the accuracy of deep learning (DL), a subtype of machine learning, in the scoring of medical short answer questions (SAQs). Methods: Fourth- and fifth-year medical students undertook an anonymised online examination. Prior to the examination, students completed a survey gauging their opinion on computer-based marking. Questions were marked by humans, and then a DL analysis was conducted using convolutional neural networks. In the DL analysis, following preprocessing, data were split into a training dataset (on which models were developed using 10-fold cross-validation) and a test dataset (on which performance analysis was conducted). Results: One hundred and eighty-one students completed the examination (participation rate 59.0%). While students expressed concern regarding the accuracy of computer-based marking, the majority of students agreed that computer marking would be more objective than human marking (67.0%) and reported they would not object to computer-based marking (55.5%). Regarding automated marking of SAQs, for 1-mark questions, there were consistently high classification accuracies (mean accuracy 0.98). For more complex 2-mark and 3-mark SAQs, in which multiclass classification was required, accuracy was lower (mean 0.65 and 0.59, respectively). Conclusions: Medical students may be supportive of computer-based marking due to its objectivity. DL has the potential to provide accurate marking of written questions, however further research into DL marking of medical examinations is required.

Joint Modelling of Cyber Activities and Physical Context to Improve Prediction of Visitor Behaviors

This article investigates the cyber-physical behavior of users in a large indoor shopping mall by leveraging anonymized (opt in) Wi-Fi association and browsing logs recorded by the mall operators. Our analysis shows that many users exhibit a high correlation between their cyber activities and their physical context. To find this correlation,propose a mechanism to semantically label a physical space with rich categorical information from DBPedia concepts and compute a contextual similarity that represents a user’s activities with the mall context. We demonstrate the application of cyber-physical contextual similarity in two situations: user visit intent classification and future location prediction. The experimental results demonstrate that exploitation of contextual similarity significantly improves the accuracy of such applications.

Review and comparison of geometric distortion correction schemes in MR images used in stereotactic radiosurgery applications

In Stereotactic Radiosurgery (SRS), MR-images are widely used for target localization and delineation in order to take advantage of the superior soft tissue contrast they exhibit. However, spatial dose delivery accuracy may be deteriorated due to geometric distortions which are partly attributed to static magnetic field inhomogeneity and patient/object-induced chemical shift and susceptibility related artifacts, known as sequence-dependent distortions. Several post-imaging sequence-dependent distortion correction schemes have been proposed which mainly employ the reversal of read gradient polarity. The scope of this work is to review, evaluate and compare the efficacy of two proposed correction approaches. A specially designed phantom which incorporates 947 control points (CPs) for distortion detection was utilized. The phantom was MR scanned at 1.5T using the head coil and the clinically employed pulse sequence for SRS treatment planning. An additional scan was performed with identical imaging parameters except for reversal of read gradient polarity. In-house MATLAB routines were developed for implementation of the signal integration and average-image distortion correction techniques. The mean CP locations of the two MR scans were regarded as the reference CP distribution. Residual distortion was assessed by comparing the corrected CP locations with corresponding reference positions. Mean absolute distortion on frequency encoding direction was reduced from 0.34mm (original images) to 0.15mm and 0.14mm following application of signal integration and average-image methods, respectively. However, a maximum residual distortion of 0.7mm was still observed for both techniques. The signal integration method relies on the accuracy of edge detection and requires 3-4 hours of post-imaging computational time. The average-image technique is a more efficient (processing time of the order of seconds) and easier to implement method to improve geometric accuracy in such applications.

Bus travel time prediction using a time-space discretization approach

The accuracy of travel time information given to passengers plays a key role in the success of any Advanced Public Transportation Systems (APTS) application. In order to improve the accuracy of such applications, one should carefully develop a prediction method. A majority of the available prediction methods considered the variation in travel time either spatially or temporally. The present study developed a prediction method that considers both temporal and spatial variations in travel time. The conservation of vehicles equation in terms of flow and density was first re-written in terms of speed in the form of a partial differential equation using traffic stream models. Then, the developed speed based equation was discretized using the Godunov scheme and used in the prediction scheme that was based on the Kalman filter. From the results, it was found that the proposed method was able to perform better than historical average, regression, and ANN methods and the methods that considered either temporal or spatial variations alone. Finally, a formulation was developed to check the effect of side roads on prediction accuracy and it was found that the additional requirement in terms of location based data did not result in an appreciable change in the prediction accuracy. This clearly demonstrated that the proposed approach based on using vehicle tracking data is good enough for the considered application of bus travel time prediction.

High-Precision Vehicle Navigation in Urban Environments Using an MEM's IMU and Single-Frequency GPS Receiver

Many applications demand high-precision navigation in urban environments. Two frequency real-time kinematic (RTK) Global Positioning System (GPS) receivers are too expensive for low-cost or consumer-grade projects. As single-frequency GPS receivers are getting less expensive and more capable, more people are utilizing single-frequency RTK GPS techniques to achieve high accuracy in such applications. However, compared with dual-frequency receivers, it is much more difficult to resolve the integer ambiguity vector using single-frequency phase measurements and therefore more difficult to achieve reliable high-precision navigation. This paper presents a real-time sliding-window estimator that tightly integrates differential GPS and an inertial measurement unit to achieve reliable high-precision navigation performance in GPS-challenged urban environments using low-cost single-frequency GPS receivers. Moreover, this paper proposes a novel method to utilize the phase measurements, without resolving the integer ambiguity vector. Experimental results demonstrate real-time position estimation performance at the decimeter level. Furthermore, the novel use of phase measurements improves the robustness of the estimator to pseudorange multipath error.

Evaluation and correction of patient-induced distortion in MR images used in stereotactic radiosurgery

Introduction Geometric disposition in MR images is partly attributed to patient/object induced distortions stemming from susceptibility differences and chemical shift artifacts. Purpose To develop and implement a procedure for patient-induced distortion characterization, as well as to evaluate and compare three distortion correction schemes in MR images used for target localization in stereotactic radiosurgery. Materials and methods A specially designed phantom was modified to accommodate two cylindrical-shape volumes lying in close proximity to control points (CPs) for distortion correction. These volumes, simulating brain metastases, were filled with Gd-solution, while the rest phantom was filled with standard copper sulfate solution. Following 1.5T MRI, Gd-induced distortion was estimated using CPs disposition analysis. Moreover, three distortion correction schemes were employed: the phase difference map (based on a dual-echo gradient echo sequence), as well as the average-image and the signal integration techniques (both based on frequency-encoding direction reversal). Residual distortion was assessed by comparing with corresponding CT scans. Results Gd-induced distortion results in a structure disposition of the of the order of 1 mm. The phase difference method proves excellent at high-signal regions but lack of phase information at low-signal areas results in unwrapping errors. The signal integration method relies on the accuracy of edge detection and noise removal at low-signal areas. The average-image technique is a simple and efficient method to improve geometric accuracy in such applications. Conclusion Patient-induced distortion in MR images should be considered and corrected for high fidelity target localization. One should be aware of the relative advantages/drawbacks of alternative correction schemes.

Radiology smartphone applications; current provision and cautions

ObjectivesMedical smartphone applications are increasingly popular amongst doctors. However, the quality of their content is variable. We assessed contemporary radiology-related smartphone applications, focussing on the level of advertised medical involvement in application development.MethodsSix major application stores were searched between 18-30 June 2012 using the terms radiology, radiation, x-ray(s), computed tomography/CT, magnetic resonance imaging/MRI, ultrasound, nuclear medicine, fluoroscopy and mammography/mammogram. Application ratings, cost and medical input in development were recorded.Results321 applications were identified. One hundred fifty-eight were "teaching" and 96 “reference”. Three of the 29 DICOM viewing applications had FDA approval for primary diagnosis, while 62 % stated they should not be used for primary diagnosis; 24 % of applications stated named medical professional involvement, 12 % had unnamed medical involvement and 4 % acknowledged guidelines or papers; 42 % did not disclose authorship.ConclusionsA large variety of radiology-related smartphone applications are available with many potential benefits. Advertised medical involvement in application design is variable, making assessment of their accuracy difficult prior to purchase. Additional measures are required to ensure the accuracy of such applications. The limitations of image interpretation using smartphones are a major drawback of DICOM viewing applications. Further research into the accuracy of primary diagnosis using such applications is needed.Main Messages • A large variety of radiology smartphone applications are available with many potential benefits • Variable medical involvement in application design limits assessment of accuracy before purchase • Limitations of image interpretation using smartphones are a drawback of DICOM viewing applications • Further work on the accuracy of primary diagnosis using these DICOM viewing applications is needed

An introduction to recursive partitioning: Rationale, application, and characteristics of classification and regression trees, bagging, and random forests.

Recursive partitioning methods have become popular and widely used tools for nonparametric regression and classification in many scientific fields. Especially random forests, which can deal with large numbers of predictor variables even in the presence of complex interactions, have been applied successfully in genetics, clinical medicine, and bioinformatics within the past few years. High-dimensional problems are common not only in genetics, but also in some areas of psychological research, where only a few subjects can be measured because of time or cost constraints, yet a large amount of data is generated for each subject. Random forests have been shown to achieve a high prediction accuracy in such applications and to provide descriptive variable importance measures reflecting the impact of each variable in both main effects and interactions. The aim of this work is to introduce the principles of the standard recursive partitioning methods as well as recent methodological improvements, to illustrate their usage for low and high-dimensional data exploration, but also to point out limitations of the methods and potential pitfalls in their practical application. Application of the methods is illustrated with freely available implementations in the R system for statistical computing.

Numerical analysis of thick plates

Bending behaviour of a rectangular plate is analysed with the help of a refined higher order theory. The theory is based on a higher order displacement model and the three-dimensional Hooke's laws for plate material, giving rise to a more realistic quadratic variation of the transverse shearing strains and linear variation of the transverse normal strain through the plate thickness. It is shown that the segmentation method for the numerical analysis of plates simply supported on two opposite edges can be considered to be the most competitive method in terms of efficiency, economy, reliability and accuracy in such applications