Category: Uncategorized

Poster accepted in Intelligent Data Analysis 2018 – Mashael Al-Luhaybi

Title: Identification of Student “Types” from Online Self- Assessment Temporal Trajectories With Dynamic Time Warping for Performance Prediction

Authors: Mashael Al-Luhaybi, Leila Yousefi, Stephen Swift, Steve Counsell and Allan Tucker

Affiliations: Brunel University London (UK)

Conference website: https://ida2018.org

Spotlight Presentation Slides: Here

Summer Short Course – Data Analysis and R (11th-12th Jul)

Making Sense out of Software Engineering Data And an introduction to R

Prof Sandro Morasca, Università degli studi dell’Insubria, Italy

The FREE summer short course (funded by Erasmus+) was organised by Prof Martin Shepperd on 11-12 July, 2018 (13:00-17:00 in WLFB208).

The course addressed the techniques that can be sensibly used to extract knowledge out of Software Engineering data acquired via experiments or routine data collection in industrial contexts, to make it practically useful. The course described and critically discussed a number of data analysis techniques, by explaining their preconditions and their outcomes. The course illustrated both basic, traditional techniques and innovative ones, like those based on Robust Regression or machine learning.  Also, it  explained how the models obtained can be validated.

A big thank you to Sandro and Martin for running this fantastic short course.

Lecture slides can be found here.

CBMS 2018 Best Student Paper – Leila Yousefi

Congratulations to Leila Yousefi who won best student paper at IEEE CBMS 2018. The paper is titled “Predicting Disease Complications Using a Step-Wise Hidden Variable Approach for Learning Dynamic Bayesian Networks”

Below is the abstract and full list of authors.

Predicting Diabetes Type 2 Mellitus (T2DM) complications such as retinopathy and liver disease is still a challenge despite being a growing public health concern worldwide. This is due to the complex interactions between complications and other features, as well as between the different complications, themselves. What is more, there are likely to be many unmeasured effects that impact the disease progression of different patients. Probabilistic graphical models such as Dynamic Bayesian Networks (DBNs) have demonstrated much promise in the modeling of disease progression and they can naturally incorporate hidden (latent) variables using the EM algorithm. Unlike deep learning approaches that attempt to model complex interactions in data by using a large number of hidden variables, we adopt a different approach. We are interested in models that not only capture unmeasured effects but are also transparent in how they model data so that knowledge about disease processes can be extracted and trust in the model can be maintained by clinicians. As a result, we have developed a step-wise hidden variable structure learning process that incrementally adds hidden variables based on the IC* algorithm. To the best of our knowledge, this is the first study for classifying disease complication using a step-wise learning methodology for identifying hidden and T2DM features with a DBN structure from clinical data. Our extensive set of experiments show that the proposed method improves classification accuracy, identifying the correct number of hidden variables, and targeting their precise location within the network structure.

Leila Yousefi, Allan Tucker, Mashael Al-luhaybi, Lucia Saachi, Riccardo Bellazzi and Luca Chiovato.

Welly done Lilly!

BIOIMAGING 2018 Best Student Paper Award – Bashir Dodo

We are pleased to announce that Bashir Dodo’s paper “Graph-Cut Segmentation of Retinal Layers from OCT Images” has won the BIOIMAGING 2018 Best Student Paper Award.

Below is the abstract and full list of authors.

ABSTRACT
The segmentation of various retinal layers is vital for diagnosing and tracking progress of medication of various ocular diseases. Due to the complexity of retinal structures, the tediousness of manual segmentation and variation from different specialists, many methods have been proposed to aid with this analysis. However image artifacts, in addition to inhomogeneity in pathological structures, remain a challenge, with negative influence on the performance of segmentation algorithms. Previous attempts normally pre-process the images or model the segmentation to handle the obstruction but it still remains an area of active research, especially in relation to the graph based algorithms. In this paper we present an automatic retinal layer segmentation method, which is comprised of fuzzy histogram hyperbolization and graph cut methods to segment 8 boundaries and 7 layers of the retina on 150 OCT B-Sans images, 50 each from the temporal, nasal and centre of foveal region. Our method shows positive results, with additional tolerance and adaptability to contour variance and pathological inconsistency of the retinal structures in all regions.

Bashir Isa Dodo, Yongmin Li, Khalid Eltayef and Xiaohui Liu.

Congratulations again!