7COM1084 Specialism Research Report

Assignment Brief: (Word Count: 1400)

In order to conduct independent research in computer science you must fully understand the research area. You must have a grasp of the current open questions in this area, as well as any common techniques used to solve problems.

In this assignment you must provide an overview of your MSc Research specialism (AI / Networking / Cyber Security / Software Engineering / Data Science). Students without a specialism may choose any one of these specialisms.

You must discuss the research question presented within the relevant specialist lecture for this specialism, propose some research approaches to investigate this question and identify further work you might undertake which builds on this. You must also explore your personal strengths in this area.



                   Topics in Computer Science

1.    Introduction

Data Science has been one of the booming areas in today’s era. The work of this particular domain can be seen in any field be it education, health sector, hospitality sector etc (Bagla, 2005). The report reviews an article from the health sector in which remarkability indicates the use of data science. The article discusses recognizing dementia in British Sign Language BSL signatories. BSL is a common biological dialect that, like some other languages, expresses itself through hand movements, body, as well as face. Recognizing dementia in BSL signatories, on the other hand, is still an accessible investigation, as there is very little data regarding cognitive impairment in this community (Sinoff and Badarny, 2014). This is aggravated further by the scarcity of clinicians who have effective communications knowledge and skills operating with BSL customers. Dementia treatment is dependent on the accuracy of intelligence tests as well as BSL translators (Bahia and Viana, 2009). As a result, the Deaf community presently has disparities in access to delivery of healthcare for procured neuromuscular disorders, leading to negative quality and higher healthcare expenses (Olson and Swabey, 2017). In response to these challenges, this paper proposed a deep learning-based scientific framework and created a toolkit process of automatically detecting early signs of dementia without using sign translation software or explanation.

2.    Research Question

  • Is the signing space envelop correlated with mild cognitive impairment and, in particular, early stages of dementia?

The research question focuses on identifying the tendency of older people associated with the increased pervasiveness of procured cognitive deficits including such dementia. While there is no cure for this disease, an early diagnosis aids in receiving the required support as well as medicines. Scientists are trying to produce efficient technology devices that will aid the physician in the early detection of cognitive disorders. (Liang, Kapetanios, Woll and Angelopoulou, 2019) conducted a study for dementia in older Deaf signers of British Sign Language (BSL). The study mainly focused on screening of dementia problems of ageing singers in BSL. Dementia in older Deaf signers of British Sign Language (BSL) presents extra challenges because the diagnosis is dependent upon several factors including translator availability and quality, and also suitable quizzes as well as memory. With the help of this research, the early symptoms of dementia in Deaf signers of BSL could be detected using advanced technologies and data science. The study suggests a multi-modal regression automated identification toolkit for BSL customers inside the stages of dementia, in which highlights from different parts of the body contributing to the sign wrapper, such as arm movements as well as facial gestures, are merged.

Data science techniques can help in identifying dementia at an early stage. Deep learning-relied approaches for image/video identification and interpretation, but on the other side, are showing promise, especially the use of Convolutional Neural Networks (CNN), which also involve huge volumes of information (Marzban et al., 2019). Thus, making it an interesting area for the study. The early identification of the disease may help in the proper diagnostic of the problem.

3.    Literature Review

The latest advancement in technology and access to high diagnostic imaging, have enhanced the support for creating deep learning methods for quantitation of illnesses including Alzheimer’s as well as dementia (Iizuka, Fukasawa and Kameyama, 2019). Most of these methods have indeed been implemented to the categorization of MR image analysis, CT scan image processing, FDG-PET scan image analysis, or the merged image analysis The study tries to compare MCI sick. To differentiate various types or phases of MCI (Fernández, Rabin, Coifman and Eckstein, 2014). Jo et al., (2016) conducted a systematic review of deep learning concerning dementia and concluded that the study was using a mixture of machine learning as well as CNN machine learning strategies. The disadvantage of the study was the limited set of data (Liu, Xi, Zhao and Li, 2019). (Shiino, Shirakashi, Ishida and Tanigaki, 2021) studied the aspects of diagnosis of dementia, different machine learning strategies were increasingly used, the method used were imaging information for diagnosis as well as disease progression and far less regularly in non-imaging research concentrated on demographic information, focuses on those aspects, and unobtrusive tracking of gait patterns. The walking pace, as well as its daily variance, were an early indicator of MCI advancement. As the real-time hand path tracking comment thread does have the possibility to monitor a patient’s overall movement patterns as well as face identification, this and many other real-time metrics of feature could over unusual ways of sensing transformation stages going to lead to dementia, that could be another possible research extended version to the toolbox (Wang, Wang, Liu and Li, 2019). AVEID, a novel approach presented in, employs an automatic system to evaluate dementia interaction, concentrating on actions on observation-based weights as well as emotion recognition. AVEID concentrated on passive interaction in gaze as well as feelings identification, whereas the method used in this paper focuses on the symbol as well as facial action detection in an energetic ability to sign discussion (Iizuka, Fukasawa and Kameyama, 2019).

Though the considered articles provide extensive knowledge about the topic these previous studies have various research gaps. The work lacks evidence about screening techniques for detecting dementia.  Various techniques of advanced data science could be optimised to increase detection efficiency. The data size is another problem identified in the existing work. In some studies, the patients suffering from dementia are taken as test cases thus it also poses a challenge to data validation and information reliability as due to the disease the information can’t be considered reliable. 

4.    Research Approach

In the selected research paper (Liang, Kapetanios, Woll and Angelopoulou, 2019). The secondary research approach has been followed. The dataset consists of video clipping of 250 Deaf signers of BSL. These videos recording have been taken from 8 areas of the United Kingdom. The subjects for the study are aged between 50-90. The study showed that there could be distinctions in the wrapper of symbol space (symbol trajectories/depth/speed) as well as facial gestures among signatories with dementia as well as healthy signatories. Such clinical experimental observations that signatories with dementia have useless sign space and less facial expression than a healthy deaf control system. In this sense, this research focused not just on arm movements, as well as on other aspects of the BSL body of the patient, such as facial gestures. A multi-modal extracting features sub-network influenced by actual clinical necessities, as well as the experimental results affiliated with the sub-network, in this article. To categorise a BSL signatory as good health or anomalous, distinct derived movement characteristics will indeed be supplied within the CNN network. The research studies various attributes which contribute to analysing the information more accurately. As signatories with dementia useless sign space and less facial expression observing hand and body movement had proven helpful in the investigation. Though the approach has many advantages there are limitations as well. Getting accurate results would be a challenge as due to dementia the respondent may select the wrong options. Secondly, there is no way to identify that the suggested technique used for analysis would result accurately.

The other method that could be followed to carry out this research is through classification. The patients suffering from dementia could be classified based on their age and stage of dementia. There is a possibility that a patient at the initial stage of dementia would respond differently in comparison to the patient at a later stage thus resulting in variation in results. Different techniques of ML can be applied to these classified groups.

5.    Personal Investment

As a student of computer science, this area has been an intriguing area of study for me. The methods proposed in the suggested research paper could be applied to another file. I would like to analyse the extent of these suggested methods. The knowledge of ML will not only help me achieve my goal but will also broaden my understanding of the area of research.


  • Bagla, P., 2005. Booming Computer Sector Seen as a Mixed Blessing. Science, 310(5755), pp.1754-1754.
  • Bahia, V. and Viana, R., 2009. Accuracy of neuropsychological tests and the Neuropsychiatric Inventory in differential diagnosis between Frontotemporal dementia and Alzheimer’s disease. Dementia & Neuropsychologia, 3(4), pp.332-336.
  • Fernández, Á., Rabin, N., Coifman, R. and Eckstein, J., 2014. Diffusion methods for aligning medical datasets: Location prediction in CT scan images. Medical Image Analysis, 18(2), pp.425-432.
  • Iizuka, T., Fukasawa, M. and Kameyama, M., 2019. Deep Learning-Based Imaging Classification Identified Cingulate Island Sign In Dementia With Lewy Bodies. Alzheimer’s & Dementia, 15, pp.P382-P383.
  • Iizuka, T., Fukasawa, M. and Kameyama, M., 2019. Deep-learning-based imaging-classification identified cingulate island sign in dementia with Lewy bodies. Scientific Reports, 9(1).
  • Liang, X., Kapetanios, E., Woll, B. and Angelopoulou, A., 2019. Real Time Hand Movement Trajectory Tracking for Enhancing Dementia Screening in Ageing Deaf Signers of British Sign Language. Lecture Notes in Computer Science, pp.377-394.
  • Liu, H., Xi, L., Zhao, Y. and Li, Z., 2019. Using Deep Learning and Machine Learning to Detect Epileptic Seizure with Electroencephalography (EEG) Data. Machine Learning Research, 4(3), p.39.
  • Marzban, E., Teipel, S., Buerger, K., Fliessbach, K., Heneka, M., Kilimann, I., Laske, C., Peters, O., Priller, J., Schneider, A., Spottke, A., Wagner, M., Wiltfang, J., Düzel, E., Jessen, F. and Dyrba, M., 2019. Explainable Convolutional Networks And Multimodal Imaging Data: The Next Step Towards Using Artificial Intelligence As Diagnostic Tool For Early Detection Of Alzheimer’s Disease. Alzheimer’s & Dementia, 15, pp.P1083-P1084.
  • Olson, A. and Swabey, L., 2017. Communication Access for Deaf People in Healthcare Settings: Understanding the Work of American Sign Language Interpreters. Journal for Healthcare Quality, 39(4), pp.191-199.
  • Shiino, A., Shirakashi, Y., Ishida, M. and Tanigaki, K., 2021. Machine learning of brain structural biomarkers for Alzheimer’s disease (AD) diagnosis, prediction of disease progression, and amyloid beta deposition in the Japanese population. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 13(1).
  • Sinoff, G. and Badarny, S., 2014. Mild Cognitive Impairment, Dementia, and Affective Disorders in Essential Tremor: A Prospective Study. Tremor and Other Hyperkinetic Movements, 4(0), p.227.
  • Wang, T., Wang, W., Liu, H. and Li, T., 2019. Research on a Face Real-time Tracking Algorithm Based on Particle Filter Multi-Feature Fusion. Sensors, 19(5), p.1245.
Scroll to Top