Development of assessment tool for dyslexia screening using fuzzy logic

• Main Author: Ng, Li Mun
• Format: Thesis
• Language: English; English; English
• Published: 2021
• Subjects: TK1001-1841 Production of electric energy or power. Powerplants. Central stations
• Online Access: http://eprints.uthm.edu.my/992/1/24p%20NG%20LI%20MUN.pdf; http://eprints.uthm.edu.my/992/2/NG%20LI%20MUN%20COPYRIGHT%20DECLARATION.pdf; http://eprints.uthm.edu.my/992/3/NG%20LI%20MUN%20WATERMARK.pdf; http://eprints.uthm.edu.my/992/
• Institution: Universiti Tun Hussein Onn Malaysia

Dyslexia is a specific reading difficulty leading to low proficiency in an individual's literacy, affecting 5 to 17.5 % of the population globally and 5 to 15 % of Malaysia's children. The current dyslexia screening test known as Ujian Pengesanan Awal Disleksia Bahasa Melayu (D-Test) is administrated manually, and the decision to determine dyslexia risk is time-consuming. Hence, this research's main objective is to develop an assessment tool by implementing fuzzy logic for rapid dyslexia risk outcomes using D-Test as the primary screening framework. The Mamdani-type Fuzzy Inference System (FIS) was developed based on 48 rule conditions, whereas the user interface was built using MATLAB App Designer. The developed assessment tool's performance was evaluated based on quantitative (accuracy, sensitivity, specificity, and precision), qualitative (Technology Acceptance Model (TAM)), and system response time assessments. Phase I study was conducted using the D-Test, and Phase II was carried out using the developed assessment tool with the recruitment of school pupils (nPhase I=117 and nPhase II=74) and teachers (n=29), respectively. The outcome from the quantitative results from Phase I and Phase II demonstrated the capability of fuzzy logic to distinguish between dyslexic and non-dyslexic subjects with an accuracy of 88.89 % and 93.24 %, respectively. Meanwhile, the finding from the qualitative approach investigated using showed the perceptions of external control (R2 =0.575, p<0.05) and perceived usefulness (R2 =0.675, p<0.05) were significantly influencing the behavioural intention of the target users towards the developed dyslexia assessment tool. The final finding on the system response time highlighted the developed tool's capability to improve the time taken when determining the dyslexia risk level (15 seconds per subject). In conclusion, the assessment tool for a rapid dyslexia risk status utilising Mamdani-type FIS had been successfully developed. The developed dyslexia assessment tool could be beneficial to assist dyslexia organisations, parents and school teachers in dyslexia screening process.