Table 5.

Review of AI assistive tools used to address learning disabilities of students with NDDs.

Author/Year	AI Tool	Features/Model Used for Training	Type of Technology	Learning Area Addressed	Effectiveness
AI Assistive Tools Used to Teach Students with ADHD.
2014 [61]	KAR robot	-	Assistive technology	Improve social skills via storytelling.	Improves children’s cognitive performance.
2015 [62]	Child activity sensing and training tool	-42 features (users’ physical and physiological) -Machine learning algorithm (not specified)	Real time assistive technology	Real-time assistive tool that tracks activities and helps students sustain attention.	An assistive intervention that is based on a smartphone and has the potential to aid a child with ADHD who has lost focus in his/her work.
2018 [63,64]	WatchMinder vibrating watch	-Sensors that collect activity and behaviour data -Bespoke algorithm/application	Wearable technology	Helps to send constant reminders to students to refocus on their work.	The watch has been effective as a simple memory aid for ADHD children with the auditory or vibrating alarm feature. The watch has been found to be affordable, durable, dependable and effective by users [65]
2018 [63]	Speech recognition software (Dragon Naturallyspeaking(Dragon Sytems company, United States, version 15, /Voice Finger/ViaTalk (LLC Company, New York/Tazti (Voice Tech Group company, United States)	-Audio data -Deep learning models (spectrograms/filter banks [66]	Assistive technology	Replaces writing activity with speech to allow students to express themselves efficiently without tiring themselves	Dragon Naturallyspeaking, Voice Finger, Via Talk, Tazti softwares have been reported to be beneficial to students with ADHD and resulted in improvement in the areas of writing, reading and spelling [65]
2018 [63,67]	Talking calculators	-User’s data such as pressing of numbers -Built-in speech synthesiser	Assistive technology	Helps students hear and process numbers easily for mathematics.	Students are able to complete assessments faster with the help of the calculator and has helped students gain independence [67]
AI assistive tools used to teach students with Dyslexia.
2013 [68]	Intelligent dyslexic system	Machine learning algorithm, visualisation concept	Assistive technology	Helps students gain knowledge on alphabets and letters	The technology has the potential to improve the reading and writing skills of students.
2014 [69]	Agent DYSL adaptive reading system	Machine learning algorithm, Mel-frequency cepstral coefficients, discrete cosine transform	Assistive technology	Enables the personalisation of reading environment of Greek students.	Students’ reading pace and accuracy were increased.
2015 [70]	Computer-based learning model	Machine learning technique	-	Explores the use of machine learning method to improve effectiveness of learning process.	-
2017 [71]	Applications for reading and writing (Learning Ally, Natural reader, dyslexia quest, sound literacy, ginger page, v books pdf voice reader, openWeb, reading intro by OZ phonics, OCR instantly pro, MindMeister)	Generation of audio files, pytorch deep convolutional text-to-speech models (PytorchDcTts)	Digital application	Helps students with reading and writing skills.	Audiobooks, such as learning Ally, has enabled students to gain confidence, independence and success [72].
2018 [73]	DIMMAND, capturaTalk application	-Chatbots	Digital application	Provides tailored interventions for difficulties encountered in literacy.	Information is not available.
2020 [74]	Voice dream reader, natural reader, web reader	-Text, audio data -Voicebot	Digital application	Helps with building reading skills.	E-readers have been found to generally improve reading speed and comprehension as compared to reading on paper [75].
2020 [76,77]	DytectiveU	-Learning patterns of students -Support vector machine algorithm	Digital application	Provides personalised game-based exercises to enhance specific cognitive skills related linked to dyslexia.	The DytectiveU application is reported to be able to offer students a variety of actions that are helpful in the learning of reading and writing [78].
2020 [79]	Generative adversarial network	Conversion of image/speech to text	Assistive technology	Converts natural language text to images to aid students in their learning.	-
AI assistive tools used to teach students with ASD.
2011 [80]	LIFEisGAME game	-Shape and appearance-based facial features -Second feedback loop, visual input from player obtained through webcam	Digital application	To teach students to recognise facial emotions.	Information is not available.
2017 [81,82]	‘Empower me’ application	-Emotion recognition features -Google smartglass, augmented reality environment, web-based dashboard to monitor progress	Wearable technology	Encourages social interaction between user and peers/educators.	Students were able to improve their social skills using the Google glass. It was also reported to be fun, useful and engaging [83].
2018 [84,85]	Kaspar robot	-Sensory data -Reinforcement learning algorithm [86]	Assistive technology	Helps enhance social interaction skills.	The human-like body and features of Kaspar have been reported to help an ASD student to be more interactive [87].
2018 [88]	ABA flashcards- Emotions, Autism emotion, conversation builder, emotions and feelings- autism, Find me, Kid in storybook maker, learning with Rufus, Look in my eyes: Train engineer, Model me going places 2, Pictello, Social stories, Special stories, The social express, Toca Boca	-Deep learning/machine learning algorithms using unstructured data	Digital application	Teaches social skills	Information is not available.
2018 [88]	ABA find it, Agnitus, Autism learning games- camp discovery, Intro to letters, Intro to Math, Math Bingo, Pop Math, Starfall ABC, Word wagon	-Deep learning/machine learning algorithms using unstructured data	Digital application	Helps in different learning areas	The camp discovery enabled participants to show high learning rates over a short period of time. It has been suggested that the application teaches the selected skills effectively [89].
2019 [90]	Emotify game	-Audio features, such as pitch of voice -Speech data to train Random forest classifier	Digital application	Helps students to recognise and express feelings.	The application caused participants to experience more engagement and exhibit higher behavioural intentions towards it [91].
2019 [92,93]	Milo, NAO, Pepper, Aisoy 1, Keepon robots	-Supervised machine learning algorithms; generalised models trained on users’ data and individualised models trained on initial subclass of users’ data [94]	Assistive technology	Helps build social and communication skills.	Social robots, such as NAO, have been reported to improve social skills in students, especially in terms of eye contact and concentration. Nonverbal children also reportedly started pronouncing some words [95].
2020 [88,96]	GoTalks speech generating device, AAC speech buddy, Proloquo2go, talking Larry, Touch chat HD, VAST autism 1-Core	-Behavioural features, such as grabbing, vocalisations -Deep learning/machine learning algorithms, neural networks [97]	Augmentative/alternative communication device	Helps with building communication skills.	Review studies report that high-technology speech generating devices are very effective in teaching manding, intraverbal and multistep tacting to ASD students [98].
2020 [96,99,100]	Facesay games	-Scores on social interactions -Facial expression recognition techniques, interactive environment with lifelike avatars	Digital application	Software games help recognise behavioural and emotional clues and enhance social skills.	Facesay application is found to be very promising, cost-effective and efficient for teaching affect recognition and mentalising constructs to high-functioning ASD students [101].
2020 [102]	Personalised ‘Kiwi’ robot for learning	-Video and audio data, such as eye contact and verbal dialogue -Supervised machine learning algorithms; generalised models trained on users’ data and individualised models trained on initial subclass of users’ data [94]	Assistive technology	Adapts lessons according to students’ changing needs.	Kiwi robot has been reported to improve the Maths skills and social skills in ASD students who were part of the study group [103].
2020 [96]	Life skills winner application	-Deep learning/machine learning algorithms using unstructured data	Digital application	Teaches students daily living skills through the application.	Information is not available.
2020 [103]	PvBOT robot	LEGO Mindstorms EV3 model	Assistive technology	Helps to teach students ‘place value’ concept in Mathematics.	PvBOT is helpful in motivating students to pay attention and stay focused for a longer period.
2021 [104]	Squizzy educational software	Scrum methodology	Assistive technology	Helps children stay focused during activities that involve cognition, such as colour selection or using pictures.	Effective in the cognitive aspect of therapy.