Authors:
Krause, André Frank, andrefrank.krause@hochschule-rhein-waal.de, HSRW
Kannen, Kyra, kyra.kannen@hochschule-rhein-waal.de, HSRW
Büscher, Sarah, sarah.buescher@hochschule-rhein-waal.de, HSRW
Ressel, Christian, Christian.Ressel@hochschule-rhein-waal.de, HSRW
Wild-Wall, Nele, Nele.Wild-Wall@hochschule-rhein-waal.de, HSRW
Keywords: ADHD, reading support, continual learning, explainable AI
Abstract:
Attention deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder often linked to reading difficulties. Symptoms such as inattention, mind-wandering, and atypical eye movements contribute to slower reading and impaired comprehension. While ADHD medication and behavioral therapy can significantly reduce symptoms, they do not directly address the specific challenges faced during reading.
Research shows that individuals diagnosed with ADHD demonstrate atypical eye movement patterns, which have been shown to contribute to difficulties experienced when reading. Compared to healthy subjects, significant disparities in fixation patterns and saccade dynamics, including increased fixation durations [d10] as well as frequency of fixations [m20], and increased forward and backward saccades [c22] were observed. EEG-based markers of neural activity linked to attention like changes in alpha power [b14] or the theta/beta ratio [s19] and cortical potentials like the P300 can be potential biomarkers for ADHD attention estimation [m22].
We aim to investigate machine learning (ML)-backed methods of neuroadaptive reading assistance that enhance reading performance and comprehension in ADHD patients. A combination of real-time neurophysiological data, including eye-tracking metrics and EEG + ECG signals, will be used to continuously monitor the user’s attentional state and cognitive engagement.
ML models will be trained to estimate the current attentional state of a person. If a decline in attention is detected, the reading conditions will be dynamically adjusted to support refocusing on the reading task, e.g., by changing the text formatting or visual highlighting of the relevant text areas. Different refocusing methods need to be evaluated to test their effectiveness in improving reading fluency and text comprehension.
To date, no reliable and objective neuromarker for ADHD is known for clinical use. Inspecting trained ML models using methods from Explainable AI may reveal previously unknown neuromarkers for ADHD, provide insights into attentional processes during reading, discover spatiotemporal dependencies among brain areas [km22] and improve transparency and interpretability of model decisions. Further, continual learning methods [w24] will be applied to improve the adaptability of ML models to individual eye tracking and EEG characteristics of a specific person.
[d10] Deans, P., O’Laughlin, L., Brubaker, B., Gay, N., & Krug, D. (2010). Use of eye movement tracking in the differential diagnosis of attention deficit hyperactivity disorder (ADHD) and reading disability. Psychology, 1(4), 238-246.
[m20] Molina, R., Redondo, B., Vera, J., García, J. A., Muñoz-Hoyos, A., & Jiménez, R. (2020). Children with attention-deficit/hyperactivity disorder show an altered eye movement pattern during reading. Optometry and Vision Science, 97(4), 265-274.
[c22] Caldani, S., Acquaviva, E., Moscoso, A., Peyre, H., Delorme, R., & Bucci, M. P. (2022). Reading performance in children with ADHD: An eye-tracking study. Annals of Dyslexia, 72(3), 552-565.
[b14] Benedek, M., Schickel, R. J., Jauk, E., Fink, A., & Neubauer, A. C. (2014). Alpha power increases in right parietal cortex reflects focused internal attention. Neuropsychologia, 56, 393-400.
[s19] van Son, D., De Blasio, F. M., Fogarty, J. S., Angelidis, A., Barry, R. J., & Putman, P. (2019). Frontal EEG theta/beta ratio during mind wandering episodes. Biological psychology, 140, 19-27.
[m22] Michelini, G., Salmastyan, G., Vera, J. D., & Lenartowicz, A. (2022). Event-related brain oscillations in attention-deficit/hyperactivity disorder (ADHD): A systematic review and meta-analysis. International journal of psychophysiology, 174, 29-42.
[km22] Kumar, N., & Michmizos, K. P. (2022). A neurophysiologically interpretable deep neural network predicts complex movement components from brain activity. Scientific reports, 12(1), 1101.
[w24] Wang, L., Zhang, X., Su, H., & Zhu, J. (2024). A comprehensive survey of continual learning: Theory, method and application. IEEE Transactions on Pattern Analysis and Machine Intelligence.