Machine learning models for the prediction of acuity and variability of eye-positioning using features extracted from oculography

During the first months of life, babies can be affected by congenital nystagmus, an ocular-motor disease making visual acuity decrease. Electrooculography (EOG) and Infrared-oculography are utilized in order to perform eye-tracking of patients, giving the possibility to extract from the signals several useful features. In the past years, different algorithms were used to perform the detection of events on these features and many researchers studied the relationships between the features and physiological values such as visual acuity and variability of eye-positioning. In this paper, machine learning techniques were used to predict visual acuity and the variability of eye positioning using features extracted from EOG. The EOG of 20 patients was acquired, signals underwent a pre-processing, and some parameters were extracted through a custom-made software. Frequency, amplitude, intensity, nystagmus foveation periods and both amplitude and frequency of baseline oscillation were the features used as input for the algorithms. Knime analytics platform was employed to perform a predictive analysis using Random Forests, Logistic Regression Tree, Gradient boosted tree, K nearest neighbour, Multilayer Perceptron and Support Vector Machine. Finally, some evaluation metrics were computed employing a leave one out cross validation. Considering the coefficient of determination, visual acuity achieved values between 0.67 and 0.85 while variability of eye positioning ranged from 0.62 to 0.79. These results were compared with past analysis with the exact same aims and dataset, obtaining a greater value as regards the variability of eye positioning and comparable results exploiting all the features related to nystagmus as regards the visual acuity. This paper showed the feasibility of a regression analysis performed through machine learning algorithms in detecting relationships among variables related to congenital nystagmus.

[1]  Arthi S V and Suresh R. Norman.  Analysis of Electrooculography signals for the Interface and Control of Appliances , 2015 .

[2]  Joseph H. Goldberg,et al.  Identifying fixations and saccades in eye-tracking protocols , 2000, ETRA.

[3]  Luigi Iuppariello,et al.  Application of data mining in a cohort of Italian subjects undergoing myocardial perfusion imaging at an academic medical center , 2020, Comput. Methods Programs Biomed..

[4]  M. Dunn.  Visual perception in infantile nystagmus , 2016 .

[5]  Arturo Brunetti,et al.  Characterization of Adrenal Lesions on Unenhanced MRI Using Texture Analysis: A Machine‐Learning Approach , 2018, Journal of magnetic resonance imaging : JMRI.

[6]  Luigi Iuppariello,et al.  Using gait analysis' parameters to classify Parkinsonism: A data mining approach , 2019, Comput. Methods Programs Biomed..

[7]  Improta Giovanni,et al.  Distinguishing Functional from Non-functional Pituitary Macroadenomas with a Machine Learning Analysis , 2019, IFMBE Proceedings.

[8]  Cynthia Rudin,et al.  Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead , 2018, Nature Machine Intelligence.

[9]  Ivo D. Dinov,et al.  Methodological challenges and analytic opportunities for modeling and interpreting Big Healthcare Data , 2016, GigaScience.

[10]  M. Juhola,et al.  Detection of nystagmus eye movements using a recursive digital filter , 1988, IEEE Transactions on Biomedical Engineering.

[11]  Richard V Abadi,et al.  The influence of nystagmoid oscillation on contrast sensitivity in normal observers , 1985, Vision Research.

[12]  Hamed Asadi,et al.  Peering Into the Black Box of Artificial Intelligence: Evaluation Metrics of Machine Learning Methods. , 2019, AJR. American journal of roentgenology.

[13]  Maria Romano,et al.  Efficacy of Machine Learning in Predicting the Kind of Delivery by Cardiotocography , 2019, IFMBE Proceedings.

[14]  M. Cesarelli,et al.  Is It Possible to Predict Cardiac Death? , 2019, IFMBE Proceedings.

[15]  R. B. Daroff,et al.  Congenital nystagmus waveforms and foveation strategy , 1975, Documenta Ophthalmologica.

[16]  M. Bracale,et al.  Eye movement baseline oscillation and variability of eye position during foveation in congenital nystagmus , 2003, Documenta Ophthalmologica.

[17]  Johan A. K. Suykens,et al.  Least Squares Support Vector Machine Classifiers , 1999, Neural Processing Letters.

[18]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[19]  Krzysztof Jaskot,et al.  Real-Time Detection and Filtering of Eye Blink Related Artifacts for Brain-Computer Interface Applications , 2015, ICMMI.

[20]  Yunhao Liu,et al.  Big Data: A Survey , 2014, Mob. Networks Appl..

[21]  Oleg V. Komogortsev,et al.  Using machine learning to detect events in eye-tracking data , 2018, Behavior research methods.

[22]  Claudia Angelini,et al.  A novel shiny platform for the geo-spatial analysis of large amount of patient data , 2017, PeerJ Prepr..

[23]  V. Gudnason,et al.  Assessing cardiovascular risks from a mid-thigh CT image: a tree-based machine learning approach using radiodensitometric distributions , 2020, Scientific Reports.

[24]  Raimondas Zemblys,et al.  Eye-movement event detection meets machine learning , 2017, BioMed 2017.

[25]  M. Cesarelli,et al.  Linear discriminant analysis and principal component analysis to predict coronary artery disease , 2020, Health Informatics J..

[26]  A. Weiss,et al.  Does eye velocity due to infantile nystagmus deprive visual acuity development? , 2017, Journal of AAPOS : the official publication of the American Association for Pediatric Ophthalmology and Strabismus.

[27]  Yodchanan Wongsawat,et al.  Hybrid EEG-EOG brain-computer interface system for practical machine control , 2010, 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology.

[28]  Kenneth Holmqvist,et al.  Eye tracking: a comprehensive guide to methods and measures , 2011 .

[29]  M. Bracale,et al.  Relationship between visual acuity and eye position variability during foveations in congenital nystagmus , 2000, Documenta Ophthalmologica.

[30]  Mario Cortese,et al.  Enhancing brain-machine interface (BMI) control of a hand exoskeleton using electrooculography (EOG) , 2014, Journal of NeuroEngineering and Rehabilitation.

[31]  Luigi Iuppariello,et al.  Classifying Different Stages of Parkinson’s Disease Through Random Forests , 2019, IFMBE Proceedings.

[32]  Liberatina Carmela Santillo,et al.  Use of the AHP methodology in system dynamics: Modelling and simulation for health technology assessments to determine the correct prosthesis choice for hernia diseases. , 2018, Mathematical biosciences.

[33]  C. Harris,et al.  Arguments for the Adoption of a Nystagmus Care Pathway , 2019, The British and Irish orthoptic journal.

[34]  Shiro Usui,et al.  Digital Low-Pass Differentiation for Biological Signal Processing , 1982, IEEE Transactions on Biomedical Engineering.

[35]  J. Erichsen,et al.  The Effect of Gaze Angle on Visual Acuity in Infantile Nystagmus. , 2017, Investigative ophthalmology & visual science.

[36]  R. Ward,et al.  EMG and EOG artifacts in brain computer interface systems: A survey , 2007, Clinical Neurophysiology.

[37]  Hari Singh Dhillon,et al.  Human Eye Tracking and Related Issues: A Review , 2012 .

[38]  C. L. Doren,et al.  The effects of afferent stimulation on congenital nystagmus foveation periods , 1995, Vision Research.

[39]  Davide Castelvecchi,et al.  Can we open the black box of AI? , 2016, Nature.

[40]  S. Brunak,et al.  Mining electronic health records: towards better research applications and clinical care , 2012, Nature Reviews Genetics.

[41]  Krzysztof Jaskot,et al.  Nystagmus Detection System , 2018 .