| CPC G06N 5/04 (2013.01) [G06N 3/044 (2023.01); G06N 20/00 (2019.01); G06N 20/10 (2019.01); G16H 50/20 (2018.01)] | 20 Claims |
|
1. A data processing system for processing a feature vector that comprises one or more features that are indicative of dyslexic behavior, the data processing system comprising:
at least one processor; and
a memory storing instructions that, when executed by the at least one processor, causes the at least one processor to perform operations comprising:
receiving, through a camera, data representing measurements of eye movements of a user of a client device as the user reads text rendered on a graphical user interface of the client device;
receiving, through the graphical user interface of the client device, data representing one or more interactions with the graphical user interface of the client device;
generating, based on the data representing one or more interactions with the graphical user interface of the client device and the data representing the measurements of the eye movements, the feature vector, wherein a feature of the feature vector represents one or more measurements of the eye movements of the user of the client device or other data indicative of dyslexic behavior;
determining, using machine learning logic, a classification metric for each feature of the feature vector, the machine learning logic being trained by operations comprising:
selecting a set of features representing performance measurements for a set of users interacting with user interfaces;
receiving, for the set of users, label data specifying whether the user is dyslexic;
training the machine learning logic using the set of features and the label data; and
iteratively removing one or more features from the set of features until an accuracy of the machine learning logic satisfies a threshold accuracy to generate a final set of features for including the feature vector; and
generating a prediction value indicative of a predicted likelihood of dyslexia for the user by performing operations comprising:
assigning to each feature of the feature vector, based on the classification metric of the respective feature, a prediction weight;
determining the prediction value based on that prediction weight for each feature of the feature vector; and
outputting, for rendering on a display, a representation of the prediction value.
|