A Robust Machine Learning Based Framework for the Automated Detection of ADHD Using Pupillometric Biomarkers and Time Series Analysis

William Das,Shubh Khanna

doi:10.1038/s41598-021-95673-5

Abstract

Accurate and efficient detection of attention-deficit/hyperactivity disorder (ADHD) is critical to ensure proper treatment for affected individuals. Current clinical examinations, however, are inefficient and prone to misdiagnosis, as they rely on qualitative observations of perceived behavior. We propose a robust machine learning based framework that analyzes pupil-size dynamics as an objective biomarker for the automated detection of ADHD. Our framework integrates a comprehensive pupillometric feature engineering and visualization pipeline with state-of-the-art binary classification algorithms and univariate feature selection. The support vector machine classifier achieved an average 85.6% area under the receiver operating characteristic (AUROC), 77.3% sensitivity, and 75.3% specificity using ten-fold nested cross-validation (CV) on a declassified dataset of 50 patients. 218 of the 783 engineered features, including fourier transform metrics, absolute energy, consecutive quantile changes, approximate entropy, aggregated linear trends, as well as pupil-size dilation velocity, were found to be statistically significant differentiators (p < 0.05), and provide novel behavioral insights into associations between pupil-size dynamics and the presence of ADHD. Despite a limited sample size, the strong AUROC values highlight the robustness of the binary classifiers in detecting ADHD—as such, with additional data, sensitivity and specificity metrics can be substantially augmented. This study is the first to apply machine learning based methods for the detection of ADHD using solely pupillometrics, and highlights its strength as a potential discriminative biomarker, paving the path for the development of novel diagnostic applications to aid in the detection of ADHD using oculometric paradigms and machine learning.

Highlights

Related to inattention and nine related to hyperactivity and impulsivity[11]
Wainstein et al showed in an experimental study that pupil-size dynamics were strong differentiators between attention-deficit/hyperactivity disorder (ADHD) positive and negative subjects after performing statistical analysis; pupil-size was shown to be strongly correlated with attentional performance in subjects[9]
State-of-the-art machine learning algorithms were evaluated based on three key medical diagnostic classification metrics: sensitivity, specificity, and area under the receiver operating characteristic (AUROC)

Summary

Introduction

Related to inattention and nine related to hyperactivity and impulsivity[11]. These subjective clinical assessments often last multiple hours[6]. Wainstein et al showed in an experimental study that pupil-size dynamics were strong differentiators between ADHD positive and negative subjects after performing statistical analysis; pupil-size was shown to be strongly correlated with attentional performance in subjects[9]. Given the vast literature highlighting associations between pupil-size dynamics and attentional performance in neurobehavioral disorders, we hypothesized that pupillometric features could be utilized as objective biomarkers to effectively characterize ADHD using a machine learning based and time series analysis methodology. We sought to develop a machine learning-based framework to analyze pupillometric variation in subjects, hypothesizing that it would accurately discriminate between ADHD positive patients and control subjects. More specific experiment details are outlined extensively in Wainstein et al.’s p ublication[25]

Methods

Results

Conclusion