Objective ADHD diagnosis using convolutional neural networks over daily-life activity records

Patricia Amado-Caballero, Pablo Casaseca-de-la-Higuera*, Susana Alberola-Lopez, Jesus Maria Andres-de-Llano, Jose Antonio Lopez-Villalobos, Jose Ramon Garmendia-Leiza, Carlos Alberola-Lopez

*Corresponding author for this work

Research output: Contribution to journalArticle

17 Downloads (Pure)

Abstract

Attention Deficit/Hyperactivity Disorder (ADHD) is the most common neurobehavioral disorder in children and adolescents. However, its etiology is still unknown, and this hinders the existence of reliable, fast and inexpensive standard diagnostic methods. Objective: This paper proposes an end-to-end methodology for automatic diagnosis of the combined type of ADHD. Methods: Diagnosis is based on the analysis of 24 hour-long activity records using Convolutional Neural Networks to classify spectrograms of activity windows. Results: We achieve up to 97.62% average sensitivity, 99.52% specificity and AUC values over 99%. Overall, our figures overcome those obtained by actigraphy-based methods reported in the literature as well as others based on more expensive (and not so convenient) acquisition methods. Conclusion: These results reinforce the idea that combining deep learning techniques together with actimetry can lead to a robust and efficient system for objective ADHD diagnosis. Significance: Reliance on simple activity measurements leads to an inexpensive and non-invasive objective diagnostic method, which can be easily implemented with daily devices.
Original languageEnglish
JournalIEEE Journal of Biomedical and Health Informatics
Early online date6 Jan 2020
DOIs
Publication statusE-pub ahead of print - 6 Jan 2020

Fingerprint

Attention Deficit Disorder with Hyperactivity
Neural networks
Actigraphy
Area Under Curve
Learning
Sensitivity and Specificity
Equipment and Supplies

Keywords

  • ADHD
  • Actigraphy
  • Deep learning
  • Concolutional neural network (CNN)

Cite this

Amado-Caballero, P., Casaseca-de-la-Higuera, P., Alberola-Lopez, S., Andres-de-Llano, J. M., Lopez-Villalobos, J. A., Garmendia-Leiza, J. R., & Alberola-Lopez, C. (2020). Objective ADHD diagnosis using convolutional neural networks over daily-life activity records. IEEE Journal of Biomedical and Health Informatics. https://doi.org/10.1109/JBHI.2020.2964072
Amado-Caballero, Patricia ; Casaseca-de-la-Higuera, Pablo ; Alberola-Lopez, Susana ; Andres-de-Llano, Jesus Maria ; Lopez-Villalobos, Jose Antonio ; Garmendia-Leiza, Jose Ramon ; Alberola-Lopez, Carlos. / Objective ADHD diagnosis using convolutional neural networks over daily-life activity records. In: IEEE Journal of Biomedical and Health Informatics. 2020.
@article{3b8433d5c082412181a44fdfa4231c0d,
title = "Objective ADHD diagnosis using convolutional neural networks over daily-life activity records",
abstract = "Attention Deficit/Hyperactivity Disorder (ADHD) is the most common neurobehavioral disorder in children and adolescents. However, its etiology is still unknown, and this hinders the existence of reliable, fast and inexpensive standard diagnostic methods. Objective: This paper proposes an end-to-end methodology for automatic diagnosis of the combined type of ADHD. Methods: Diagnosis is based on the analysis of 24 hour-long activity records using Convolutional Neural Networks to classify spectrograms of activity windows. Results: We achieve up to 97.62{\%} average sensitivity, 99.52{\%} specificity and AUC values over 99{\%}. Overall, our figures overcome those obtained by actigraphy-based methods reported in the literature as well as others based on more expensive (and not so convenient) acquisition methods. Conclusion: These results reinforce the idea that combining deep learning techniques together with actimetry can lead to a robust and efficient system for objective ADHD diagnosis. Significance: Reliance on simple activity measurements leads to an inexpensive and non-invasive objective diagnostic method, which can be easily implemented with daily devices.",
keywords = "ADHD, Actigraphy, Deep learning, Concolutional neural network (CNN)",
author = "Patricia Amado-Caballero and Pablo Casaseca-de-la-Higuera and Susana Alberola-Lopez and Andres-de-Llano, {Jesus Maria} and Lopez-Villalobos, {Jose Antonio} and Garmendia-Leiza, {Jose Ramon} and Carlos Alberola-Lopez",
year = "2020",
month = "1",
day = "6",
doi = "10.1109/JBHI.2020.2964072",
language = "English",
journal = "IEEE Journal of Biomedical and Health Informatics",
issn = "2168-2194",
publisher = "IEEE",

}

Objective ADHD diagnosis using convolutional neural networks over daily-life activity records. / Amado-Caballero, Patricia; Casaseca-de-la-Higuera, Pablo; Alberola-Lopez, Susana; Andres-de-Llano, Jesus Maria; Lopez-Villalobos, Jose Antonio; Garmendia-Leiza, Jose Ramon; Alberola-Lopez, Carlos.

In: IEEE Journal of Biomedical and Health Informatics, 06.01.2020.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Objective ADHD diagnosis using convolutional neural networks over daily-life activity records

AU - Amado-Caballero, Patricia

AU - Casaseca-de-la-Higuera, Pablo

AU - Alberola-Lopez, Susana

AU - Andres-de-Llano, Jesus Maria

AU - Lopez-Villalobos, Jose Antonio

AU - Garmendia-Leiza, Jose Ramon

AU - Alberola-Lopez, Carlos

PY - 2020/1/6

Y1 - 2020/1/6

N2 - Attention Deficit/Hyperactivity Disorder (ADHD) is the most common neurobehavioral disorder in children and adolescents. However, its etiology is still unknown, and this hinders the existence of reliable, fast and inexpensive standard diagnostic methods. Objective: This paper proposes an end-to-end methodology for automatic diagnosis of the combined type of ADHD. Methods: Diagnosis is based on the analysis of 24 hour-long activity records using Convolutional Neural Networks to classify spectrograms of activity windows. Results: We achieve up to 97.62% average sensitivity, 99.52% specificity and AUC values over 99%. Overall, our figures overcome those obtained by actigraphy-based methods reported in the literature as well as others based on more expensive (and not so convenient) acquisition methods. Conclusion: These results reinforce the idea that combining deep learning techniques together with actimetry can lead to a robust and efficient system for objective ADHD diagnosis. Significance: Reliance on simple activity measurements leads to an inexpensive and non-invasive objective diagnostic method, which can be easily implemented with daily devices.

AB - Attention Deficit/Hyperactivity Disorder (ADHD) is the most common neurobehavioral disorder in children and adolescents. However, its etiology is still unknown, and this hinders the existence of reliable, fast and inexpensive standard diagnostic methods. Objective: This paper proposes an end-to-end methodology for automatic diagnosis of the combined type of ADHD. Methods: Diagnosis is based on the analysis of 24 hour-long activity records using Convolutional Neural Networks to classify spectrograms of activity windows. Results: We achieve up to 97.62% average sensitivity, 99.52% specificity and AUC values over 99%. Overall, our figures overcome those obtained by actigraphy-based methods reported in the literature as well as others based on more expensive (and not so convenient) acquisition methods. Conclusion: These results reinforce the idea that combining deep learning techniques together with actimetry can lead to a robust and efficient system for objective ADHD diagnosis. Significance: Reliance on simple activity measurements leads to an inexpensive and non-invasive objective diagnostic method, which can be easily implemented with daily devices.

KW - ADHD

KW - Actigraphy

KW - Deep learning

KW - Concolutional neural network (CNN)

U2 - 10.1109/JBHI.2020.2964072

DO - 10.1109/JBHI.2020.2964072

M3 - Article

JO - IEEE Journal of Biomedical and Health Informatics

JF - IEEE Journal of Biomedical and Health Informatics

SN - 2168-2194

ER -