Functional imaging derived ADHD biotypes based on deep clustering: a study on personalized medication therapy guidance

Aichen Feng; Dongmei Zhi; Yuan Feng; Rongtao Jiang; Zening Fu; Ming Xu; Min Zhao; Shan Yu; Michael Stevens; Li Sun; Vince Calhoun; Jing Sui

doi:10.1016/j.eclinm.2024.102876

Functional imaging derived ADHD biotypes based on deep clustering: a study on personalized medication therapy guidance

EClinicalMedicine. 2024 Oct 10:77:102876. doi: 10.1016/j.eclinm.2024.102876. eCollection 2024 Nov.

Authors

Aichen Feng^{1

2}, Dongmei Zhi³, Yuan Feng⁴, Rongtao Jiang⁵, Zening Fu⁶, Ming Xu^{1

2}, Min Zhao^{1

2}, Shan Yu^{1

2}, Michael Stevens^{7

8}, Li Sun⁴, Vince Calhoun⁶, Jing Sui³

Affiliations

¹ Brainnetome Center and National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing, 100190, China.
² School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, 100049, China.
³ State Key Laboratory of Cognitive Neuroscience and Learning, Beijing Normal University, Beijing, 100875, China.
⁴ Peking University Sixth Hospital/Institute of Mental Health, National Clinical Research Center for Mental Disorders (Peking University Sixth Hospital), Beijing, 100191, China.
⁵ Department of Radiology and Biomedical Imaging, Yale University, New Haven, CT, USA.
⁶ Tri-institutional Center for Translational Research in Neuroimaging and Data Science (TReNDS), Georgia Institute of Technology, Emory University and Georgia State University, Atlanta, GA, 30303, United States.
⁷ Department of Psychiatry, Olin Neuropsychiatry Research Center, Institute of Living, Hartford Healthcare Corporation, Hartford, CT, United States.
⁸ Department of Psychiatry, Yale University School of Medicine, New Haven, CT, United States.

Abstract

Background: Attention deficit hyperactivity disorder (ADHD) is one prevalent neurodevelopmental disorder with childhood onset, however, there is no clear correspondence established between clinical ADHD subtypes and primary medications. Identifying objective and reliable neuroimaging markers for categorizing ADHD biotypes may lead to more individualized, biotype-guided treatment.

Methods: Here we proposed a graph convolution network for biological subtype detection (GCN-BSD) using both functional network connectivity (FNC) and non-imaging phenotypic data for ADHD biotype. We applied GCN-BSD to ADHD patients from the ABCD study as the discovery dataset and a validation ADHD dataset with longitudinal medication treatment from Peking University Sixth Hospital.

Findings: We identified two biotypes based on 1069 ADHD patients selected from Adolescent Brain and Cognitive Development (ABCD) study, which were validated on independent ADHD adolescents undergoing longitudinal medication treatment (n = 130). Interestingly, in addition to differences in cognitive performance and hyperactivity/impulsivity symptoms, biotype 1 demonstrated a significantly better recovery rate in psychosomatic problems score (p < 0.05, baseline symptom score adjusted) when treated with methylphenidate than with atomoxetine.

Interpretation: Our results suggested that such an imaging-driven, biotype-guided approaches hold promise for facilitating personalized treatment of ADHD and exploring possible boundaries through innovative deep learning algorithms to improve medication treatment effectiveness.

Funding: Science and Technology Innovation 2030 Major Projects, the National Natural Science Foundation of China, the Startup Funds for Talents at Beijing Normal University, China Postdoctoral Science Foundation, and the National Institutes of Health.

Keywords: Adolescent brain and cognitive development (ABCD) study; Attention deficit hyperactivity disorder (ADHD); Biological subtype detection; Deep clustering; Graph convolutional network (GCN).