Previously submitted to: JMIR Mental Health (no longer under consideration since Mar 06, 2020)
Date Submitted: Feb 15, 2020
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Validity of diagnosis support for attention deficit/hyperactivity disorders - Model discrimination of machine learning approach
ABSTRACT
Background:
Attention deficit hyperactivity disorder (ADHD) is a common neurobehavioral disorder characterized by inattention, hyperactivity, and impulsivity. It is a chronic disorder and often persists into adulthood. Long-term follow-up studies showed that children with ADHD were more impaired in psychosocial, educational, and neuropsychological functioning, and had higher risks for antisocial disorders, major depression, and anxiety disorders as adults. Proper and early diagnosis would save medical resources and assist policy making for ADHD.
Objective:
In this research, we developed a diagnosis decision model using machine learning approach to effectively screen the disorder potentials. The model is based on three machine learning algorithms: logistic regression, classification and regression tree (CART), and neural network. They were compared for analysis of the disorder diagnosis with receiver operating characteristic curve.
Methods:
There were 74 participants in the ADHD group, while 21 participants in non-ADHD control group. The performance of three algorithms is evaluated by receiver operating characteristic (ROC) curve.
Results:
The results showed that the CART outperformed the other two, and the region values of receiver operating characteristic were in the following order: CART (0.848) > logistic regression model (0.826) > neural network (0.67). The sensitivity and specificity of the CART were 88% and 50%, respectively.
Conclusions:
In the future, this model can also be used in other neuroscience fields, such as diagnosing Asperger Syndrome, Tourette Syndrome, and Dementia. Thereby, it can exert practical value and benefits of the research results.
Citation
