Centre for Smart Systems, AI and Cybersecurity (SSAICS)
This project aims to improve the diagnosis of bipolar disorder and borderline disorder by utilizing machine learning techniques and various cognitive and behavioural criteria, such as family history and smartphone mood ratings, to improve the accuracy of the classification algorithm.
Computational modelling for Accurate Differential Diagnosis of Bipolar and Borderline Personality Disorders
This project aims to improve the diagnosis of bipolar disorder (BP) and borderline disorder (BPD) by utilizing machine learning (ML) techniques. Distinguishing between these two disorders is a complex task due to their clinical overlap and similar symptoms. The current methods of diagnosis rely heavily on clinical interviews, which can be time-consuming and prone to subjective biases. Our proposed method for differentiating these two disorders uses electroencephalogram (EEG) signals and cognitive features obtained from the Wisconsin Card Sorting Test (WCST) and the Independent Components Analysis (ICA) test. Previous research has suggested that these two disorders are biologically indistinguishable from each other in electrophysiological data. However, family history has been identified as the most crucial feature in the classification of these two disorders, with 92-95% accuracy. Therefore, we will include family history as a feature in our ML algorithm. In addition, we will use other cognitive and behavioural criteria, such as emotion regulation strategies and parents' behaviours during childhood, to improve the accuracy of our algorithm. To further improve the accuracy of our algorithm, we will also use smartphone mood ratings. We will include this feature in our algorithm to determine if it improves the accuracy of our classification.
Our study has significant implications for the clinical diagnosis and treatment of these two disorders. Improving the accuracy of diagnosis can lead to better treatment outcomes and prevent misdiagnosis, which can result in inappropriate treatment and medication. Additionally, our study can contribute to a better understanding of the biological and cognitive underpinnings of these two disorders.
In conclusion, our project aims to improve the diagnosis of bipolar disorder and borderline disorder by utilizing machine learning techniques. We will use EEG signals, cognitive features, family history, and smartphone mood ratings to improve the accuracy of our classification algorithm. Our study has significant implications for the clinical diagnosis and treatment of these two disorders, as well as advancing our understanding of their underlying biology and cognition.