Content area
Full text
Backgrounds
Bipolar disorder (BD) is a complex chronic disease characterized by recurring episodes of depression and mania or hypomania [1]. The disorder is widespread, with an estimated lifetime prevalence of the bipolar spectrum ranging from 2.8 to 6.5% [2]. BD is the sixth leading cause of disability worldwide among young adults [3], and the risk of suicide in patients with this disorder is significant, estimated at around 15% [4, 5]. Numerous studies have highlighted BD’s substantial social and economic costs, particularly regarding the impact on patients’ productivity and caregivers’ well-being [6, 7–8].
The primary symptoms of BD are severe mood swings. Treatment often involves medications that stabilize mood and alleviate symptoms during both depressive and manic episodes. Among the drugs prescribed for this purpose is lithium. Lithium is considered the appropriate treatment for managing BD and has been used for over six decades. Lithium also reduces suicidal thoughts or intentions in bipolar patients [9, 10]. However, despite its widespread use, the precise biological mechanisms underlying BD and the response to lithium remain poorly understood, prompting ongoing research into the genetic factors involved[11].
In recent years, significant efforts have been made to gain more information about the biological responses associated with lithium treatment, focusing particularly on the effects of lithium on gene expression. A promising approach to studying these effects is the use of Bayesian network models, which estimate gene networks from microarray gene expression data. The concept of Bayesian networks was first introduced in 1988 by Judea Pearl. After more than 30 years, Bayesian networks have since become a powerful tool in various fields of science [12]. Considerable attention has been directed toward utilizing the Bayesian network model to estimate gene networks from microarray gene expression data. These networks are useful for analyzing gene expression data for several reasons:
The expression levels of genes are inherently random, and the data contains errors. Given the high cost of experiments, it is often impractical to repeat them. With their probabilistic and statistical foundation, Bayesian networks are well-suited for handling such uncertainty [13].
Gene expression is fundamentally a random phenomenon, for the randomness of the model is a core property. Bayesian networks are designed to accommodate this by incorporating error-prone data [14].
Bayesian networks use conditional probabilities to...