Creativity is notoriously difficult to define and measure as it is a multi-dimensional construct that encompasses aspects of cognitive processes, personality traits, and environmental factors (Said-Metwaly et al., 2017). In other words, creativity can be described as having novel ideas or alternatives to problems, exhibiting cognitive flexibility, and possessing the ability to synthesize and organize information. Scholars from the fields of economics, arts, neuroscience, medicine, and psychology have studied what creativity is and how to cultivate and apply it to daily life. Several individual and societal factors can influence creativity, including intrinsic motivation, cultural environment, and socioeconomic status. Although creativity is not solely associated with mental disorders, one of the most fervent discussions about it has focused on its association with psychopathology. Artistic and literary professionals, who are traditionally considered highly creative, have often been associated with psychopathology; this adds to the significance of creativity as a clinically relevant trait (Kyaga, 2014).

The association between creativity and psychopathology has been prompted by anecdotal reports of artistic individuals exhibiting psychiatric symptoms (Ourtani, 2021). Reports of creative geniuses, such as Wolfgang Amadeus Mozart and Vincent van Gogh, exhibiting symptoms that seemingly align with the diagnostic criteria for psychiatric illnesses have garnered attention. Previous studies have used various methods to evaluate this link (Kyaga, 2014; Ourtani, 2021; Power et al., 2015; Rajagopal et al., 2023). Nevertheless, there is a paucity of literature elucidating the genomic basis of creativity and its genetic relationship with psychiatric disorders. Although previous studies have shown that the heritability of creativity is moderate to high (Kyaga, 2014), its genetic architecture has not yet been elucidated. Li et al. (2020) examined the genetic architecture of creativity through a genome-wide association study (GWAS) but did not find any significant loci or polygenic overlap with psychiatric disorders due to a small sample size. Although the polygenic risk score (PRS) for schizophrenia (SCZ) and bipolar disorder (BD) has been shown to predict creativity (Power et al., 2015), an exploration of the broad landscape of associations between creativity and various psychiatric disorders is necessary.

Among the various methods used to define creativity (Said-Metwaly et al., 2017), occupation has often been utilized (Kyaga, 2014), particularly in a binary manner, by considering those with artistic and scientific occupations as creative and others as non-creative. Creativity has continuous properties as a polygenic trait; therefore, utilizing a binary system (i.e., artistic or scientific professions as creative and others as non-creative) helps simplify its characteristics but reduces the statistical power of analysis. In this study, we integrated a machine learning (ML)-based phenotyping of creativity to define occupational creativity (OC) as a continuous trait (Bakhshi et al., 2015). By conducting an ML-based GWAS of OC using the UK Biobank (UKB), we aimed to (1) identify OC-associated genetic variants and genomic basis via GWAS and post-GWAS analyses; (2) investigate the relationships between OC and psychiatric disorders using linkage disequilibrium score regression (LDSC), PRS analyses, and bivariate causal mixture modelling (MiXeR); (3) explore the shared genetic basis of OC and psychiatric disorders using the conditional and conjunctional false discovery rate (cond/conjFDR) approach; and (4) validate our main GWAS, in comparison to the GWASs using or excluding traditional definitions of creative occupations and controlling for the genetic effects of education years.