Sex differences in schizophrenia spectrum disorders: insights from the DiAPAson study using a data-driven approach

Abstract

Purpose: Schizophrenia Spectrum Disorders (SSD) display notable sex differences: males have an earlier onset and more severe negative symptoms, while females exhibit affective symptoms, better verbal abilities, and a more favourable prognosis. Despite extensive research, areas such as time perception and positivity remain underexplored, and machine learning has not yet been adequately utilised. This study aims to address these gaps by examining sex differences in a sample of Italian patients with SSD using a data-driven approach.

Methods: As part of the DiAPAson project, 619 Italian patients with SSD (198 females; 421 males) were assessed using standardised clinical tools. Data on socio-demographics, clinical characteristics, symptom severity, functioning, positivity, quality of life (QoL), and time perspective were collected. Descriptive and regression analyses were conducted. Principal Component Analysis (PCA) and the Gaussian Mixture Model (GMM) was used to define data-driven clusters. A leave-one-group-out validation was performed.

Results: Males were more likely to be single (p < 0.001) and less educated (p = 0.006), while females smoked more tobacco (p = 0.011). Males were more frequently prescribed antipsychotics (p = 0.022) and exhibited more severe psychiatric (p = 0.004) and negative symptoms (p = 0.013). They also had a less negative perception of past events (p = 0.047) and a better view of their psychological condition (p = 0.004). Females showed better interpersonal functioning (p = 0.008). PCA and GMM revealed two main clusters with significant sex differences (p = 0.027).

Conclusion: This study identifies sex differences in SSD, suggesting tailored treatments such as enhancing interpersonal skills for females and maintaining positive self-assessment for males. Using machine learning, we highlight distinct SSD phenotypes, emphasising the need for sex-specific interventions to improve outcomes and QoL. Our findings stress the importance of a multifaceted, interdisciplinary approach to address sex-based disparities in SSD.

Trial registration: ISRCTN registry ID ISRCTN21141466.

Keywords: Machine Learning; Positivity; Schizophrenia spectrum disorder; Sex differences; Time perception.

Fig. 1

Parallel coordinate plots with confidence intervals. median values and 95% confidence intervals for normalized clinical scores in the two clusters are shown. the cluster assignment was performed in the logo-cv scenario. panel a shows the training features of the pca-gmm, panel b shows the features not used in the training that differed significantly (p < 0.05) between the clusters

Fig. 2

Intra- and inter-sex differences between two clusters. the figure shows the statistical differences between each permutation of the two clusters and sexes. the p-values were calculated with ancova using age as a covariate and bonferroni correction. acronyms: c0: cluster 0, c1: cluster 1, m: men, w: women