J Psychiatr Res. 2026 Mar 26;198:203-209. doi: 10.1016/j.jpsychires.2026.03.043. Online ahead of print.
ABSTRACT
BACKGROUND: Identifying baseline clinical characteristics associated with treatment response is crucial for optimising intervention strategies in people with autism spectrum disorder (ASD). This study aimed to distinguish responders from non-responders to a multisession prefrontal transcranial direct current stimulation (tDCS) protocol and to explore how baseline individual characteristics relate to treatment outcomes.
METHODS: Using complementary inferential statistics and predictive modelling approaches, we analysed baseline clinical characteristics of adolescents and young adults with ASD who underwent cathodal tDCS (1.5 mA) over the left dorsolateral prefrontal cortex, coupled with cognitive remediation training. Baseline symptom severity was assessed using the Autism Diagnostic Interview-Revised (ADI-R). Normalised feature importance and Shapley additive explanations (SHAP) were used to identify features associated with restricted and repetitive behaviour (RRB) reduction, and a model-derived classification threshold was estimated to stratify potential responders and non-responders within this protocol.
RESULTS: Inferential statistics revealed that responders had significantly lower social interaction symptom severity than non-responders (p < .001; Cohen’s d = 0.41). Predictive modelling further indicated that lower baseline verbal communication severity was associated with greater RRB reduction (averaged SHAP value across models = -0.017). A model-derived classification threshold of 19.6 on the ADI-R verbal communication sub-score was identified to differentiate responders and non-responders within this protocol.
CONCLUSION: Lower baseline verbal communication severity is associated with greater behavioural improvement following this multisession prefrontal tDCS protocol. These findings should be interpreted as exploratory and hypothesis-generating. Replication in larger, independent cohorts is required to assess the robustness and generalisability of these findings.
TRIAL REGISTRATION: ClinicalTrials.gov (ID: NCT05035511).
PMID:41921246 | DOI:10.1016/j.jpsychires.2026.03.043
