Abstract Background and aims Intracerebral haemorrhage (ICH) is associated with high mortality and long-term disability. Evidence for surgical benefit remains inconsistent across ICH subtypes. Advances in artificial intelligence (AI) and minimally invasive approaches promise greater precision in haematoma evacuation, but randomised evidence remains limited. Methods To evaluate efficacy and safety of AI-assisted robotic-guided minimally invasive neurosurgery versus conventional care across major ICH subtypes within a single master protocol. Results This is a multicentre, randomised, controlled umbrella trial with four parallel substudies targeting large basal ganglia haemorrhage, moderate basal ganglia haemorrhage, intraventricular haemorrhage, and brainstem haemorrhage. Within each substudy, participants are allocated 1:1 using stratified block randomization to AI-assisted robotic-guided intervention (substudy-specific endoscopic evacuation or stereotactic puncture) or the relevant conventional control (craniotomy, external ventricular drainage, or standard medical management). Outcome assessment is blinded. The planned sample size is approximately 1000 participants. Follow-ups are up to 6 months. Primary analyses will be conducted on an intention-to-treat basis. Conclusions In the master protocol, the primary outcome is the utility-weighted modified Rankin Scale (UW-mRS) at 6 months; key secondary outcomes include ordinal shift in modified Rankin Scale (mRS), excellent outcome (mRS 0–1), functional independence (mRS 0–2), health-related quality of life (EuroQol five-dimension five-level questionnaire; EQ-5D-5L), length of hospital stay, and mortality at 6 months. Substudy-specific outcomes are defined where applicable. Conflict of interest
Yu et al. (2026) studied this question.