To explore neurocritical care (NCC) providers’ perceptions of potential integration of artificial intelligence (AI)-based prognostic tools in withdrawal of life-sustaining treatment (WLST) and goals-of-care (GOC) decisions.

Discussion about AI-based prognostic tools is emerging in NCC. These tools hold potential benefits, including standardizing decision-making in complex scenarios; however, they also raise ethical concerns regarding transparency, bias, and autonomy distress. Understanding providers’ views is essential for the responsible integration of these tools. This study was conducted as part of the ºÃÉ«ÏÈÉú Neurologist-in-Training Clinical Ethics Elective.

A 28-item mixed-methods survey was distributed electronically to NCC providers in September 2025. The survey collected demographics, experience with AI-based prognostic tools, ethical concerns, and perceptions toward their use in WLST/GOC. Quantitative data were summarized using descriptive statistics. Qualitative data were analyzed using Braun and Clarke’s inductive approach, along with reflexivity and cross-checking. Data saturation was achieved after approximately 30 responses.

Thirty-eight providers participated (attendings: 39%; residents: 39%; fellows: 10%; advanced practice: 8%; 82% neurology). Overall views were neutral (45%) to positive (39%). Most respondents (65%) believe AI tools can be helpful in WLST/GOC discussions. Reported benefits included reduced variability (58%), more accurate prognosis (45%), reduced human bias (42%), support for less experienced clinicians (39%), and faster decisions (32%). Main concerns involved over-reliance on AI (87%), errors with existing data/publications (76%), algorithmic bias (61%), family mistrust (58%), and lack of transparency (53%). Thematic analysis of open-ended questions revealed that communication was the dominant theme, highlighting providers’ concern about how AI-generated prognostic information could affect conversations with families regarding WLST/GOC. Respondents also emphasized that AI should remain a supportive tool rather than a determinant of WLST/GOC decisions.

NCC providers identified both potential benefits and ethical concerns regarding AI in WLST/GOC decision-making. Responsible integration requires communication support, preservation of provider autonomy, and transparency.