What is the potential for the introduction of an artificial intelligence (AI) chatbot to support medical abortion at home? Findings from a patient survey in England and Wales.

BACKGROUND

Most patients in Britain undergoing medical abortion under 10 weeks' gestation manage the entire process at home, with access to clinical support from qualified staff via telephone if needed. Artificial intelligence (AI) 'chatbots' are potential tools to support patients; however, little is known about patient views on their use in this regard.

METHODS

We invited British Pregnancy Advisory Service (BPAS) patients who had a medical abortion under 10 weeks' gestation to participate in an online, English-language, cross-sectional questionnaire (August 2025-October 2025). The questionnaire assessed demographics, self-reported likelihood of using a potential AI chatbot, perceived appealing features, reasons for hesitation and preferred features. Descriptive analysis, Chi-square statistics and p values were used to analyse responses. The desired sample size was 380.

RESULTS

Of the 4684 patients invited, 396 completed the questionnaire (8%). Overall, 22.4% self-reported as very likely to use an AI chatbot, 13.1% as likely, 20.8% as neutral, 17.5% as unlikely and 26.2% as very unlikely. Self-reported likelihood of use did not vary by demographics. Perceived appealing features included: 24-hour access (57.6%), reduced wait time (47%), anonymous care (25.7%) and access to reliable personalised information (17.2%). Reasons for hesitation included: misinformation (54.2%), preference for human interaction (51.9%), conversations that loop (42.2%), data privacy (40.4%) and lack of trust in AI (38.3%). Preferred features emphasised flexibility, customisation and ability to contact staff.

CONCLUSIONS

Respondents' views on a potential AI chatbot for medical abortion support were mixed, shaped by trust, reliability, flexibility and access to clinical staff. Such tools may be most acceptable as optional adjuncts within existing care pathways, requiring real-world evaluation.