RBMO Study Led by Belén Lledó Compared General AI Chatbots on IVF/PGT Questions – Fertiligent AI
Fertiligent AI shared a post on LinkedIn about a paper by Belén Lledo et al. published in RBMO:
“A new RBMO study compared general AI chatbots on IVF/PGT questions (ChatGPT-3.5, Gemini 1.5, and Llama-2). Useful—but the takeaway isn’t ‘AI is ready out-of-the-box.’ It’s that generalist LLMs struggle on domain nuance—and why a fertility-trained LLM is the right path forward.
Why a Fertility-trained LLM outperforms generalists:
- Guideline-true by design. Tuned to ASRM/ESHRE guidance and updated clinic SOPs; answers reflect current practice, not internet averages. (RBMO found quality varied with complexity.) (ScienceDirect)
- Lower hallucinations, higher citations. Retrieval from curated fertility corpora (patient leaflets, consent forms, lab standards) with inline sources for clinical review.
- Context that actually matters. Embryology, PGT-A/PGT-M nuance, ovarian reserve, AMA counseling, medication protocols—handled with structured reasoning, not generic medical talk.
- Safety rails and scope control. Built-in disclaimers, triage triggers, and handoff to clinicians when questions cross into diagnosis.
- Clinic-grade privacy. Architected for PHI handling and audit trails; responses are logged and reviewable.
- Localization for real patients. Multilingual outputs, readability tuning, and culturally sensitive counseling for high-anxiety moments.
- Continuous updates. Domain models are retrained/augmented as guidance changes—avoiding the ‘snapshot’ problem of frozen general models.
Note: the RBMO study used older baselines (ChatGPT-3.5 and Gemini 1.5). Newer general models have improved—but specialization still wins when accuracy, guardrails, and citations are non-negotiable in patient education and clinic workflows.
At Fertiligent AI, we’re building exactly this: a fertility-trained LLM with verified sources, clinic-first safety, and seamless clinician handoff—so teams can scale guidance without compromising care.
Sources: RBMO comparative evaluation of chatbots for PGT (uses ChatGPT-3.5, Gemini 1.5, Llama-2); RBMO study on reliability of AI counseling for fertility patients.”
Title: Assessing the performance of generative AI chatbots in preimplantation genetic testing: a comparative study of expert evaluations
Authors: Belén Lledo, Paola Carbone, Jose A. Ortiz, Ruth Morales, Adoración Rodríguez-Arnedo, Leyre Herrero, Elisa Alvarez, Jorge Ten, Lydia Luque, Juan C. Castillo, Jordi Suñol, Annalisa Racca, Andrea Bernabeu
-
May 6, 2026, 01:35Alison Wright: BSGE 2026 Highlights Leadership Change and Future Vision
-
May 6, 2026, 01:30Short Menstrual Cycles May Signal Early Ovarian Ageing – Fertility Plus
-
May 6, 2026, 01:27Asma Khalil: Maternal Mental Health Remains a Leading Risk
-
May 5, 2026, 16:55Pregnancy Outcomes in a Recurrent Miscarriage Population After Uterine Surgeries – Fertility and Sterility
-
May 5, 2026, 16:43MAGENTA Oocyte Scores, Morphokinetics and Live Birth – Future Fertility
-
May 5, 2026, 16:34Join Us on 16 September to Explore Challenges in Managing a Successful Fertility Service – BFS
-
May 5, 2026, 16:17Mature Oocyte Retrieval Without an Exogenous Trigger During Stimulation – Fertility and Sterility
-
May 5, 2026, 16:08We Are Delighted to Announce that Fertility 2027 Registration Opens soon – Fertility Conferences
-
May 5, 2026, 15:59Lillian’s Story Highlights Urgency of Recognising Postpartum Preeclampsia – Preeclampsia Foundation