TL;DR
Get school and study supplies delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
In a guest report published by Astral Codex Ten, Drew Housman says he and his wife used ChatGPT for fertility questions after years of unsuccessful embryo transfers. He says the chatbot suggested an MRI, which revealed a 6-centimeter uterine fibroid; the account does not establish that the chatbot caused a pregnancy or that the same approach is appropriate for other patients.
Drew Housman says he and his wife, after years of unsuccessful embryo transfers and a diagnosis of unexplained infertility, asked ChatGPT fertility questions and followed up when it suggested an MRI. In a guest report for Astral Codex Ten, Housman says the scan found a previously undetected 6-centimeter uterine fibroid that their fertility doctor then recommended removing. The account is a personal report, not evidence that an AI chatbot can reliably diagnose infertility or replace medical care.
Housman writes that the couple could create embryos, but repeated transfers did not lead to a pregnancy. Over the course of about six years, they saw multiple doctors and underwent numerous scans, tests and procedures without finding an explanation. He describes the label “unexplained infertility” as a point at which the couple felt less attention was being paid to investigating their case.
The couple began using ChatGPT, initially GPT-4, to ask questions they did not have time to raise during short appointments. Housman says they asked about imaging, hormone levels, follicle counts, inflammation, procedures and other aspects of fertility treatment. He says answers they checked with their doctors were consistent with what those clinicians told them, while also recounting occasional errors, including the chatbot misreading an ultrasound as showing a pregnancy.
According to Housman, the chatbot—given the name “Dr. Reid”—raised the possibility of an MRI. The couple asked their fertility doctor about it; he writes that the doctor was initially reluctant, and the couple insisted before the scan was ordered. Housman says the MRI identified a 6-centimeter submucosal fibroid in the uterine wall, and that the doctor recommended removing it. The report does not provide medical records or independent confirmation of the sequence.
A Second Opinion in a Long Search
The report illustrates why some patients turn to chatbots: they can ask follow-up questions at any hour, revisit explanations and organize concerns between appointments. For people facing long, uncertain medical investigations, that access may help them formulate questions or seek clarification from clinicians. Housman says the tool gave the couple a persistent place to explore possibilities they felt had not received enough attention.
But the story also shows the limits of drawing broad conclusions from one outcome. The MRI and fibroid are reported by the patient, and the article does not show that the chatbot identified the growth with diagnostic certainty; it suggested an examination that a doctor ultimately ordered and interpreted. The case does not establish that an MRI is indicated for every patient with infertility, or that a chatbot can determine which test is appropriate. Clinical decisions require qualified medical judgment, including consideration of a patient’s history, risks and other findings.
For readers, the practical relevance is less that an AI system should direct care than that patients can use its suggestions as prompts to discuss with their own clinicians. Chatbot answers can be mistaken, incomplete or poorly calibrated, as Housman’s ultrasound example acknowledges. The account describes one couple’s experience, not a validated treatment pathway or a measured improvement in fertility outcomes.
As an affiliate, we earn on qualifying purchases.
Six Years Before the MRI
Housman frames the episode against the couple’s experience of fertility care in the United States. He says appointment time was limited and that the couple had many detailed questions between visits. Their use of ChatGPT began after GPT-4 became available, with the system answering questions in the persona of a fictional medical-drama doctor. He describes the chatbot as patient and available, but also says the older models sometimes resisted medical queries and could make obvious mistakes.
The couple’s clinical experience, as recounted in the guest post, included repeated embryo transfers that did not implant and a wide range of investigations. Housman says an MRI had appeared as a possible option in one set of office notes, but was not emphasized, and that the couple did not pursue it at that time. He does not identify the dates of the appointments, the specific models used for each exchange or the full medical record. The central development is therefore a patient account of an overlooked possibility coming back into discussion—not a formal evaluation of AI-assisted diagnosis.
“That’s when our LLM doctor stepped in.”
— Drew Housman, in a guest report for Astral Codex Ten
What the Account Cannot Establish
The guest report does not include medical records, the MRI images or independent confirmation of the diagnosis and treatment plan. It is not clear from the account why the fertility doctor initially discouraged the MRI, what other factors informed that decision, or whether the fibroid was conclusively responsible for the failed embryo transfers. Housman says the doctor judged it needed removal, but the report does not provide the procedure’s outcome.
It is also unclear whether the chatbot’s MRI suggestion followed from specific details entered by the couple or from a broader response to their questions. The report supplies no comparison with other patients and no evidence about how often similar AI suggestions are useful or misleading. Housman’s description of the chatbot matching doctors’ diagnostic skill is his assessment of their experience, not a result of a clinical study. The account does not say whether the couple later achieved a pregnancy.
The Outcome After Fibroid Removal
The account says the fertility doctor recommended removing the fibroid, but it does not describe whether or when the operation took place, what clinicians found afterward, or whether the couple resumed embryo transfers. Those details would clarify the medical outcome for this family, though they would not by themselves establish how well AI chatbots work for other patients.
For now, the reported next step is a clinical one: the couple’s treatment plan depends on their medical team’s assessment of the MRI and the fibroid. Housman’s report leaves the broader question open—whether chatbots can consistently help patients identify useful questions without prompting unnecessary tests or giving false reassurance. That would require evidence beyond a single personal account.
Key Questions
What did ChatGPT suggest?
Housman says the chatbot suggested that his wife get an MRI during their investigation of repeated unsuccessful embryo transfers.
What did the MRI find?
According to Housman, the scan showed a 6-centimeter submucosal fibroid in his wife’s uterus. He says their fertility doctor recommended removing it.
Does the report prove the chatbot diagnosed infertility?
No. It is a personal account of a chatbot suggesting a scan. The report does not establish that the AI diagnosed the fibroid, that the growth caused the failed transfers or that the tool is reliable for other patients.
Did the couple have a baby after the MRI?
The supplied report does not say whether the fibroid was removed or whether the couple later achieved a pregnancy.
Source: rss
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
