Partial artificial gestation is used clinically for premature humans by 2126.

Current estimate

Resolution criteria

2126

Approved extra-uterine support replaces a meaningful portion of late gestation for premature human infants.

Probability history

Sources

1. A live query against the ClinicalTrials.gov API returns no registered human trial of any artificial womb, artificial placenta or extra-uterine support device, and the leading developer's own website states its technology has no market clearance from the FDA or any other regulator

   clinicaltrials.gov · 2026 · accessed Sep 12, 2026 · trial_registry_query_and_company_disclosure

   A direct query of the ClinicalTrials.gov version 2 API for artificial womb, extrauterine and artificial placenta terms combined with premature returned only unrelated studies - milk protein supplementation, gut microbiome, caesarean partographs, endometriosis diagnostics and IVF trials. No study touching artificial womb, EXTEND, artificial placenta or ectogenesis devices appears. Vitara Biomedical's own resources page states: 'Vitara Biomedical does not currently offer medical devices for sale or distribution and the company's technology has not received market clearance status by the U.S. FDA or any other country's regulatory authority.' The FDA's Pediatric Advisory Committee discussed the technology on 19 and 20 September 2023 under docket FDA-2023-N-2680; the 19 September session was open and the 20 September session closed for confidential commercial information. No vote, formal recommendation, guidance document or further advisory-committee meeting dated 2024, 2025 or 2026 was found on fda.gov.

   Limit: A registry query cannot exclude a trial registered under a sponsor-obscured or device-generic title, and it cannot see an FDA Investigational Device Exemption, which is not public. Four national registries were not queried. So this establishes the absence of a publicly registered trial in the one registry that was queried primarily, not the global absence of any authorised trial.

   The registry result is a primary-source negative: the ClinicalTrials.gov v2 API was queried live and the full result set read, not inferred from a news claim or a search-engine snippet. Vitara's disclaimer was fetched from the company's own site and read verbatim. The FDA meeting announcement page was fetched and read. The limits are stated rather than hidden. WHO ICTRP could not be searched at all - its portal returned a server error. ISRCTN and ANZCTR were covered only by web search, not by direct registry query. EU CTIS, the Netherlands register, Japan's jRCT and the Chinese registry were not searched. The FDA absence for 2024 to 2026 rests on two targeted site-scoped searches, not a page-by-page crawl of the advisory-committee calendar.

2. The European Union's Perinatal Life Support project closed on 30 September 2024 having produced a spin-out company rather than a device, and that successor was still developing a training manikin before any animal testing as of June 2025

   cordis.europa.eu · 2025 · accessed Sep 12, 2026 · government_project_record_and_institutional_announcement

   CORDIS records Horizon 2020 project 863087, Perinatal Life Support, as running from 1 October 2019 to 30 September 2024, status Project Closed, total cost and EU contribution 2,876,131.25 euro at 100 percent funding, coordinated by Eindhoven University of Technology with Universitatsklinikum Aachen, Politecnico di Milano, LifeTec Group BV and Nemo Healthcare BV. Eindhoven's own news post of 11 June 2025 describes the commercial successor AquaWomb as developing a training manikin first, with animal testing a later and deliberately reduced step made possible by mathematical modelling, and states the company expects to run on innovation and research subsidies in its first few years. No human trial timeline is stated.

   Limit: A roadmap stated in a university news post is a plan, not a measurement of progress, and a project closing on its scheduled end date is the normal outcome for a fixed-term Horizon grant rather than evidence of failure. Nothing here establishes what the consortium's deliverables actually demonstrated; it establishes only that the European effort has no device at trial stage.

   Both sources were fetched and read directly. CORDIS is the European Commission's own record, so the dates and the funding figure are primary. The Eindhoven post is the institution's own announcement about its own spin-out, so it is primary for the company's stated roadmap and self-interested in the optimistic direction, which strengthens rather than weakens the reading here: an institution promoting its spin-out placed it before animal testing.

3. The claim that Vitara Biomedical's artificial womb holds FDA breakthrough device designation traces to a single local news item from September 2023, is repeated by aggregators, and appears in nothing the company itself has published

   vista.today · 2026 · accessed Sep 12, 2026 · press_characterisation_versus_company_disclosure

   A September 2023 Philadelphia local outlet, Vista Today, appears to be the origin of the characterisation that Vitara's device holds FDA breakthrough device designation, and aggregator sites have repeated it since. A breakthrough designation is normally announced by the company that receives it, because it is a commercial asset. Vitara's own resources and press page carries no such claim and instead states that its technology has not received market clearance from the FDA or any other regulator. Its November 2024 financing release, which announces a 50 million dollar round led by Sands Capital with GV, First Spark Ventures, Flerie and Khosla Ventures, and names Kim Rodriguez as chief executive, states the money will advance efforts leading to a first-in-human clinical study and does not mention a designation either. The FDA does not publish a searchable device-level list of breakthrough designations, so the claim can be shown to be uncorroborated but cannot be shown to be false.

   Limit: This record establishes that the claim is unsourced, not that the designation does not exist. A company may hold a designation and decline to publicise it, and a startup between financing rounds has ordinary reasons to say little. The finding is about the evidence chain, not about the regulatory fact.

   Vitara's own resources page and the BioSpace release were fetched and read in full. The FDA's public Breakthrough Devices Program pages were searched and no company-issued or FDA-issued announcement of a designation for Vitara was found. This is a negative from absence of corroboration, and its ceiling is stated: because FDA maintains no public device-level list, absence from public sources is consistent with a designation that was simply never announced.

4. No peer-reviewed animal study published between 2023 and 2026 reports parity with in-utero controls on growth or IGF-1, and the one 2024 study reporting body-weight gain reports a fallen brain-to-body ratio consistent with oedema and does not measure IGF-1 at all

   pmc.ncbi.nlm.nih.gov · 2026 · accessed Sep 12, 2026 · peer_reviewed_animal_studies_and_absence_search

   The 2024 EXTEND study (Pediatric Research 2024 Jun 4;96(7):1616-1625, doi 10.1038/s41390-024-03287-0) supported ten lambs cannulated at 90 to 95 days gestational age for a mean of 13.5 days, range 10 to 21, with 9 of 10 completing the protocol, and reports weight rising from about 0.98 kg at cannulation to 1.57 kg at necropsy and brain weight of 26.6 grams. Against its extremely preterm control cohort the brain-to-body ratio fell, 17.6 percent against 22.7 percent, which the authors themselves treat as a signal that body-weight gain is confounded by oedema rather than somatic growth. The paper reports no IGF-1, no liver, kidney or lung weight and no humerus length, so it did not re-measure the markers on which the 2023 deficit was established and cannot be read as closing that gap. The 2023 Frontiers in Physiology study remains the reference for the deficit: lung 37.0 against 45.7 grams, liver 53.0 against 75.25 grams, kidney 9.5 against 12.9 grams, humerus growth 2.9 against 4.8 centimetres, IGF-1 17 against 140 micrograms per litre, all at p below 0.05, with six supported animals against eight controls. A 2024 review reports a separate 14-day cohort reproducing the same pattern. A search across PubMed, Europe PMC, bioRxiv, Frontiers and Pediatric Research for the period 2023 to 2026 found no study reporting parity.

   Limit: An absence established by literature search is bounded by the search. Unpublished second-generation results may exist, particularly inside a commercially funded programme, and would not appear. A Japanese and Western Australian programme published a result described in institutional material as sustaining a 24-week-equivalent lamb in a stable state for five days, but the underlying paper was not read and five days is in any case too short an interval for somatic growth to be a measurable endpoint, so it neither closes nor bears on this gap.

   The 2024 EXTEND paper and the 2023 Frontiers paper were both read in full from their PMC records; the EXTEND figures required three retrieval passes because the first mis-ordered a table, and the resolution was that the study defines two separate control cohorts at different gestational ages. Point estimates are reliable; the error bars are lower-confidence because they passed through a summarisation step. The 2024 review at PubMed 39597933 could not be read in full - a reCAPTCHA wall and an empty Europe PMC record - so its figures come from an abstract-level snippet corroborated by a second review that paraphrases the same study, and it is treated here as supporting rather than load-bearing. A January 2026 systematic review reaches the same top-line conclusion, but its per-study table returned every sample size as unspecified and contains a sentence contradicting its own conclusion, so it is deliberately not cited as a source here.

5. IGF-1 infusion improved organ weights but not body weight in normally growing fetal sheep and failed entirely in growth-restricted fetal sheep, and no published study has attempted IGF-1 or placental hormone supplementation inside an artificial placenta circuit

   journals.physiology.org · 2025 · accessed Sep 12, 2026 · peer_reviewed_intervention_studies_and_absence_search

   In normally growing late-gestation fetal sheep, chronic intravenous recombinant IGF-1 at 25 micrograms per kilogram per hour for one week raised heart weight from 21.9 to 27.6 grams at p 0.0202, kidney from 19.3 to 23.9 grams at p 0.0072, adrenal from 0.38 to 0.57 grams at p 0.0011 and spleen from 5.8 to 8.2 grams at p 0.0146, while leaving total body weight unchanged at 3.021 against 3.032 kilograms, p 0.909. In growth-restricted fetal sheep - the physiological state that actually resembles an animal on extra-uterine support - IGF-1 LR3 at 1.17 micrograms per kilogram per hour for one week did not improve fetal growth at all, in contrast to its effect in normal fetuses, and the authors conclude that amino acid, insulin and oxygen levels would also need correcting. The 2023 artificial placenta study itself attributes its deficit to reduced IGF-1, hypoalbuminaemia-driven oedema and hydrocortisone dosing interfering with the growth hormone to IGF-1 axis. No published study has attempted exogenous IGF-1, placental lactogen or placental growth hormone inside an artificial placenta circuit.

   Limit: The null in growth-restricted sheep is abstract-level and unread, and placental insufficiency is a related but not identical state to extra-uterine support - the fetus on a circuit has adequate oxygen delivery, which a growth-restricted fetus does not, so one of the co-factors the authors name is already corrected in the artificial placenta case. The absence of an in-circuit attempt is a gap in the literature, which is not the same as a failed attempt, and it cuts both ways.

   The 2022 Frontiers study in normal fetal sheep was read in full from its PMC record and its figures are internally consistent. The growth-restriction study could not be read - the journal returned HTTP 403 - so its dose, duration and null result come from search-result snippets at abstract level and are labelled as such rather than presented as read. The absence of any in-circuit supplementation attempt was established by searches across PubMed, PMC, Europe PMC and bioRxiv combining artificial placenta and artificial womb and EXTEND with IGF-1, placental lactogen and growth hormone.

6. Among 22,953 United States infants born at 22 to 25 weeks in 2020 to 2022, survival to discharge at 22 weeks rose from 10.9 to 35.4 percent against the 2013 to 2018 cohort while survival without severe complications at 22 weeks was 6.3 percent

   publications.aap.org · 2025 · accessed Sep 12, 2026 · multi_centre_registry_cohorts

   The Vermont Oxford Network cohort covers 22,953 infants born at 22 to 25 weeks gestational age across 636 United States hospitals in 2020 to 2022. Among those receiving postnatal life support, survival to discharge was 35.4 percent at 22 weeks, 54.8 at 23, 71.3 at 24 and 81.9 at 25. The corresponding figures for the 2013 to 2018 cohort were 10.9, 49.4, 69.9 and 79.2, so the 22-week figure roughly tripled while 23 to 25 weeks moved by a few points. Across all live births at 22 weeks rather than only those given life support, survival was 24.9 percent. Survival without severe complications ranged from 6.3 percent at 22 weeks to 43.2 percent at 25. The international iNeo comparison, 5,019 infants across 11 networks including 846 at 22 weeks and 4,173 at 23, published in JAMA Pediatrics on 25 August 2025, found survival to discharge ranging from 9 to 64 percent at 22 weeks and 16 to 80 percent at 23 between networks, with survival free of severe brain injury ranging from 7 to 53 percent and 9 to 69 percent.

   Limit: The comparison between the 2020 to 2022 and 2013 to 2018 cohorts is between two published cohorts and not a controlled before-and-after, and the tripling at 22 weeks is confounded by a simultaneous rise in the rate at which 22-week infants are offered active care at all - a denominator that changes when practice changes. The severe-complication definitions differ between the two registries cited, so the 6.3 percent and the 7 to 53 percent range are not directly comparable to each other.

   The iNeo paper was read in full from its PMC record. The Vermont Oxford figures were obtained at abstract level by two independent routes - a web search snippet and an NCBI eutils XML fetch - which agreed with each other on the 22-week and 25-week figures; the full paper's methods and tables were not read, and this is stated rather than implied. Neither cohort was reached through the AAP page directly.

7. The FDA withdrew approval of hydroxyprogesterone caproate in April 2023 after its confirmatory trial failed to show any reduction in preterm birth, and no replacement prevention candidate was found, so the route by which prevention makes an artificial womb unnecessary currently has no live candidate

   federalregister.gov · 2023 · accessed Sep 12, 2026 · regulatory_withdrawal_and_official_statistics

   The FDA withdrew approval of Makena and all generic hydroxyprogesterone caproate on 6 April 2023, for both intramuscular and subcutaneous formulations, after a nearly three-year hearing and appeal process, because the post-approval confirmatory PROLONG trial failed to demonstrate that the drug reduced preterm birth risk in women with a prior spontaneous preterm birth. The drug had been on the market for roughly two decades under accelerated approval. The World Health Organization records 13.4 million preterm births globally in 2020, 9.9 percent of all births, with more than 90 percent of extremely preterm babies in low-income countries dying within days against under 10 percent in high-income settings. A Lancet systematic analysis places extremely preterm births below 28 weeks at 4.2 percent of all preterm births, 95 percent confidence interval 3.1 to 5.0, roughly 567,800 babies.

   Limit: The most important number for this route was not obtained. The trend direction for extremely preterm birth incidence, globally and in high-income countries, is not established from any directly read primary source, because the one paper that answers it was inaccessible. A search-tool paraphrase suggesting no measurable global change with increases in some high-income countries was deliberately not treated as verified. Absence of a replacement candidate rests on one retrieval pass and is weak evidence about a century.

   The WHO fact sheet was fetched and read directly and is the source of the 13.4 million and 9.9 percent figures and of the survival disparity. The Federal Register withdrawal was confirmed across the notice and multiple state Medicaid bulletins citing the same FDA decision, but the notice itself was not fetched in full. The Lancet paper returned HTTP 403 on every attempt, so its 4.2 percent and 567,800 figures come from a search-engine synopsis and are labelled as unread. No pharmacological or mechanical pipeline candidate replacing hydroxyprogesterone caproate was found; cervical cerclage was named but not searched, so it is uncovered rather than absent.

8. Perfluorocarbon partial liquid ventilation, a perinatal technology with strong preclinical evidence and a funded developer, was formally discontinued in May 2001 after a truncated phase three trial produced a mortality signal in the wrong direction

   nejm.org · 2001 · accessed Sep 12, 2026 · clinical_trial_history_and_systematic_review

   Partial liquid ventilation with perflubron entered human study in premature infants with severe respiratory distress syndrome, reported in the New England Journal of Medicine in September 1996. The multicentre randomised trial enrolled 182 patients across 65 centres and was stopped early, before reaching target enrolment and before it could detect a clear difference against conventional mechanical ventilation. Twenty-eight-day mortality was 22 percent on partial liquid ventilation against 14 percent on control, not statistically significant on the truncated sample but not excluding clinically meaningful harm. A separate adult acute respiratory distress trial reported 28-day mortality of 15 percent conventional, 26.3 percent low-dose and 19.1 percent high-dose. Alliance Pharmaceutical formally discontinued LiquiVent for acute lung injury, acute respiratory distress syndrome and neonatal respiratory distress syndrome on 29 May 2001, after phase three failed to show superiority over high-frequency oscillatory ventilation. Cochrane's standing conclusion across both its paediatric and adult reviews is no evidence of benefit with some evidence of increased harm.

   Limit: One analogue is not a base rate, and the analogy is imperfect in a way that matters. Partial liquid ventilation was a modification of ventilation for lungs that had already been born into air; extra-uterine support proposes not to ventilate at all. The failure mode that killed it - a harm signal against an improving comparator - is nonetheless available to any device tested against a standard of care that is itself moving.

   The NEJM citation and the Cochrane review identifiers were located and cross-confirmed across three independent sources. Neither full text was fetched directly, so the percentages are secondary-sourced and internally consistent rather than read from the primary papers, and this is stated rather than implied. The 29 May 2001 discontinuation date and its stated reason rest on the same secondary chain.

9. Of 1,041 FDA breakthrough device designations granted between 2016 and September 2024, 127 or 12.2 percent were authorised in that window, and of 418 high-risk cardiovascular devices approved between 1977 and 2021 only 36 or 8.6 percent were approved for paediatric use, with 19 of those 36 supported by no clinical study containing a paediatric patient

   pmc.ncbi.nlm.nih.gov · 2025 · accessed Sep 12, 2026 · peer_reviewed_regulatory_cohort_analyses

   Between 1 January 2016 and 30 September 2024 the FDA granted 1,041 breakthrough device designations, of which 127, or 12.2 percent, were authorised within that window; 75 of the 127 were therapeutic devices analysed in depth. Among those 75, mean review time was 225.8 days, only 57 percent were reviewed within statutory timelines, 89 percent underwent some premarket clinical testing but nearly half of primary effectiveness endpoints used surrogate measures and 19 percent lacked statistical testing, median follow-up for implantables was 6 months with an interquartile range of 3.8 to 12 months, no postmarket study was required for 60 percent including 7 of the 11 that missed their premarket effectiveness endpoints, and 9 of 75, or 12.0 percent, had been recalled by November 2024. Separately, of 418 cardiovascular devices approved by the FDA between 1977 and 2021, 36, or 8.6 percent, were approved for paediatric use; 19 of those 36 were supported by no clinical study that included a paediatric patient; and across all trials for those 36 devices, 846 of 23,581 enrolled patients, or 4 percent, were paediatric.

   Limit: Neither cohort is the right reference class. The breakthrough programme's population is not committee-gated, and the cardiovascular cohort is a different device class. More importantly, both are censored: a designation granted in 2023 has had one year to reach authorisation, so the 12.2 percent is a rate within a window and not a lifetime conversion rate.

   Both papers were fetched and read from their PMC records and the figures are direct. Two claims were checked and could not be established, and are recorded as absent rather than approximated: no published statistic exists for the fraction of devices reaching an FDA advisory committee discussion that subsequently reach market, which is the population this forecast actually sits in; and the frequently repeated claims that paediatric devices lag adult devices by five to ten years and that 79 percent of high-risk devices lack paediatric approval trace to a paper whose full text was behind a cookie wall and are therefore unverified.

10. A 247-paper scoping review maps the ethics of artificial womb technology onto consent, moral status and legal status without finding a prohibition, and the condition the FDA's own panel named as blocking a first-in-human trial is the absence of a validated animal model rather than any legal barrier

    fda.gov · 2026 · accessed Sep 12, 2026 · regulatory_briefing_and_bioethics_scoping_review

    A scoping review in the Journal of Perinatology identified 247 papers on the ethics of artificial womb technology, of which 114, or 46.2 percent, address the extreme prematurity application and 133, or 53.8 percent, address speculative full ectogenesis, across seven theme categories including benefits and harms, decision-making, moral and legal status, justice and access, research ethics and speculative concerns. The recurring substantive problems are that the pregnant person must consent to transfer, and that the supported entity occupies an ambiguous category that is neither clearly fetus nor clearly neonate, which destabilises viability as the threshold United States constitutional law uses. Against this, the FDA's September 2023 briefing framework for a first-in-human trial names four conditions - prospect of direct benefit to the subject, risk justified by anticipated benefit, parental consent, and a validated animal model - and it was the fourth that the panel identified as not yet met, noting that lamb, piglet and non-human primate models carry between-species differences that may prevent direct representation of the human neonate. No jurisdiction was found to have legislated a prohibition on extra-uterine gestation research specifically.

    Limit: An absence of prohibition established by one search pass across a small number of jurisdictions is thin evidence about a hundred years, during which any number of jurisdictions may legislate in either direction. The validated-animal-model condition is also not a permanent barrier; it is a description of what the current evidence base lacks, and it is exactly what a second generation of animal work would supply.

    This is the weakest-verified record in this cycle and the weakness is stated rather than hidden. The FDA ethics briefing document at media/172267 and the meeting minutes at media/174946 were both attempted directly and both returned unparseable binary content, so their contents here are reconstructed from three independent journalistic accounts that themselves cite the primary documents, not from a direct read. The scoping review's figures come from its abstract by search, not from a fetched full text. The absence of an AWT-specific prohibition was established by search and is explicitly not extended to the 14-day embryo rule, which governs early embryo culture and is a different technology at a different stage; conflating the two would manufacture a finding.