← All programs
Program 02 · snapi

snapi

snapi is an investigational microbiome-based program for identifying women at elevated risk of preterm birth in early pregnancy. A microbiome matrix has been established in a Korean cohort, and Saem Research is designing a clinical study in Australia to extend the dataset across additional ethnic populations.

The market gap

A category open for redefinition

Current preterm-risk screening relies on cervical length ultrasound (sensitivity ~38% at the ≤25mm cutoff, Kuusela 2020), and the last commercially available rapid fFN biomarker test was discontinued in September 2024.

A $1.7B obstetric biomarker market (2025), projected to reach $2.5B by 2031 — without a standard product. The first validated workflow becomes the new clinical default.

Until 2024

Hologic Rapid fFN was the standard preterm-risk biomarker for ~25 years.

September 2024

Hologic discontinues global sales of the Rapid fFN® 10Q cassette — leaving no commercially available rapid fFN point-of-care test.

Now

Saem Research is advancing R&D on two products to fill this gap.

Sources: WHO Preterm birth fact sheet (updated 2024); Kuusela et al., BJOG, 2021; ESS Application April 2026.

The solution

A maternal-microbiome read for preterm-birth risk

snapi reads a focused vaginal-microbiome panel and integrates it with a locked machine-learning prediction model — returning a composite PTB risk score in hours, on standard hospital qPCR equipment, without new capital expenditure.

Step 01 · Sample
Vaginal swab

Clinician-collected vaginal swab during the standard prenatal exam at weeks 16–24. No new device. No blood draw.

Step 02 · qPCR
Multiplex panel

Targeted multiplex qPCR panel on existing hospital qPCR equipment. No central-lab dependency.

Step 03 · Report
Risk score

Composite PTB risk score returned within four hours on standard hospital qPCR.

Step 04 · Action
Clinical decision

Risk band informs clinical interventions: monitoring cadence, progesterone supplementation, cervical cerclage, and referral.

Existing PTB workflows are symptoms-driven. Risk is identified only once contractions, cervical change, or other clinical signs appear — typically too late for non-emergency intervention. snapi reads PTB risk from the maternal microbiome at week 16–24, before any clinical sign. The OB-GYN gains weeks of clinical runway to plan monitoring, progesterone, or cerclage.

The evidence

Validated. Locked. Externally tested across two years.

snapi's prediction model achieves AUROC 0.80 on an external temporal hold-out cohort (two years post training-set lock). TRIPOD+AI Type 3 framework. The algorithm and the decision threshold are both pre-locked at training; updates are governed by a Predetermined Change Control Plan (FDA GMLP-aligned). +0.150 AUROC over cervical length screening — externally validated, two years apart.

Cervical length ≤25mm — asymptomatic screening (Kuusela et al., 2020)
38.5%
snapi (microbiome + ML) — external temporal hold-out at the locked decision threshold
>79%
0.81
Training AUROC
0.80
External hold-out AUROC
0.949
NPV at 10% prevalence

A regulatory-grade cohort: Korea training + external temporal hold-out cohorts, two-year gap between training-set lock and external validation.

The pipeline

Built around a single longitudinal cohort

PTB Prediction is snapi's first product — externally validated and entering clinical trials. The platform is being designed so the same enrolled mother contributes to two additional diagnostic products as her pregnancy and her baby's first months unfold.

  • Current · Preterm birth prediction. Vaginal microbiome qPCR + locked machine-learning prediction model. Externally validated. MFDS pathway in flight.
  • Next · Breast Milk Quality Score (BMQS). ELISA + qPCR for sIgA, lactoferrin, and selenium in breast milk. In R&D.
  • Later · Infant gut health. Bifidobacterium and commensal markers from meconium and infant stool, tracked over 0–24 months. In development.

Platform vision: one enrolment. Five sample types — vaginal, maternal gut, meconium, breast milk, infant stool. One longitudinal cohort. One Master IRB. Each future product inherits the prior product's data.

The moat

The data, the workflow, and the algorithm

Reproducing this asset requires a paired mother-infant longitudinal microbiome cohort, a locked predictive model, and an MFDS-aligned regulatory pathway. None of these can be shortcut with capital alone.

  • Patents — assay + locked model. Priority date secured. Patent applications filed covering the assay design and the locked predictive model.
  • Cohort asset — cannot be retrospectively collected. A regulatory-grade longitudinal mother-infant cohort. Time, not capital — capital cannot shortcut it.
  • Regulatory — MFDS as APAC keystone. Hong Kong · Australia · UK · China. MFDS pathway as the keystone for an APAC cascade.
  • Modality — no new device, no new lab. Vaginal swab on installed hospital qPCR infrastructure. No new device, no central-lab dependency.
The science

Built on peer-reviewed research

The vaginal microbiome's role in preterm-birth risk is established science. snapi translates this body of evidence into a clinically validated diagnostic.

Fettweis JG et al. · Nature Medicine · 2019
The vaginal microbiome and preterm birth.

Foundational evidence that vaginal microbiome composition in early-to-mid pregnancy is an independent predictor of preterm-birth risk.

Park S et al. · American Journal of Reproductive Immunology · 2021
Prediction of preterm birth based on machine learning using bacterial risk score in cervicovaginal fluid.

Study from Ewha Womans University — snapi's clinical partner — demonstrating feasibility of qPCR + machine learning prediction (10-bacteria panel, sensitivity 71%).

Kuusela P et al. · BJOG · 2021
Second-trimester transvaginal cervical-length measurement for preterm-birth prediction.

Cervical length ≤25mm predicted spontaneous preterm birth at <33 weeks with sensitivity 38.5% — the current clinical benchmark snapi is positioned to improve.

Additional reading: Callahan et al., PNAS 2017 (independent cohort replication); Kindinger et al., BMC Medicine 2017 (microbiome × cervical length × progesterone); Ravel et al., PNAS 2011 (vaginal community state types).

See Program 01 · EveryBaby →