NIPT in 2026: How AI and next-gen sequencing are changing prenatal screening

Article produced in association with Jeen Health

Non-invasive prenatal testing has been part of UK antenatal care since the early 2010s, but the technology underneath it has changed considerably.

Two developments in particular are reshaping the field in 2026: advances in next-generation sequencing (NGS) that widen what the test can detect, and the addition of artificial intelligence at the analysis stage.

Together, they are making NIPT more accurate on the conditions it has always screened for, and more capable on conditions it could not previously reach.

The practical change for patients is quieter than the headlines suggest. The blood draw and the laboratory workflow look the same. What has changed is what happens after that, and what the result can reliably say.

What NIPT does, and what NGS made possible

NIPT screens cell-free DNA (cfDNA) in maternal blood. Around 10 per cent of that cfDNA comes from placental tissue, and in most pregnancies it accurately reflects the fetal genome.

By sequencing the fragments and counting how many map to each chromosome, a laboratory can assess whether any chromosome is over- or under-represented, which is the signal used to identify trisomy 21, 18 and 13.

NICE Diagnostics Guidance DG46 sets out the evidence base that underpinned the NHS adoption of NIPT for these three conditions.

Next-generation sequencing is what made this work at clinical scale.

Most NIPT platforms use one of two sequencing approaches: massively parallel sequencing, which reads the entire genome at low depth, or targeted sequencing, which focuses on specific chromosomal regions. Reliable results are possible from around ten weeks of pregnancy, earlier than the combined screening test.

The AI layer: how machine learning is changing analysis

The conventional way of interpreting NIPT data is statistical. Read counts are compared with expected distributions using a Z-score or similar metric, and a threshold determines whether a chromosome is flagged as higher-chance.

This works well for the common trisomies, but it is limited by noise in the signal – particularly in samples with low fetal fraction or subtle structural variants.

Machine learning sits alongside that approach rather than replacing it.

The aiD-NIPT algorithm, published in Frontiers in Genetics, illustrates the direction of travel: a convolutional neural network trained on cfDNA fragment distance patterns.

In an analysis of 17,678 clinical samples, the ensemble version showed sensitivity of 99.07 per cent across trisomy 21, 18 and 13, with a positive predictive value of 88.43 per cent.

The conventional Z-score method applied to the same samples showed a PPV of 40.77 per cent.

The sensitivity figures are close; the difference in PPV is large, and it translates directly into fewer false-positive results and fewer unnecessary diagnostic procedures for patients who would otherwise be sent on to amniocentesis or CVS.

The widening scope of NIPT

Alongside the analytical changes, the range of what NIPT can detect has widened in stages. The first generation screened only for trisomies 21, 18 and 13.

Sex chromosome aneuploidies came next.

Microdeletion syndromes – 22q11.2 (DiGeorge), 1p36, Prader-Willi and Angelman, Cri-du-Chat and others – have since moved into routine private panels.

The newest tier reaches single-gene (monogenic) conditions, where the paternal contribution can be isolated from the maternal cfDNA background for selected autosomal dominant variants.

Each expansion has been driven by improvements in next-generation sequencing – higher coverage, better read accuracy, and bioinformatics pipelines capable of separating true biological variation from sequencing noise.

Niptify Focus Plus: where UK screening technology sits in 2026

A useful way to see how far the technology has moved is to look at one of the most capable NIPTs currently available privately in the UK.

Niptify, offered in UK by Jeen Health, is based on whole-genome sequencing of cfDNA, but its distinguishing feature is what happens before the sequencing stage – a proprietary enrichment step called Focus Plus.

Focus Plus filters the cfDNA sample by fragment size, retaining the shorter fragments that are disproportionately of fetal origin and reducing the maternal DNA background.

The practical effect is an increase in the effective fetal DNA fraction of roughly 3.6× compared with conventional NIPT workflows.

Two clinical consequences follow from that enrichment.

The first is a markedly lower no-call rate. Niptify reports a first-attempt success rate above 99 per cent, falling to under 0.1 per cent after retesting – placing it at the bottom of the range for UK-available private NIPTs, where no-call rates of 1-3 per cent are more typical.

For patients, this matters because a no-call result usually means a repeat blood draw, delayed results and an anxious wait – all avoidable when the fetal fraction is adequate first time.

The second is that pregnancies which traditionally produce less reliable results on standard NIPT – higher maternal BMI, donor-egg conceptions, surrogacy – become routinely screenable.

Focus Plus retains enough fetal DNA to be assessed directly, instead of relying on a contrast between fetal and maternal genomes.

The screening scope matches the technical uplift.

Alongside the common trisomies (at a 99.9 per cent detection rate), Niptify covers sex chromosome aneuploidies, rare autosomal trisomies and monosomies, segmental aneuploidies, more than twenty clinically relevant microdeletion syndromes, and genome-wide copy number variants larger than 1 Mb.

Unusually among UK-available NIPTs, it also screens for three mitochondrial DNA point mutations (1095T>C, 1494C>T, 1555A>G) linked to sensorineural hearing loss – clinically useful findings, because certain aminoglycoside antibiotics can trigger hearing damage in carriers and can be avoided if the variant is known.

Samples are processed in a CE-IVD certified, ISO 15189 accredited laboratory, and the test has consistently scored well in EMQN external quality assurance.

Put against other expanded panels available in the UK, Niptify’s combination of Focus Plus enrichment, whole-genome sequencing base and 20+ microdeletion coverage is distinctive.

Most competing panels cover the common trisomies and a handful of microdeletions; few screen the genome for CNVs above 1 Mb, and mtDNA screening remains rare.

The UK clinical pathway

On the NHS in England, NIPT is offered via two defined pathways, as summarised in the NHS Genomics Education Programme knowledge hub.

The Fetal Anomaly Screening Programme (FASP) route offers NIPT from 11 weeks as a second-line test following a higher-chance result on the combined or quadruple screen.

The R445 pathway, open from 10+0 weeks, is available to women with a history of a full trisomy 21, 18 or 13 pregnancy.

Laboratory delivery runs through NHS Genomic Laboratory Hubs, using both massively parallel sequencing and microarray technologies following the 2019 UK NSC decision to allow both approaches in procurement.

Private NIPT operates on different terms: no contingent requirement, earlier gestational start, broader condition menus.

The technology gap between what the NHS currently commissions and what is available privately – particularly on microdeletions and CNV detection – has grown wider as panels like Niptify have come to market.

Looking further ahead: whole genome sequencing after birth

The technology visible in NIPT today is a preview of a larger trend. Genomics England is running the Generation Study, a research programme in partnership with NHS England that uses whole genome sequencing on 100,000 newborns to screen for variants linked to more than 200 rare conditions.

The Generation Study is research rather than routine care, but the direction is clear: the boundary between what NIPT screens for in pregnancy and what newborn sequencing screens for after birth is narrowing, and the underlying pieces – high-depth sequencing, AI-driven interpretation, expanded panels – are shared.

Limitations that still matter

None of this changes the underlying truth that NIPT is a screening test, not a diagnostic one.

A higher-chance result continues to require confirmation by amniocentesis or chorionic villus sampling.

Confined placental mosaicism remains a recognised cause of false positives, because the placental cfDNA does not always match the fetal genotype.

Positive predictive values fall as the base rate of a condition falls, which is why rare microdeletion panels tend to have lower PPVs than trisomy 21.

Low fetal fraction samples still produce ‘no-call’ results, though workflows like Focus Plus have reduced that failure rate substantially.

AI models, like any trained algorithm, only perform well on populations that resemble their training data – so continued evaluation on diverse UK cohorts matters for equitable performance.

What is reasonable to say about NIPT in 2026 is this: the test is more accurate on the conditions it always screened for, and clinically useful on conditions it could not previously reach.

The clinical conversation around it is more nuanced, and pre-test counselling more important than ever.

These changes have come less from a single breakthrough than from steady, incremental progress – in NGS read quality, in fetal DNA enrichment workflows, and in the AI methods now applied to interpretation.

If you are considering NIPT or a broader prenatal genomic assessment, speaking with a genetic counsellor is a sensible first step, as they can explain the differences between the available panel options clearly

Disclaimer: This article is produced for informational purposes only and does not constitute medical advice, diagnosis or treatment. Clinical guidance referenced reflects published NICE, NHS and Genomics England information as at April 2026. Individual circumstances vary; readers are advised to consult a qualified healthcare professional before acting on any information in this article. This piece was produced in association with Jeen Health, which provided background clinical information for editorial purposes. Hyperlinks to external sources are included for reference only and do not represent an endorsement of any product, service or organisation.