Table of Contents
- The 30-second summary: the seven figures that matter
- 1. The trial: how it was done
- 2. The missing figure: interval cancers
- 3. What happens outside the trial
- 4. What the systematic review qualifies
- Practical applications for this
- Frequently asked questions
- Can artificial intelligence identify a higher number of breast cancers than radiologists?
- Do false positives increase?
- What is an interval cancer and why does it matter so much?
- How many hours of labor does it eliminate?
- Can AI read mammograms on its own?
- What we do not know
- References
Of everything promised about artificial intelligence in medicine, one case has completed the full course: randomised trial, prespecified primary endpoint, published result. It is breast cancer screening, and the trial is called MASAI. With 105,934 women randomised within Sweden’s national programme, AI-supported reading detected 29% more cancers than standard double reading, without increasing false positives, and cut reading workload by 44%. And in 2026 came the missing figure: interval cancers, the endpoint that actually measures whether a screening programme is safe. This article explains why this case counts and what sets it apart.
The 30-second summary: the seven figures that matter
Each row lists the study alongside its publication year. Be aware that the initial three entries represent consecutive papers stemming from the exact same trial.
| Finding | Result | Study and year |
|---|---|---|
| Cancers detected per 1,000 women screened | 6.4 versus 5.0; relative proportion of 1.29 (95% CI: 1.09-1.51; p = 0.0021) | Hernström et al., Lancet Digital Health (2025) |
| False positives | No elevation: ratio standing at 1.01 (95% CI: 0.91-1.11; p = 0.92) | Hernström et al., Lancet Digital Health (2025) |
| Radiologist reading workload | 44.2% reduction in scans examined | Hernström et al., Lancet Digital Health (2025) |
| Interval cancers, the primary endpoint | 1.55 compared to 1.76 per 1,000; proportion of 0.88, remaining non-inferior | Gommers et al., The Lancet (2026) |
| Screening sensitivity | 80.5% versus 73.8% (p = 0.031), maintaining identical specificity | Gommers et al., The Lancet (2026) |
| Real-world rollout in Germany, 463,094 women | 6.7 against 5.7 per 1,000: marking a 17.6% increase in detection | Eisemann et al., Nature Medicine (2025) |
| Reading reduction across 31 reviewed studies | Ranging from 40% to 90% via triage methods | Jassim et al., BMJ Open (2025) |
What you will discover here
- How MASAI is designed and why its design differs from the rest.
- What an interval cancer is and why it was the missing figure.
• What happened when it was taken to 463,094 women outside a trial.
• What the systematic review says is still unresolved.
1. The trial: how it was done
MASAI was conducted as part of the national screening initiative in Sweden, taking place across four distinct facilities situated in the country’s south-western region. During the timeframe spanning from April 2021 through December 2022, 105,934 female participants were allocated at random in a 1:1 ratio to undergo either artificial intelligence-assisted evaluation or conventional double reading devoid of AI. The median participant age stood at 53.7 years. Assignment was carried out automatically via the image archiving platform immediately following acquisition, ensuring that the interpreting radiologists remained blinded to the group assignment of each individual case.
The utility operates under a dual approach, and drawing a distinction matters since the final outcome hinges on it. The software rates every assessment on a ten-tier scale to determine if the screening requires single or double evaluation, while simultaneously highlighting suspicious areas for the attending radiologist. It never supplants human judgment; rather, it reorganizes workflow and directs attention.
The detection results, published in Lancet Digital Health in 2025: 6.4 cancers per 1,000 women screened in the AI group against 5.0 in the control group, a ratio of 1.29 (95% CI: 1.09-1.51; p = 0.0021).
What kind of cancers were additionally detected
This question is what separates a useful screening programme from one that merely generates diagnoses. The additional detection was mostly of small invasive cancers without nodal involvement: 58 more T1 tumours, 46 more lymph-node negative, and 21 more of subtypes other than luminal A. In situ carcinomas also increased, and about half of that increase was high grade.
And what did not happen: neither the recall rate nor the false-positive rate rose significantly, while the positive predictive value of recall improved with a ratio of 1.19 (95% CI: 1.04-1.37; p = 0.012). Reading workload fell from 109,692 to 61,248 reads: 44.2% fewer.
| Study | What it contributed | Result | What it cannot support |
|---|---|---|---|
| Lång et al., Lancet Oncology (2023) | Safety analysis with the first 80,033 women | Similar or higher detection; 44.3% fewer reads | An interim safety analysis, not the final endpoint |
| Hernström et al., Lancet Digital Health (2025) | Secondary outcomes with the full sample | 29% more detection without more false positives | Detection is not the same as patient benefit |
| Gommers et al., The Lancet (2026) | Primary endpoint: interval cancer | Non-inferior, with higher sensitivity and equal specificity | It does not measure mortality; follow-up is two years |
| Eisemann et al., Nature Medicine (2025) | Real-world rollout at 12 German centres | 17.6% more detection without worsening recall | Radiologists voluntarily chose whether to use the tool |
| Jassim et al., BMJ Open (2025) | Review of 31 studies and over 2 million examinations | Reading reduction of 40% to 90% with triage | Interval-cancer follow-up incomplete in several datasets |
2. The missing figure: interval cancers
Detecting more cancers does not by itself show that a screening programme is better. A programme can raise its detection figure by finding tumours that would never have caused trouble, while still missing the aggressive ones. The measure that catches this is the interval cancer: the one that appears between two screening rounds, that is, the one screening missed.
That constituted the main goal of MASAI, featured in The Lancet in 2026, utilizing a predefined non-inferiority margin of 20%.
| Measure | With AI | Standard double reading |
|---|---|---|
| Interval cancers per 1,000 participants | 1.55 (95% CI: 1.23-1.92) | 1.76 (95% CI: 1.42-2.15) |
| Proportion ratio | 0.88 (95% CI: 0.65-1.18) | p = 0.41: non-inferior |
| Invasive interval cancers | 75 | 89 |
| Interval cancers at stage T2 or above | 38 | 48 |
| Sensitivity | 80.5% (76.4-84.2) | 73.8% (68.9-78.3); p = 0.031 |
| Specificity | 98.5% | 98.5%; p = 0.88 |
The result meets the requirement and adds something more. Not only were there no additional interval cancers, but those that did occur in the AI group tended to be less invasive and at a lower stage, and programme sensitivity rose by nearly seven points without losing a tenth of a point of specificity. It is a clean result, and a rare one in this field.
3. What happens outside the trial
A clinical trial is a controlled environment. The next question is whether this holds up in a real programme, and there is a sizeable answer. The PRAIM study, published in Nature Medicine in 2025, followed 463,094 women aged 50 to 69 screened at 12 German centres between July 2021 and February 2023, read by 119 radiologists.
The detection rate in the AI-supported group was 6.7 per 1,000 against 5.7 in the control group: 17.6% more (95% CI: 5.7% to 30.8%), statistically superior. The recall rate was slightly lower and non-inferior, and the positive predictive value rose from 14.9% to 17.9%.
There is one restriction that the researchers themselves point out and which must not be overlooked: radiologists voluntarily chose whether to use the system, meaning this is not a randomized trial and disparities might exist between the group that utilized it and the one that did not.
4. What the systematic review qualifies
The most extensive synthesis currently accessible was brought to light by Ghufran Jassim and co-authors in BMJ Open in 2025: 31 studies covering more than two million screening examinations across Europe, Asia, North America and Australia.
Their conclusions align with the points above on the core aspects. Serving as a second reader or embedded in double-reading processes, artificial intelligence sustains or slightly boosts sensitivity—by as much as nine percentage points—while maintaining or enhancing specificity; additionally, triage setups cut reading volume by 40% to 90% while keeping detection non-inferior when thresholds are set conservatively.
However, the authors raise a pair of caveats. The initial point involves general reliability: variations among methodologies alongside reliance on selected populations restrict certain precision measures, while follow-up data regarding interval cancers remains partial across several key investigations. The subsequent point addresses independent deployment: clear documentation, uniform assessment frameworks, and extended demographic research must be established prior to evaluating artificial intelligence as an independent diagnostician.
Practical applications for this
Ordered by the strength of the evidence behind it:
• Always ask for the interval-cancer figure. It represents the critical metric that distinguishes finding more from finding better, and the exact data MASAI delayed publishing for three years.
• The role that is demonstrated is triage and support, not sole reader. That is what the trial evaluated and what the systematic review considers safe today.
• Look at what kind of tumour is additionally detected. In MASAI it was mostly small invasive cancers without nodal involvement, which is exactly what screening is for.
- Verify that the recall does not increase. The worth of the outcome stems from spotting more instances without incurring a penalty in false positives.
- Distinguish trial from rollout. The German study featuring 463,094 women boasts a massive scale, yet adoption of the tool remained voluntary, lacking true random assignment.
Frequently asked questions
Can artificial intelligence identify a higher number of breast cancers than radiologists?
In supported screening, yes, and it is measured in a randomised trial. In MASAI, with 105,934 women from Sweden’s national programme, detection was 6.4 cancers per 1,000 screened with AI against 5.0 with standard double reading, a ratio of 1.29 (95% CI: 1.09-1.51; p = 0.0021). The AI was not reading alone: it sorted examinations and marked findings for the radiologist.
Do false positives increase?
Not significantly. In the same trial, the false-positive ratio was 1.01 (95% CI: 0.91-1.11; p = 0.92) and the recall ratio 1.08 (0.99-1.17; p = 0.084), while the positive predictive value of recall improved with a ratio of 1.19 (p = 0.012).
What is an interval cancer and why does it matter so much?
It is the cancer that appears between two screening rounds, that is, the one screening did not detect. It was MASAI’s primary endpoint, and the result published in The Lancet in 2026 was 1.55 per 1,000 with AI against 1.76 without it, a ratio of 0.88 (95% CI: 0.65-1.18) meeting the non-inferiority criterion. Without that figure, detecting more cancers does not show that screening is better.
How many hours of labor does it eliminate?
Around half the reads. MASAI had 61,248 reads in the AI group against 109,692 in the control group, 44.2% fewer, and the review of 31 studies puts the reduction between 40% and 90% depending on the triage strategy used.
Can AI read mammograms on its own?
Not yet, on the available evidence. The systematic review published in BMJ Open in 2025 concludes that transparent reporting, standardised evaluation and long-term population studies are needed before considering AI as a stand-alone reader, and notes that interval-cancer follow-up is incomplete in several of the available datasets.
What we do not know
• Whether it reduces mortality. No study has measured it: the hardest endpoint available so far is interval cancer at two years.
• What happens across several screening rounds. The published follow-up covers one interval, not a lifetime of successive screens.
- Whether the outcome relies on the specific setup. MASAI utilized a distinct release of a commercial software, and comparative, randomised evaluations between platforms are absent.
• How it behaves in different populations. The large studies are Swedish, Danish and German, with organised screening programmes and relatively homogeneous populations.
References
1. Hernström, V., Josefsson, V. et al. Screening performance and characteristics of breast cancer detected in the Mammography Screening with Artificial Intelligence trial (MASAI). The Lancet Digital Health, 2025. DOI: 10.1016/s2589-7500(24)00267-x
2. Gommers, J., Hernström, V. et al. Interval cancer, sensitivity, and specificity comparing AI-supported mammography screening with standard double reading without AI in the MASAI study. The Lancet, 2026. DOI: 10.1016/s0140-6736(25)02464-x
3. Lång, K., Josefsson, V. et al. Artificial intelligence-supported screen reading versus standard double reading in the MASAI trial: a clinical safety analysis. The Lancet Oncology, 2023. DOI: 10.1016/s1470-2045(23)00298-x
4. Eisemann, N., Bunk, S. et al. Nationwide real-world implementation of AI for cancer detection in population-based mammography screening. Nature Medicine, 2025. DOI: 10.1038/s41591-024-03408-6
5. Jassim, G., Otoom, O. et al. Performance of artificial intelligence in breast cancer screening programmes: a systematic review. BMJ Open, 2025. DOI: 10.1136/bmjopen-2025-111360
This material is purely informational and should not be considered medical guidance. The outcomes detailed originate from structured screening initiatives employing dedicated systems and protocols, meaning they cannot be applied directly to alternative clinical environments or personal choices regarding diagnostic procedures.