Every day, one story that matters, broken down into three numbered scenarios, each with its own probability. Never fixed — it shifts as the situation changes.
Scénario
Friday, science
Breast cancer screening: is the model set to shift?
Faced with stagnating participation (45.7% in 2024–2025) and the rise of AI and risk personalization, is breast cancer screening in France shifting toward a new model?
Publié le 2 octobre 2026
Share:
The facts
Organized breast cancer screening in France is built on a universal, standardized model: every woman aged 50 to 74 receives an invitation every two years for a fully covered mammogram. Yet the program faces persistent declining participation. According to figures published by Santé publique France, the standardized uptake rate in the national organized breast cancer screening program* stood at 45.7% over 2024–2025, down significantly from 46.6% in 2022–2023. Despite recurring awareness campaigns, more than one in two eligible women do not attend their appointments scheduled by regional screening centers, undermining public health goals.
This apparent drop-off masks a more complex reality: many women choose to undergo screening outside the public framework, directly with their usual gynecologist or radiologist. A survey by DREES published in February 2026 highlights a major social divide in preventive care uptake. Overall screening participation—whether through public programs or individual care—ranges from 42% among women in the lowest income decile to 67% among the top 10%, a massive 25-point gap. Higher-income households often bypass the public track in favor of individually prescribed exams, while underprivileged groups fall through the cracks.
Understanding it
Understanding mass screening and why it is running out of steam
Public screening is like a periodic vehicle inspection notice sent on a fixed schedule to everyone. This uniform model treats every woman identically regardless of genetic profile, but it runs into cultural hurdles, fear of the exam, and a shortage of doctors in medical deserts.
Yet the medical stakes across the country remain huge. With 61,214 new diagnoses and 12,765 deaths recorded in 2023, breast cancer remains the leading cause of cancer mortality among women in metropolitan France. Early detection is the single most powerful tool to save lives: the 5-year standardized net survival rate reaches 88% when the disease is caught at an early stage. When a tumor is detected before palpable symptoms appear, treatment regimens are far less invasive, surgeries less extensive, and long-term remission rates significantly higher.
As the universal model runs out of steam, technological advances are opening up an unprecedented alternative powered by image analysis algorithms. The MASAI clinical trial conducted in Sweden on nearly 106,000 women and published in The Lancet in January 2026 showed that AI-assisted screening achieved a 12% reduction in interval cancers*, the aggressive tumors that emerge between two routine mammograms. The detection rate reached 80.5% with AI support compared to 73.8% with the standard approach, all while substantially cutting down the reading time required by radiologists.
At the same time, medical research is steadily moving away from a one-size-fits-all screening schedule toward risk-stratified screening based on genetic profile, breast density, and family history. Programs such as Interception, led by Unicancer, are evaluating targeted prevention pathways for high-risk profiles. The 2026–2030 roadmap published by the French National Cancer Institute (INCa) acknowledges this shift by explicitly planning to support personalized screening and define the role of AI and tomosynthesis* within the national program.
Understanding it
Understanding the role of AI and risk stratification
Instead of relying solely on age milestones, personalized screening combines medical history, genetics, and breast density to adjust the schedule. AI acts as an expert second set of eyes to spot subtle abnormalities without replacing the practitioner, allowing advanced examinations to focus on the highest-risk patients.
France is thus at a strategic crossroads. The current system, designed over twenty years ago on an egalitarian mass-screening model, struggles to engage the entire population and does not yet harness individual risk-assessment tools. Faced with demographic constraints in radiology and the urgent need to bridge regional divides, three forward-looking scenarios emerge: a complete transition to an AI-driven personalized model, the lasting continuation of a hybrid framework without a clean break, or the emergence of two-tier preventive medicine deepening social inequalities.
Participation in organized screening (PNDOCS)45.7% down from 46.6% in 2022–2023
Overall screening gap by income level (DREES)25 points 42% among the bottom 10% vs. 67% among the top 10%
Universal mammograms or AI-assisted tailored monitoring: the future of screening
What we're assessing
FavorableLe dépistage stratifié sur le risque individuel et assisté par l'IA remplace la mammographie systématique.
StableLa mammographie biennale reste la norme générale tandis que l'IA s'intègre par touches progressives.
DégradéLes innovations profitent aux populations aisées et creusent la fracture territoriale et sociale.
Favorable
30%
Likely
Shift toward personalized screening guided by risk and AI
In this forward-looking scenario, health authorities completely overhaul public screening by incorporating genetic data, family history, and automated image analysis. Routine biennial mammograms cease to be the sole standard, making way for a tailored schedule: women identified as low risk space out their checkups, while high-risk profiles receive close monitoring combining MRI and tomosynthesis*. Artificial intelligence becomes a frontline filter in regional centers, ensuring image-reading accuracy and freeing up valuable time for radiologists.
Compared to maintaining a uniform routine, this approach optimizes medical resource allocation and curtails overdiagnosis in low-risk patients. However, it requires a complete overhaul of hospital information systems and enhanced training for primary care physicians in assessing individual risk. If this restructuring succeeds in reaching underserved women more effectively, the overall survival rate could rise from 88% to unprecedented levels of 92% to 95% by 2030.
Indicators affected
Participation au dépistage organisé (PNDOCS)58 %↑45,7 % (2024-2025)
Écart de participation générale selon le niveau de vie (DREES)12 points↓25 points (2015-2020)
The France angleOptimized imaging spending and lower breast cancer mortality. ↑ Rather favorable for France.
Stable
50%
Likely
Prolonged coexistence of the traditional model and digital decision-support tools
The current mass-screening framework remains the benchmark for the vast majority of women aged 50 to 74. Technological innovations such as diagnostic software and pilot targeted programs like Interception are integrated in phased steps without altering the standard biennial schedule. Radiologists incorporate digital assistance into their daily practice to make double reading more reliable, but invitations continue to be issued based solely on age, maintaining a reassuring yet rigid institutional framework.
Unlike an immediate, full-scale overhaul, this incremental adaptation scenario avoids organizational disruptions but prolongs the extra costs generated by overlapping public and private systems. Participation in the national program caps within a narrow band around 50% due to persistent personal hesitancy and specialist shortages across several regions. Equipment upgrades would then progress at a measured pace without quickly bridging the preventive divide across social classes.
Indicators affected
Participation au dépistage organisé (PNDOCS)49 %↑45,7 % (2024-2025)
Écart de participation générale selon le niveau de vie (DREES)22 points↓25 points (2015-2020)
The France angleUniversal baseline maintained, but transitional extra costs and capped participation. ↓ Rather unfavorable for France.
Degraded
20%
Unlikely
Preventive divide and the rise of two-tier advanced screening
In this drifting scenario, cutting-edge technologies remain concentrated in major urban centers and private clinics accessible to well-informed populations with strong supplementary insurance coverage. Women from privileged backgrounds benefit from accurate genetic profiling and screening assisted by the latest AI tools, while patients in rural areas and disadvantaged neighborhoods face growing wait times for a basic standard mammogram in the public sector.
Far from democratizing medical breakthroughs, this path steepens the social gradient highlighted by DREES, turning risk personalization into a privilege for the well-informed. Regional coordination centers, strained by budget constraints and healthcare worker shortages, would see their role reduced to running a residual system. The premature mortality gap between socioeconomic groups would then risk widening significantly over the decade.
Indicators affected
Participation au dépistage organisé (PNDOCS)41 %↓45,7 % (2024-2025)
Écart de participation générale selon le niveau de vie (DREES)32 points↑25 points (2015-2020)
The France angleWidening regional health inequalities and loss of collective efficiency. ↓ Rather unfavorable for France.
Ordres de grandeur indicatifs pour les 3 scénarios ci-dessus, estimés avec l'information disponible à la publication et réévalués si la situation change — jamais des prévisions garanties. Learn more about our method →
Key takeaways
Should organized breast cancer screening abandon routine biennial mammograms in favor of an AI-assisted personalized model?
Participation in the national program is stagnating at 45.7% over 2024–2025, while the screening gap reaches 25 points between the bottom 10% and top 10% income brackets.
No, the model will not shift overnight: our most likely scenario (50%) anticipates a hybrid transition where biennial mammography remains the common baseline while AI is integrated gradually, compared to a full personalized shift (30%) or a two-tier divide (20%).
The rollout of the French National Cancer Institute’s 2026–2030 roadmap and the release of findings from regional AI trials will serve as a decisive test for updates to national protocols.
Slightly negative
Our assessment of the impact for France: slightly negative.the predominance of the hybrid adaptation scenario (50%) and the risk of an unequal drift (20%) limit immediate public health gains compared to the potential of a full personalized overhaul (30%).
Si tu devais retenir 1 chose
Avec seulement 45,7 % de participation au dépistage organisé et un écart de 25 points selon le revenu, le dépistage du cancer du sein cherche son second souffle entre intelligence artificielle et personnalisation du risque.
