Series: 'Medical desertification and care wandering' — Article 5/5
This first part of article 5 addresses the operational question: what needs to be funded, who should lead the initial investment, how to align incentives, and what credible roadmap can take us from experimentation to national scale? Predictive prevention is not a mobile app or a digital gadget — it is a major socio-technical infrastructure, on par with the power grid, highways, or broadband.
The French healthcare system excels in acute care management — post-stroke survival in the EU Top 3, cancer outcomes in the global Top 5, organ transplantation as European leader. But it fails to anticipate, prevent, and support over the long term: healthy life expectancy stands at 64 years in France versus 72 in Sweden, and over 30 % of hospitalisations are avoidable — representing €4.5 billion wasted annually.
RPM devices, even when supervised by human professionals, enable a 25 to 45 % reduction in hospitalisations. Dozens of structuring experiments are underway (Kaiser Permanente, ETAPES, PREDICARE, Article 51), but all remain fragile and without national-scale deployment. Three-layer predictive prevention — risk stratification, digital twin monitoring, targeted human intervention — is technically mature. The barrier is organisational and political.
The transition requires five structural transformations. First, national governance through the creation of a National Agency for Preventive and Predictive Medicine. Second, readable and multi-year funding via dedicated ONDAM allocation and recycling of hospital savings. Third, massive professional training with 140 hours on AI and chronic care pathways in initial medical education. Fourth, an interoperable and ethical information system with real-time ENS and mandatory FHIR APIs. Fifth, outcomes-based management with a public dashboard and independent audits.
The challenge is no longer scientific — the evidence is established. It is no longer economic — the ROI is demonstrated. It is political. The question is not “can we?” but “do we want to?”. Article originally published on LinkedIn (in French).