Synthesis for public decision-makers · v2 · May 2026 · 11 pages
The French healthcare system is entering a zone of durable structural tension. As of January 1, 2024, 13.8 million patients held ALD status (long-term condition registry). Chronic trajectories already concentrate close to two-thirds of reimbursed expenditure, with a projection approaching 75% by 2035. This evolution unfolds against a cumulative backdrop of medical demographic contraction, organizational fragmentation, professional cognitive saturation, and growing territorial heterogeneity.
This brief defends a precise proposition: medical wandering and drift result less from local malfunctions than from a progressive inadequacy between the historical architecture of the system — optimized for the acute episode — and the epidemiological nature of the trajectories it must now govern. It deduces from this diagnosis the minimal properties that a sustainable vigilance regime should satisfy, and exposes the political arbitrations that no technical analysis can settle in lieu of the public decision-maker.
Medical drift is not the failure of insufficiently attentive actors; it is the structurally expected effect of an intermittent vigilance regime confronted with trajectories that have become longitudinal, multifactorial, and dynamic.
Public decision-makers and their cabinets (Health, Public Accounts), central administrative directorates (DGOS, DGS, DSS), Regional Health Agencies (ARS), the National Health Insurance Fund (CNAM), HAS-CEESP authorities, territorial conferences, learned societies and professional orders, and territorial elected officials with public-health competencies. It aims to structure the political discussion upstream of any deployment, not to propose a program.
Four structural determinants converge: demographic constraints, economic incentives oriented toward acute care (T2A), cognitive load exceeding human capacities, organizational fragmentation of care pathways. Taken together, they produce a structurally reactive system. Wandering (disorganized over-utilization) and drift (structural under-utilization, progressive invisibility) are not independent malfunctions: they designate two states of the same observation device that has become inadequate.
Comparative examination of four contrasted configurations (British NHS, Netherlands, Nordic models, integrated American models Kaiser and VHA) shows that none constitutes a directly transposable solution. Each reduces certain vulnerabilities at the cost of residual tensions. The conclusion is precise: the transformation required is not a matter of adopting a foreign architecture, but a matter of changing the surveillance regime — a change that the foreign configurations themselves have not fully effected.
The brief introduces, as a doctrinal heuristic, a conceptual operator: clinical vigilance density, defined as the effective proportion of a pathological trajectory that is the object of an interpretable observation. This operator provides a comparison tool across national regimes, pathologies, and temporalities of a single trajectory. It is not, at this stage, a calibrated metric — its translation into a quantifiable indicator belongs to subsequent methodological work.
The brief states six minimal properties — strictly bounded longitudinal visibility, interpretive continuity of trajectories, explicit prioritization under resource constraints, cognitive sustainability, preserved centrality of human clinical reasoning, compatibility with a fragmented system — and five non-negotiable legitimacy constraints that are not secondary safeguards but structural conditions. Credible alternative architectures (reinforcement of proximity medicine, vertical integration, precision medicine, patient platforms) are examined and situated.
Who governs alert thresholds? Who bounds the uses of longitudinal trajectories? Who decides on prioritization under insufficient resources? What form of State is capable of governing a predictive vigilance infrastructure without tipping into actuarial surveillance? These four questions cannot receive a purely technical answer. The brief establishes that no sustainable architecture can avoid making them explicit — and that, in the absence of an explicit public mandate, they will be settled by default at the technical level, without transparency.
PREDICARE · predictive medicine · health policy · algorithmic governance · longitudinal vigilance · long-term conditions (ALD) · medical wandering · medical drift · chronicity · actuarial surveillance · political arbitration · public good · partial inalienability of data