PUBLIC HEALTH · ANTICIPATION
A national health authority was conducting a preparedness exercise for a scenario involving a moderate resurgence of a respiratory disease with community transmission, comparable in intensity to severe seasonal influenza. Recommended, non-mandatory protective measures. Post-COVID context marked by structural health-measure fatigue.
The question was not epidemiological : existing models were sufficient to project viral circulation. It was behavioral. How will a population that has already lived through a crisis react if the risk returns without returning to the same level of severity?
THE PROBLEM
A population that has lived through a health crisis does not meet the next signal as a blank slate. It carries experience, memories, habits acquired then abandoned, trust or distrust in institutions, a perception of the social and economic costs of measures, and fatigue with certain practices.
The question posed by the national health authority was therefore not “ will people follow the recommendations ? ”. It was : what would effective adoption of recommended protective behaviors be in a context of post-COVID institutional distrust, and how would institutional communication strategies influence those behaviors ?
The health authority used our system to test four institutional communication strategies on a differentiated synthetic population, with behavioral trajectories projected over twelve weeks.
WHAT THE SCENARIO ASSUMES
WHAT DIFFERS FOR EACH GROUP
A VARIABLE : MEMORY OF RISK
STATED FATIGUE, ACTUAL FATIGUE
Individuals have learned what they consider effective, what they now refuse, and what they are willing to resume temporarily. What the case documents is not a level of weariness, but the gap between what is stated and what is actually reactivated.
A MECHANISM : THE PERSONAL THRESHOLD
Conceptual diagram. No threshold value is measured : the diagram only shows that the same rise in the signal does not trigger the same behaviors in everyone.
What the case documents is the dispersion of these shifts : in 5 territories out of 10, part of the population adapts its behavior 8 to 14 days before any official measure, while others change nothing under the same conditions.
WHAT IS TESTED
What varies is not the recommended health content, but the architecture of public messaging : who speaks, from what level, and in what register. No additional health measure is simulated.
4 200 synthetic adults were interviewed individually on the 4 strategies tested. Dynamic agents conducted behavioral interviews, following up with each adult on tipping points : perception of the threat, adoption of protective behaviors, resistance to collective arrangements, mobilizable health fatigue. Trajectories were projected over 12 weeks, modeling territorial and intergenerational dynamics.
SIMULATED POPULATION
The reconstructed population covers 4,8 million European adults, calibrated on public INSEE data, Santé publique France data and European post-COVID studies of health behavior.
It is structured into 9 profiles of relationship to health risk, crossing age, epidemiological vulnerability, occupational exposure, institutional trust and engagement in post-COVID social movements. This heterogeneity determines what shifts and what does not.
AGE AND VULNERABILITY
stated epidemiological position changes how the same signal is interpreted
OCCUPATIONAL EXPOSURE
some work situations make adaptation difficult, regardless of intention
INSTITUTIONAL TRUST
it determines what is heard in a national message
POST-COVID ENGAGEMENT
involvement in social movements arising from the previous crisis structures reception
MOBILIZABLE FATIGUE
willingness to readopt an already-tested practice, distinct from stated fatigue
SPONTANEOUS SELF-REGULATION
behaviors adopted before any official announcement
TERRITORIAL EMBEDDEDNESS
local and national voices do not carry in the same way
INTERMEDIATE TYPOLOGIES
nine profiles of relationship to risk, crossing these dimensions rather than opposing them
These configurations reveal part of the population's heterogeneity. The simulation operates on synthetic individuals, not a handful of persona types : these are not representative figures, but distinct situations, constraints and options.
REACTIONS
THE POPULATION DOES NOT WAIT FOR OFFICIAL ANNOUNCEMENTS TO ACT
The simulation identifies a structural post-COVID dynamic: in 5 territories out of 10, behavioral self-regulation — voluntary mask wearing, interpersonal distancing, spontaneous remote work — precedes official measures by 8 to 14 days. The population has behavioral capital that it mobilizes without waiting for instructions.
VALIDATING WHAT IS ALREADY BEING DONE CARRIES MORE WEIGHT THAN PRESCRIBING WHAT SHOULD BE DONE
A register that gets ahead of behavior generates 34 % effective adoption of protective behaviors. A register validating observed self-regulation — “more and more of you are wearing masks on public transport: this behavior makes a collective difference” — generates 76 %. The expected behavioral content is identical in both messages.
STATED HEALTH-MEASURE FATIGUE IS MISLEADING
68 % of adults report high fatigue with health measures; only 42 % show effective resistance to recommended protective behaviors in behavioral simulation. This asymmetry between stated and behavioral responses is structural, and means setups should not be calibrated to opinion surveys.
THE PREFECTURAL CASCADE CARRIES MORE WEIGHT THAN A FIRM NATIONAL MESSAGE
Firm national communication delivered by the minister generates 42 % average adoption. The same communication rolled out as a territorial prefectural cascade, with each prefect communicating the setup in their department alongside local elected officials, achieves 68 %. The gap is greatest in territories most distrustful of national messaging.
34 % versus 76 % adoption depending on the register, 42 % versus 68 % depending on the channel : these gaps concern the same population, the same health scenario and the same expected behavioral content.
RECOMMENDATION AND BEHAVIOR
Between the two sit risk perception, trust, constraint, habit and observation of others. This is the layer explored by the simulation : message, interpretation, individual decision, behavior.
THE MESSAGE
A public recommendation, identical in its expected behavioral content.
THE INTERPRETATION
It depends on past experience, trust in the sender and what the individual observes around them.
THE INDIVIDUAL DECISION
A register that gets ahead of behavior — “you must wear a mask” — produces 34 % effective adoption. A register validating observed self-regulation produces 76 %.
THE BEHAVIOR
The expected content is the same in both messages. The effect is opposite. A recommendation is not a behavior.
THREE MOMENTS
Behavioral reading across the three moments of the tested scenario. This diagram describes neither the circulation of a pathogen nor an epidemiological curve.
It makes visible a dynamic documented by the case : when self-regulation becomes observable — voluntary mask wearing, distancing, spontaneous remote work — it itself becomes a signal. A message validating this movement achieves 76 % adoption ; a message that gets ahead of it, 34 %.
MODULAR ADAPTATION
Nobody shifts from a “ protected ” state to an “ unprotected ” state. The documented adaptations are separable : a precaution on public transport, occasional remote work, particular care around a relative.
WHAT THE BINARY MODEL ASSUMES
A population split between those who follow the recommendations and those who do not.
WHAT THE SIMULATION SHOWS
Observed behaviors are partial: voluntary mask wearing in certain situations, interpersonal distancing, occasional spontaneous remote work.
WHAT THIS CHANGES
The same individual can reactivate one practice and reject another. Adoption is not measured as a state, but as a set of separable behaviors.
WHAT THIS IMPLIES FOR THE SETUP
A uniform message addresses a population that adjusts in fragments. A system for listening to actual behavior becomes a condition of calibration.
COMPARISON
Qualitative comparison derived from the simulated strategies. The levels reflect documented gaps : 34 % versus 76 % effective adoption depending on the register, 42 % versus 68 % depending on the channel. No strategy is cost-free : a firm register produces a clear signal but feeds reactance in distrustful territories ; an informative approach limits social disruption but relies on unevenly distributed self-regulation ; the territorial cascade is more demanding to operate.
THE MOST ROBUST
Territorial prefectural cascade register validating observed self-regulation system for listening to actual behavior
THE MOST FRAGILE
Firm, prescriptive national messaging no territorial relay or observation of behaviors already adopted
DECISION
PROJECTION, THEN OBSERVATION
The health authority integrated the results into its national preparedness setup for an epidemic-resurgence scenario. The validated setup retains the territorial prefectural cascade as the primary channel, a register validating observed self-regulation, a system for listening to actual behavior through prefects to adjust messages in real time, and dedicated training for prefects in this register.
The setup did not need to be activated during a real epidemic situation over the observation period. It was, however, used during two occasional health alerts : regional food contamination and an air-pollution alert. The results were consistent with projections, with adoption of recommendations 32 points higher than the conventional setups used during previous alerts.
The methodology was referenced in the national health-crisis preparedness plan. These values are those documented by the commissioning organization.
TAKEAWAY
The population did not react to the announced level of risk. It reacted to what it had already experienced, what it observed around it, and how public messaging positioned itself relative to a movement already under way. In 5 territories out of 10, adaptation precedes the announcement by 8 to 14 days : the institutional message does not initiate behavior, it confirms or opposes it.
Two consequences for public decision-making. Messages that validate observed behaviors achieve higher adoption than those that prescribe them — 76 % versus 34 % — which requires real-time observation rather than communication written in advance. And the asymmetry between stated fatigue (68 %) and behavioral fatigue (42 %) must be integrated into calibration : a setup sized to opinion surveys underestimates the population’s mobilizable capacities.
POSSIBLE FUTURES
A
REACTIVATE STRONGLY
Firm, prescriptive national communication close to the register used in the previous crisis
B
INFORM AND EMPOWER
Limited communication leaving individuals to make their own trade-offs
C
TERRITORIALIZE AND VALIDATE
Territorial prefectural cascade, register validating observed self-regulation, listening to actual behavior
METHOD
4,8 million reconstructed synthetic adults, 9 profiles of relationship to health risk, 4 200 individuals interviewed individually across 4 communication strategies, and behavioral trajectories projected over 12 weeks with modeling of territorial and intergenerational dynamics.
This case adds a capability to the library : simulating a population that has already experienced an event. Past event, learning, new event, modified reaction. The agents are not static respondents ; they carry memory, fatigue and thresholds that shift the response to the next signal.
This case comes from a preparedness exercise conducted for a national health authority. The commissioning organization is not named, no personally identifiable data entered the system, and the detailed results remain its property. The comparisons between strategies published here are qualitative ; the quantitative values cited are those documented by the commissioning organization.
The simulation explores the behavioral reactions of synthetic populations to different epidemic-resurgence and public-response scenarios. It is neither epidemiological modeling, nor a health prediction, nor medical advice, nor an evaluation of the clinical or public-health effectiveness of a measure. The public decision belongs to the decision-maker.
Your next decision
Describe your need. We can point you to the right level of support.
State your decision. See the future it produces.
