Proactive personalized early warning systems: the next frontier in climate–health preparedness

Under the impacts of escalating climate change, the decade 2014–2024 stands as the warmest on record, with global mean surface temperatures approaching 1.5 °C above pre-industrial levels [1]. As a result, populations worldwide face significantly longer and more frequent dangerous heat exposure than in previous decades [2]. The health consequences of hot temperature are severe: heat-related mortality among adults aged 65 years and older in 2023 has risen by 167 % since the 1990s [3], and the 2022 European summer demonstrated that high-income health systems are not immune, with an estimated 62,000 excess deaths attributed to extreme heat [4]. Beyond mortality, heat exacerbates cardiovascular and respiratory diseases, renal failure, and adverse pregnancy outcomes [5]. Outdoor workers, the elderly, individuals with chronic conditions, pregnant women, and urban residents bear a disproportionate share of this burden.

Governments worldwide have established heat-health early warning systems (HHEWS) to tackle the accelerating global warming. In Europe, operational HHEWS now cover the majority of national meteorological services. In North America, the National Weather Service HeatRisk tool provides color-coded risk forecasts. In China, the national high-temperature health risk warning launched in July 2025 generated more than 1.8 billion digital impressions in its first week [6]. These achievements align with the United Nations Early Warnings for All (EW4All) initiative, announced at COP27 in 2022 with US$3.1 billion in implementation investment [7]. Countries with robust multi-hazard early warning systems experience disaster-related death rates nearly six times lower than those without [7].

Yet a critical gap persists between alert issuance and health protection. Most HHEWS calibrate alerts using aggregated, population-level data. As a result, the reasoning of population based HHEWS did not adequately consider the differences and heterogeneity among individuals. A previous study has showed that fewer than 30 % of individuals at elevated heat risk take recommended protective measures after receiving warnings [8]. Population-level alerts often feel irrelevant: the young, healthy office worker dismisses the warning as inapplicable; the elderly hypertensive resident – the very person for whom it is most urgent – receives no message distinguishing her specific vulnerability.

Moreover, the proliferation of concurrent alert streams creates an overwhelming information environment. During the 2024 Chengdu-Chongqing heatwave, over 40 % of local residents deleted alert messages without reading them [6]. Frequent warning issuance leads to public desensitization, reducing perceived urgency over successive events. Warning fatigue is a systems failure that arises when population-level science is transmitted without individual-level translation. The EW4All initiative itself recognizes this gap – Pillar 3 (warning dissemination) presently tracks system coverage metrics but lacks measurable behavioral uptake targets, leaving the critical translation from alert receipt to protective action systematically unaccounted for [7].

Recent years have seen encouraging progress in addressing these limitations. Personalized digital heat warning systems are being developed and piloted across multiple settings. The Ethos (Extreme Heat and Older Persons) system in Australia delivered individualized heat risk alerts to older adults based on physiological modelling; more than 90 % of recipients took protective action, and heat-preparedness scores rose substantially [8]. O’Connor and colleagues proposed a dual-track framework calling for population-level and individual-level warning systems to operate in tandem [9]. These advances motivate our proposal of a Proactive Personalized Early Warning System (PP-EWS).

PP-EWS rests on three integrated components (Figure 1). The first is to develop a suite of individual risk assessment tools. Rather than one-size-fits-all solutions, PP-EWS comprises diverse instruments accommodating different populations, geographic contexts, and use cases. Different age cohorts, sexes, disease histories, occupational groups, and regional populations each require tailored risk models that account for their specific vulnerability profiles. Multiple data sources support these tools: atmospheric temperature and humidity with high spatial resolution, individual-level demographic and health information, medication regimens, and habitual activity patterns. Users who input their individual information could receive targeted vigilance and actionable guidance spanning immediate protective actions and longer-term adaptive strategies.

Figure 1:

Proactive personalized early warning system (PP-EWS) framework.

The second component is risk communication and health literacy. Technical sophistication is wasted on recipients who distrust the source or lack the vocabulary to interpret risk scores. Effective PP-EWS requires sustained health education achieving three objectives. First, it must build public awareness of climate change’s concrete, personal health consequences – not as a distant environmental problem but as an immediate physiological reality for each individual. Second, it must instil a clear understanding that heat-related illness is preventable, and that early warning signals are the principal tool for prevention rather than passive advisories to be scrolled past. Third, it must cultivate what we term proactive risk ownership: the disposition to actively and regularly assess one’s own heat exposure and to take adaptive action ahead of symptoms rather than only in response to them. Integrating PP-EWS education into community health visits, workplace safety programs, and chronic disease management builds the foundation that personalized alerts require.

The third component is multi-sectoral governance. PP-EWS requires collaboration across meteorological services, public health agencies, healthcare systems, urban planning authorities, and education departments. China’s existing infrastructure provides an enabling context: sub-seasonal forecasting with high-resolution temperature predictions; national heat-health surveillance encompassing extensive county-level station networks; and community health worker networks reaching the vast majority of urban and rural populations [6]. Privacy-preserving federated learning architectures enable individual risk modelling at population scale without centralizing sensitive health data.

We call on governments, international organizations, and research funders to commit to rapid development and equitable deployment of proactive personalized heat-health early warning systems. As the global community approaches the 2027 EW4All deadline, universal coverage of early warnings is becoming achievable. But coverage is necessary, not sufficient. PP-EWS redefines success: not the transmission of a signal, but the prevention of a death. Without transformative intervention, heat-related mortality is projected to increase by 370 % by 2050 [10]. The evidence is accumulating. The infrastructure exists. What remains is the political will to treat each person’s heat risk as the specific, computable, and actionable quantity it is, and to build systems that act on that recognition before the next heatwave arrives.