A Global Dataset of Cost-of-illness for Climate-sensitive Diseases through Automated Literature Extraction
This paper presents a globally comparable, harmonized dataset of cost-of-illness for over 80 climate-sensitive diseases, derived from nearly 900 publications using automated large language model extraction and a three-tier confidence scoring framework to facilitate robust cross-country economic modeling and policy design.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
When the weather turns extreme, the human body pays a price. Heatwaves can trigger heart attacks, rising temperatures can spread mosquito-borne illnesses, and shifting seasons can worsen asthma. For decades, scientists have tracked how many people get sick or die from these climate-driven health crises. But for policymakers and economists, knowing the number of sick people is only half the story. The other half is the cost: the money spent on hospital beds, medicines, and lost workdays. Until now, calculating this financial burden has been a patchwork effort. Researchers often rely on abstract estimates of the value of a human life, a method that critics say feels disconnected from the actual bills families and governments face. What is needed is a clear, global ledger of the real money spent to treat climate-sensitive diseases, but gathering this data has been like trying to assemble a massive puzzle where every piece is in a different language, printed on different paper, and scattered across thousands of books.
A team of researchers has now built that ledger. By teaching computers to read and understand thousands of scientific papers, they have created the first global dataset that translates the economic cost of climate-related illnesses into a single, comparable language. The project, led by scholars from Tsinghua University and other institutions, focused on over eighty diseases that are known to be influenced by the climate, ranging from dengue fever and malaria to stroke and diabetes. Instead of manually reading every study, which would take a human lifetime, the team used advanced artificial intelligence to scan nearly nine hundred peer-reviewed publications. These papers, published between 2015 and 2025, contained the actual financial records of how much it costs to treat patients in different parts of the world. The computer extracted twenty-seven thousand specific cost observations, converting every currency from rupees to dollars and adjusting every price to reflect the value of money in 2024. This process turned a chaotic collection of local reports into a unified map of global health expenses.
The resulting dataset reveals a world of stark contrasts. The researchers found that the cost of treating the same disease can vary wildly depending on where a patient lives. For instance, the annual cost of treating chronic obstructive pulmonary disease, a lung condition often linked to air quality and temperature, was found to be as low as three hundred and fifty dollars per patient in one country, while in another, it exceeded twenty-four thousand dollars. These differences are not just about how expensive hospitals are; they reflect how different countries count their costs. Some studies include the money lost when a patient cannot work, while others only count the hospital bill. The new dataset helps untangle these differences by categorizing costs into clear groups, such as money spent on hospital stays, medicines, or transportation, and by noting whether the patient paid out of their own pocket or through insurance.
However, the map also shows where the data is missing. The researchers discovered that most of the financial evidence comes from wealthy nations and a few large countries like China, India, and the United States. Vast regions, particularly in Africa and parts of South America, remain largely blank on this economic map. This gap is significant because it means we cannot fully understand the financial impact of climate change on the world's most vulnerable populations. The study also highlighted that while we have plenty of data on common chronic conditions like diabetes and heart disease, there is very little information on the costs of mental health issues or neurological disorders linked to the climate. The team did not just collect numbers; they also built a system to rate the reliability of each piece of information. They assigned confidence scores based on how many studies supported a finding, how consistent the numbers were, and how trustworthy the original data sources were. This allows anyone using the data to know which figures are solid and which are based on thin evidence.
The work does not claim to have solved the problem of measuring climate health costs, but it provides the first solid foundation for doing so. The researchers acknowledge that their automated system, while powerful, is not perfect. Sometimes the computer might miss a specific detail in a complex table or misinterpret a method described in the text. To address this, they designed the dataset to be open and updatable, inviting other scientists to add new findings and correct errors as they arise. The ultimate goal is to move beyond abstract projections and give governments and health organizations a concrete tool for decision-making. With this dataset, leaders can finally compare the cost of treating a heat-related illness in one country against another, or weigh the financial benefits of climate action against the cost of inaction. By turning fragmented reports into a clear, global picture, this research offers a new way to see the true price of a changing climate, grounded in the reality of actual medical bills rather than theoretical estimates.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.