Air pollution and meteorological drivers of hospital medical consumables demand: a hybrid forecasting approach for inventory optimization

Hospitals are vast, complex ecosystems where the flow of people and the flow of supplies must move in perfect sync. When a patient arrives with a sudden illness, the hospital needs the right tools immediately. But predicting exactly when those tools will be needed is a difficult puzzle, especially because the demand for medical supplies is not just driven by the number of patients, but also by the world outside the hospital walls. The air we breathe and the weather we experience can subtly, and sometimes dramatically, change the number of people seeking medical care. For years, hospital administrators have relied on simple guesses based on past months to order their inventory, a method that often leads to either running out of critical items or letting expensive supplies sit unused on shelves. The question of whether the sky and the air itself could be used to predict these needs has remained largely unexplored in the context of hospital logistics.

A team of researchers at the Air Force Medical Center in Beijing decided to investigate this connection by looking at two very different types of medical supplies: interventional guidewires and disinfection supplies. Guidewires are thin, flexible tools used by doctors to navigate blood vessels during complex procedures; they are expensive, high-value items that are typically used in planned surgeries. Disinfection supplies, on the other hand, are the everyday materials used to clean and sterilize equipment and rooms; they are lower in cost but are consumed in massive, unpredictable quantities whenever infection risks rise. The researchers gathered daily records of how many of each item were used over a period spanning from January 2023 to July 2025. They paired this data with detailed records of the weather and air quality in Beijing, including temperature, wind speed, and levels of various pollutants like nitrogen dioxide and particulate matter. Their goal was to see if the environment could explain why the hospital used more or fewer of these items on any given day, and to build a system that could predict these needs more accurately than current methods.

The researchers found that these two types of supplies behave in completely different ways. The demand for guidewires was remarkably stable and predictable, following a steady rhythm that mirrored the hospital's scheduled surgeries. In contrast, the use of disinfection supplies was chaotic and volatile, swinging wildly from day to day. On average, the hospital used about 15.6 guidewires per day, with the highest single-day usage reaching 49 units. The amount of money spent on these wires was significant, averaging over 10,000 yuan daily. However, the usage of disinfection supplies was far more erratic, with an average of 165.2 units per day but with daily peaks that could spike to 3,400 units. This huge gap between the average and the peak suggested that while guidewires were used in a consistent, planned manner, disinfection supplies were subject to sudden, large-scale events, likely driven by outbreaks of infection or urgent changes in hospital protocols.

When the team looked at how the environment influenced these supplies, the results were striking. For the guidewires, the quality of the air and the weather had almost no impact; even on days with heavy pollution, the usage of these surgical tools dropped by less than 2 percent. This confirmed that planned surgeries are not easily disrupted by the weather. However, the story was entirely different for disinfection supplies. On days with poor air quality, the hospital used significantly fewer of these items, with a drop of more than 10 percent. This counterintuitive finding suggests that when the air is bad, people might stay home, or hospital procedures might be delayed, leading to a temporary lull in the need for sterilization. The researchers used a statistical tool called LASSO to sift through dozens of environmental factors and identify the specific drivers for each supply type. They found that five factors, including air quality index and wind speed, were key for disinfection supplies, while eight factors, including temperature and precipitation, influenced the guidewires.

To turn these insights into a practical tool, the researchers built several computer models to forecast future demand. They tested a standard time-series model, a newer model designed to handle seasonal patterns, and a hybrid model that combined the strengths of both. For the guidewires, the hybrid model worked beautifully. It successfully learned the regular, rhythmic patterns of the data and predicted future usage with high accuracy, outperforming the other models by a significant margin. It could anticipate the weekly and monthly cycles of surgical activity with a level of precision that the older models could not match. However, when it came to the chaotic world of disinfection supplies, the results were more modest. While the advanced models reduced the error rate by about 30 percent compared to the basic model, they still struggled to predict the extreme spikes in demand. The data for these supplies was so dominated by random, sudden shocks that no mathematical formula based on history alone could perfectly foresee them.

The study also uncovered how the effects of the environment unfold over time. Using a method that tracks how an exposure today affects outcomes over the next few weeks, they discovered that the relationship between pollution and supply use is not immediate. For disinfection supplies, certain pollutants like nitrogen dioxide showed a strong effect within a few days, while other factors like wind speed had impacts that lingered for weeks. For guidewires, the lag patterns were different, reflecting the longer planning cycles of surgery rather than immediate reactions to the weather. The researchers also examined how multiple pollutants work together. They found that during periods of good air quality, different factors dominated the demand for each supply, but during polluted episodes, nitrogen dioxide became the primary driver for changes in disinfection supply usage. This suggests that the mix of environmental factors matters just as much as the individual elements.

Ultimately, this research highlights that a single strategy cannot manage all hospital supplies. The study suggests that hospitals should treat these two categories differently. For stable, high-value items like guidewires, advanced forecasting models that incorporate environmental data can help optimize inventory, ensuring that expensive tools are available when needed without tying up too much capital in excess stock. For volatile, low-cost items like disinfection supplies, however, relying solely on prediction models is insufficient. The researchers propose that hospitals should instead focus on building robust emergency reserves to handle the unpredictable surges that models cannot catch. By understanding that the sky and the air influence the hospital in different ways for different tools, administrators can move away from one-size-fits-all inventory management toward a more refined, responsive system that keeps patients safe and supplies flowing.