|
|
|
|
|
AI
Predicts Outbreaks of Diarrheal Illness Using Weather Patterns and Disease
Data |
|
Interviewer:
Madeline Roberts, PhD, MPH The recent study noted cyclical patterns in historical diarrheal disease data, meaning certain months exhibited patterns of consistently higher disease rates. This information alone is helpful for epidemiologists; when paired with weather data, model accuracy improved. The authors withheld the 12 most recent months of data, then tested neural network-based model performance for disease rate prediction over that time period. The model with the best predictive performance included: ♦ weather data (precipitation, miniminum and maxiumum temperatures, and El Niño Southern Oscillation phases), ♦ historical diarrheal disease data (data from the same month for the previous two years) ♦ most recent diarrheal disease data (i.e., from the preceding month—data which is not always available in lower-resourced areas). Seasonal-to-subseasonal (S2S) research has origins in weather prediction, where it can provide researchers with two weeks to two months and up to two years advance indication of adverse weather events. Decision-makers can utilize this information to prepare and mobilize resources for predicted events. Environmental epidemiologists can build models which include climate forecast data and disease data and apply S2S methods to predict health threats ranging from malaria to heatwave morbidity and mortality. It amounts to a powerful prediction tool to improve preparation and response times. The authors underscore the value of leaders having three to six months lead time for resource allocation and understanding how near-term future disease threats compare to historical averages. “NN-based S2S early warning systems can be developed to reliably predict [diarrheal] disease risk for various regions with diverse characteristics…Such systems will enable communities to anticipate climate change–related health threats ahead of time, adequately prepare for them, and respond when necessary rather than simply reacting to them.” When accompanied by rapid mobilization, early warning systems for diarrheal illness have high-impact potential in low and middle-income countries where diarrheal disease is a leading cause of mortality. We reached out to Dr. Amir Sapkota, one of the paper's senior authors, with a few questions regarding his recent publication and his broader research program, which explores the intersection of climate change and human health. EpiMonitor: A publicly disseminated weekly/monthly disease forecast, as suggested in your study, is such a powerful tool. Are there diseases or health effects within the US that you think would be well-suited for this kind of forecasting? Sapkota: There are many, ranging from seasonal flu, diarrheal diseases to heat stroke and asthma exacerbations. These tools are designed to enhance public health preparedness and community resilience. EpiMonitor: Seasonal-to-subseasonal (S2S) prediction can range from two weeks to two years. Was there a time span in mind for the predictive models developed in your study? (Perhaps you can speak to any evidence on timing aimed toward optimal public health mobilization) Sapkota: It takes time to mobilize public health resources. So knowing how things are going to be tomorrow or the day after does not give public health practitioners enough time to prepare. Disease outlooks with lead times from a few weeks to a few months are ideal. EpiMonitor: Based on your research experience, are there any policy changes or public health interventions you would like to see that could potentially mitigate some of the health effects of extreme weather events (either within the US or globally)? Sapkota: We need a forward-looking public health system, where we can anticipate these threats ahead of time, prepare for them, and respond to them when the time comes, instead of simply reacting after the fact. EpiMonitor: Less related to your recent publication, your research team partners with the Maryland Public Health Department. Can you talk a bit about your work in that area? Sapkota: As the flagship university in the state of Maryland, we are constantly partnering with our state agencies to address pressing public health issues in Maryland. We are working very closely with the Maryland Department of Health to understand how ongoing climate change is impacting the health of Marylanders and identify the most vulnerable communities. ■ |
|
|
