Dieseases like Diarrhea are one of the leading causes of death in countries with lack of clean drinking water and lack of functioning sanitation. And when there are harsher weathers like flooding and drought due to climate -change, then the risk of outbreaks of deadly water-borne dieseases increases too. Hopefully, that may change with the help of an AI-based model.

Amir Sapkota, a senior author from University of Maryland’s School of Public Health (UMD SPH) has now developed an AI-driven model that may predict deadly outbreaks with greater accuracy not weeks but even months before the actual outbreak.

AI based Climate projection–based warning systems

The authors argue that there is a strong need for a AI embedded warning system instead of the currently available weather or climate-based projection systems as they can give indications of the actual outbreak only only 7 – 10 days before. But by using an AI-based warning system built on extensive datasets, the system may warn about the actual harsher outbreak due to climate change like flooding and drought 50 – 100 years in advance.

A team from multiple disciplines collected extensive data related to temperature, precipitation, previous disease rates, and El Niño Southern Oscillation phases and tried to predict diarrhoeal disease rates that happened in Nepal (2002 -2014), (2000–2015) and in Taiwan (2008–2019) and found that their Neural Network based early warning system excellent response by analyzing data from the past.

“Knowing expected disease burden weeks to months ahead of time provides public health practitioners crucial time to prepare. This way, they are better prepared to respond when the time comes” excerpt from the conversation of Sapkota with Phys.org on the same study.

The authors used four different models for predicting the previous diarrhoreal disease rates of Nepal, Taiwan and Vietnam and found that the Model 4 gave the most accurate prediction. But according to the authors, even Model 3 is able to provide reasonable insights about the upcoming disaster which could cause outbreak of Diarrhea.

Model 1 : – Included Weather Variables only.
Model 2 : – Included Historical Data only.
Model 3 : – Included both weather and historical data.
Model 4 : – Combination of Model 3 along with recent diesease data.

You can read the study in detail here.