AI-driven wastewater analysis enhances disease forecasting, expert reveals at BIWWEC 2024

KUCHING (Oct 24): The integration of artificial intelligence (AI) with wastewater-based epidemiological (WBE) data presents a promising strategy for improving public health responses to communicable diseases, said Prof Dr Norhayati Abdullah, secretary of the Asean Learning Network at Universiti Teknologi Malaysia (UTM).

Delivering a presentation on behalf of Dr Arash Zamyadi from Monash University at the Borneo International Water and Wastewater Exhibition and Conference (BIWWEC) 2024 here today, she discussed the topic ‘AI Applications for Forecasting the Spread of Communicable Diseases via Wastewater-Based Epidemiological Data and Beyond’.

Dr Norhayati highlighted the collaborative work of Dr Arash and his team, who use advanced AI techniques to analyse large datasets, particularly in sewage surveillance of Covid-19 RNA fragments, which played a crucial role in pandemic response efforts.

“In their study, a key question arose: While numerous samples have been collected, how do we manage such vast amounts of data effectively using AI?” she said.

By incorporating additional parameters like volumetric flow rate and catchment population, the team could calculate the active Covid-19 cases in communities and predict the spread of communicable diseases.

“This approach allows for the efficient allocation of medical resources, alleviates pressure on healthcare systems, and keeps health authorities informed about the spread of diseases in communities,” Dr Norhayati explained.

She further detailed the role of machine learning models in managing massive amounts of data.

These models, she said, can identify patterns within datasets and make accurate predictions with minimal human intervention.

Their ability to self-improve based on supplied data enhances their value in managing public health crises, she added.

“With such vast amounts of data, manual management becomes impractical. Machine learning allows us to select specific models, identify patterns, and make predictions with minimal human input,” she said.

She said Dr Arash’s work also emphasises the use of big data for disease forecasting, where predictions based on daily sampling can help forecast trends and inform future data collection efforts.

Dr Norhayati concluded that the accuracy of WBE forecasting can be improved through optimal sampling frequency, enabling AI algorithms to identify patterns more effectively.

However, forecasting accuracy may also be influenced by factors such as population mobility, sampling methods, wastewater composition, and environmental conditions.

“Interestingly, too much data can sometimes negatively impact forecast performance.”

While machine learning algorithms are highly effective, Dr Norhayati cautioned that their success depends on the context in which they are used.

“A single algorithm may not be suitable for all datasets or use cases,” she noted, stressing the importance of choosing the right tools for each scenario.