WBE joins AI: improving WBE protocols to support the construction of predictive ML models

Abstract

Wastewater Based Epidemiology (WBE) supports sanitary surveillance enablingan early identification of viral spread. The procedure involves the genome copy(GC) of SARS-CoV-2 capture from sewage samples infected by symptomatic andasymptomatic people. During the pandemic of COVID-19 in Brazil(2020/2021)the WBE studies followed different guidelines still incipient, revealinglow concern for a common agreed-upon procedure. As a result, when compilingthe available WBE data for the training of Artificial Intelligence (AI) models forCOVID-19 number of cases prediction, we found few quantity, obtained throughdifferent adopted procedures and difficult to co-related to extra information. Thelack of a common WBE procedure makes it hard to build useful predictiveMachine Learning (ML) models. In this context, guidelines that link ML andWBE are explored here. We aim at raising an alert highlighting the relevance onthe design of useful strategies to join WBE and IA. The proposal aims atstandardization and consistency without any detriment to the initial objective ofsurveillance. The approach includes processes related to: the definition of samplecollection approach, sampling frequency, related information, physical andtemporal sample characteristics, laboratory methods for genes amplification anddetection and results dissemination.