When the pandemic began, there was a lot we didn't know about the SARS-CoV-2. The virus was constantly changing its game, constantly challenging healthcare workers, medical researchers, scientists and hapless people worldover. But 15 months into the pandemic and we have all come a long way. Like any typical pandemic, the ongoing pandemic too follows certain behaviours and patterns. It appears in waves, there are multiple strains present across the world and new symptoms are being reported as well. Last year, India, like every other country, was on high alert. Lockdowns have become a part of life now, and COVID-appropriate behaviour is the first thing doctors lead medical bulletins with. 

As India continues to battle a brutal and devastating second wave, health experts and policymakers realise that if they need to beat the virus, staying ahead of it is absolutely imperative. Last week, Principal Scientific Advisor to the Prime Minister Dr. KV Vijayaraghavan said that a third wave in India is inevitable, and that India must brace itself with all countermeasures as there is no saying when this can happen or what the effects are going to be. But this may not be altogether true - a certain degree of preparedness for a possible third wave can be achieved by leveraging technologies, specifically AI. 

Researchers Experimenting with AI To Predict Possible Third Wave 

Studying pandemics and disease outbreaks isn't new; but advances in AI can provide a new dimension to disease tracking. Applying advanced algorithmic calculations and predictive modelling can give a fairly good idea of the things to come, allowing governments, healthcare workers and businesses to take appropriate anticipatory measures. Scientists at IIT Kanpur, following a mathematical study, estimate that India will witness a third wave in October 2021. In South Africa, an AI-based algorithm designed by the University of Witwatersrand in partnership with the National Research Foundation’s iThemba LABS; the Provincial Government of Gauteng; and York University in Canada, shows that there is a low risk of a third infection wave of the COVID pandemic in all provinces of South Africa. The AI-powered early detection system functions by predicting future daily confirmed cases, based on historical data from South Africa’s past infection history, that includes features such as mobility indices, stringency indices and epidemiological parameters. The AI-based algorithm works in parallel and supports the data of an existing algorithm that is based on more classical analytics. Both algorithms work independently and are updated daily. These two independent algorithms add to the robustness of algorithm's predictive capacity, stated a report. 

The SARS Effect on Data Driven Approaches to COVID19 Management 

Certain Asian nations like China and Taiwan have made major strides in infectious disease preparedness, based on prior experience with SARS in 2003-04. They have documented outbreaks using data and analytics, and have fashioned these findings to manage the spread of COVID19. In this research paper, China implemented unprecedented restrictive measures from January 23, 2020 after the outbreak that began in Wuhan in December 2019 had claimed 1,000 lives and impacted more than 50,000. Cities were quarantined, the ongoing national holiday extended, public spaces closed and rigourous temperature checks were initiated. But even then, the Chinese authorities weren't sure how effective these steps were or how long they should be enforced. They used the Susceptible Exposed Infectious Removed (SEIR) model to derive the curve of the epidemic. Then, they integrated population migration data before and after January 23rd 2020 along with the updated epidemiological data. An AI model, trained on SARS data, was used to then predict how the epidemic would forge its path ahead. This led the authorities to understand that the epidemic in China would peak by late February, and gradually decline by end April. Even a 5-day delay in implementation would increase the cases in mainland China by threefold. In addition, had they lifted the quarantine in Hubei - the epicentre of the outbreak - a second peak would have hit the area in March and extended the epidemic to late April. These results were further corroborated by a machine learning prediction. Even as COVID19 started impacting other countries like Italy, USA, India and Brazil, China had slowly started moving towards a semblance of normalcy by mid 2020, resuming key business activities. 

Major Challenges To Predictive Modelling 

In this paper authored by Eric Paternoster and Dr Suman De of Infosys, there is sufficient material to show that an AI-based predictive model will be essential to managing a third wave of COVID19. This continuous learning model would be based on real-time and robust data that would aid governments and policymakers with guiding economies to a phased reopening. Like any AI expert will tell you, data lies at the core of sound AI applications. The biggest challenge that countries face is gathering quality data. Ideally, to build robust pandemic prediction models, data like fatality rates, R-values, virus transmission rates and herd immunity levels need to be collated. In additional, quality observational data is essential but hard to gather in countries where mass testing protocols are poor. What's being seen during this second wave in India atleast is that even the younger, and likely healthier population band is also getting adversely affected, unlike the first wave where those aged, and with comorbidities were most vulnerable. Doctors say that in many cases, COVID19 exposes underlying health conditions even in young people, and this has also largely overwhelmed health systems. There aren't viable data repositories of existing health conditions of large populations, and this can cause seriously impede data integrity. Farr's Law states that epidemics are symmetrical and scale uniformly, but the gross inequities in the way the disease is affecting different countries seems to indicate otherwise. Population variances, COVID19 testing propensity, access to critical care facilities are all key determinants of how different populations are dealing with the disease. This means, every country needs to focus on data parameters specific to its own societal structure and aggressively gather this data to build competent predictive models. 

The UK's NHS is banking on technology like AI to restore disrupted workflows, and its time other healthcare systems take a cue from how NHS is doing this. If some of the above challenges are overcome, atleast partially, India can bank on limited but accurate datasets to build predictive models that can aid governments with preparations for a third wave. These models can help hospitals plan for contingencies better like management of essential supplies, assigning emergency care workers on priority, manage non-COVID19 healthcare emergencies with adequate staff and supplies, plan vaccination programmes and ensure supply of life saving drugs. 

Too many lives have been lost since the start of the pandemic, and India's healthcare system is under a great deal of duress. If we hope to conquer a debilitating third wave, a data driven approach is imperative.