The recent surge in JN.1 variants of Covid-19 have added to the worries in countries across the world. While the WHO has notified it as ‘Variant of Interest’, the fast-paced spread of JN.1 has raised concerns about the fresh wave of the pandemic. However, a latest discovery can prove crucial in potentially safeguarding us from future eventualities.
A recent study has discovered that an Artificial Intelligence model can predict which variants of SARS-CoV2 can bring a fresh wave of infection. The joint study was conducted by Massachusetts Institute of Technology (MIT) in the US in collaboration with the Hebrew University-Hadassah Medical School in Israel.
The AI model has the ability to detect up to 73% of the Covid-19 variants in each country around the world that will cause a minimum 1,000 cases in every 10 Lakh population in the span of three months, following a one-week observation period. Whereas it can detect up to 80% variants after two weeks.
For the purpose of the study, the research teams analysed 9 million SARS-COV-2 genetic sequences found in 30 countries throughout March 19, 2022. Among these countries, the United States has the highest number with 2.9 million reported sequences, and Luxembourg has the smallest number with 25,000 reported sequences. These genetic sequences were collected by Global Initiative on Sharing Avian Influenza Data (GISAID). The researchers then combined the data with various other factors including vaccination rate, infection rate, etc.
GISAID is an initiative that promotes rapid sharing of data from priority pathogens including influenza, hCoV-19, respiratory syncytial virus (RSV), hMpxV as well as arboviruses including chikungunya, dengue and zika
Based on the patterns emerging from the study, the team has built a risk assessment model powered by Machine Learning. This AI algorithm helps in studying vast sets of past historical data and make predictions. Their findings have been published in a PSNAS Journal.
As per the study findings, the most prominent factor that influences a particular variant’s infectiousness is the early trajectory of the infections it causes, its spike mutations, and how different its mutations are from those of the most dominant variant during the observation period.
The researchers underlined that although pre-existing models predict the dynamics and trends of viral transmission, they fall short of predicting variant-specific spread. The authors mentioned that their study leverages variant-specific genetic data coupled with epidemiological information. This can provide improved early signals and predict the future spread of newly detected variants.
They also highlighted that this approach is not only limited to SARS-COV-2. This model can potentially be used to predict the spread of other respiratory illness-causing viruses, like- Influenza, Avian Flu and other Coronaviruses.
The authors wrote in their study, "These results support the hypothesis that the infectious new variants are those that acquire enough mutations which either can lead to reinfections or enable targeting new subgroups of the population that were naturally immune to previous variants."
In the past, MIT researchers had developed an AI model that could determine whether a person had Covid-19 or not with the sound of their cough. The researchers then claimed to have achieved striking 97.1% accuracy in their study.
Since the onset of the pandemic in 2020, researchers have been working on developing various AI-based solutions to detect/diagnose and in some cases even predict Covid-19 disease.
For example, Covid-Net, developed by researchers at the University of Waterloo and DarwinAI, is a deep learning model designed for the detection of COVID-19 from chest X-ray images. DeepMind's AlphaFold, initially created for protein folding, has been used to predict the 3D structures of proteins associated with the SARS-CoV-2 virus.
CORD-19 is a large dataset that has been used by researchers worldwide to develop AI models for information extraction, summarization, and other tasks related to COVID-19 research.
Some of the AI models have been employed for forecasting the spread of the virus, helping public health officials and policymakers make informed decisions. These models analyse a variety of data, including infection rates, mobility patterns, and other factors, to predict the future course of the pandemic.
The College of Healthcare Information Management Executives (CHIME) is an executive organization dedicated to serving senior digital health leaders. CHIME includes more than 5,000 members in 56 countries and two US territories and partners with over 150 healthcare IT businesses and professional services firms. CHIME enables its members and business partners to collaborate, exchange ideas, develop professionally and advocate the effective use of information management to improve the health and care throughout the communities they serve. CHIME's members are chief information officers (CIOs), chief medical information officers (CMIOs), chief nursing information officers (CNIOs), chief innovation officers (CIOs), chief digital officers (CDOs), and other senior healthcare leaders. The CHIME India Chapter became the first international chapter outside North America in 2016 and is now a community of over 70+ members in India. For more information, please visit www.chimecentral.org