COVID-19 SURGE PREPAREDNESS WITH AI
New and emerging variants of the SARS-CoV-2 virus continue to pose a threat to the health of populations across the globe. Counting the unique and observable changes in the sequences available till January 2022 shows more than 6,000 mutations have accumulated in the spike gene of the virus.
Initial studies claimed SARS-CoV-2 to be a fast-mutating virus which may make the virus fitter over time. A preprint claimed that the fitness of the SARA-CoV-2 virus is increasing because of the natural phenomenon of purifying selection of the spike protein.
GENOME-BASED APPROACH TO PREDICT COVID-19 SURGE-
- Delta and the Omicron surges highlight the crucial need to use genomic features to predict surges.
- Most of the currently available models predict future trends based on the reported infections and deaths.
- These models do not incorporate features from the virus sequences in a predictive manner.
- The strain flow model plugs this gap by taking a sequence-driven approach to predict future surges using a novel AI pipeline.
- The three-base codons in these sequences were treated as words in the document with each nucleotide base as a letter.
- The best model compressed the viral sequence in 36 dimensions, and the authors proposed that some of these may encode the patterns that make the virus spread faster.
- The models were trained to extract the dimensions that predicted the number of cases in the 17countries with a two-month lead time.
- These models were accurate in predicting the surges during the Delta, omicron and the current surges happening in India.
- The Strainflow model does not predict the actual number of cases but we do get an accurate sense of how sharp the surge might be.
OPEN-DATA INITIATIVES:
These observations when seen in the light of newer waves of infection such as in Delta and the Omicron surges highlight the crucial need to use genomic features to predict surges.
Consortia and open-data initiatives across the globe, such as the Indian SARS-CoV-2 Genomics Consortium (INSACOG) and Global Initiative on Sharing Avian Influenza Data(GISAID) have been instrumental in the identification of new variants. However, most of the inferences from genomic surveillance have so far been retrospective in nature, explaining the past rather than predictive of the future.
Most of the currently available predictive models are variations of standard epidemiological models such as compartmental or agent-based models, which predict the future trends based on the reported infections and deaths. These models do not incorporate features from the virus sequences in a predictive manner.
A recently published model, Strainful, plugs this gap by taking a sequence-driven approach to predict future surges using a novel ARTIFICIAL INTELLIGENCE pipeline.
- It is based on a simple hypothesis-virus sequence and can be treated as documents that can be read like books by Natural Language Understanding (NLU) models. These models can discover the underlying grammar patterns which are casually predictive of future surges.
- The three-base codons in this sequence were treated as words in the document with each nucleotide base as a letter.
- A caveat with NLU models is that these need millions of documents for training.
- Fortunately, GISAID plugged this gap. More than 2.8 million high-quality sequences are processed.
- These included data from 16 countries and India.
- Experiments were done with several NLU models optimised for efficiently learning the grammar of the spike gene.
- The best model compressed the viral sequences in 36 dimensions, technically known as low-dimensional embeddings.
- Each of these 36 dimensions is a different cocktail mix of codon-level relationships.
- These 36 cocktail mixtures may encode the patterns that make the virus spread faster.
CLEAR TEMPORAL PATTERNS:
- The models are trained to extract the dimensions that predicted the number of cases in the 17 countries with a two-month lead time.
- Surprisingly, these models were accurate in predicting the surges during the Delta, Omicron and current situations.
- The features of these cocktails mixes were examined and found that the diversity of some of these, technically captured as a quantity known as entropy showed clear temporal patterns predictive of surges.
- The entropy was seen to start dipping sharply up to two months before the cases surge.
- This might be a biological plausible because the virus strain evolution takes an explore-exploit pattern, exploring the possible combinations of mutations and then exploiting some to establish itself as the dominant strain.
- Although the Strainflow approach does not predict the actual number, the accurate sense of how sharp the surge can be known.
- The Strainflow model has proven to be effective for predicting whether there is a likely surge with a 2-month lead time, which could help the healthcare systems to be prepared.
DE-NOVO APPROACH:
A key feature of Strainflow was its data-driven, de-novo approach without the need for expert understanding.
The strongest learning from this exercise has been the power of open data and interdisciplinary thinking. Strainflow is just one piece in the puzzle for solving the surge-preparedness of COVID-19 which could lay the foundation for general infectious disease preparedness.
Let's imagine the future as multiple signals such as Strainflow, traditional epidemiological models, testing results and non-conventional sources such as mobility, demographics, and social media signals feeding into a unified model for preparedness.
This is exactly our current endeavour, with projects geared towards district level preparedness in collaboration with ICMR and understanding the relationship between mobility patterns and strain emergence in collaboration with Meta(formerly Facebook) as a part of the data for the good initiative.
