Complex Analysis of COVID Data? AI to the Rescue!

"The human brain becomes pretty quickly overwhelmed by the permutations and combinations of these things [a range of data gleaned by studies to determine vital details about SARS-CoV-2]."

"But when you put AI into it, it can run countless simulations and can home in on important stuff very quickly and effectively."

Bill Kapogiannis. data scientist, National Institutes of Health, U.S.

Data scientists are scrambling to enlist the power of artificial intelligence in unlocking some of the novel coronavirus secrets to better understand how it functions and how it can be stopped. Some of the biggest mysteries are awaiting being unravelled by AI with the proper inputs; questions such as how it is that this disease appears so at variance in its effect on children, as opposed to adults; what it is about the virus that creates "superspreaders" of some people, even as others fail entirely to transmit the virus to others. The questions are endless, the answers elusive.

 

And often entirely unexpected results can be gleaned from coronavirus data fed into a machine. Data scientist A.J.Venkatakrishnan fed genetic data from 10,967 samples of the virus into a computer hoping the artificial intelligence might highlight possible weaknesses which could be exploited in developing therapies, so when the program reported that the new virus appeared to have a snippet of DNA code -- "RRARSAS" -- distinct from the predecessor coronavirus, he was dumbstruck at the sequence that mimics a protein to help the human body regulate salt and fluid balance. 

As director of scientific research and partnerships at AI startup Nference, Dr.Venkatakrishnan wondered whether this mutation might permit the virus to behave in the manner of a Trojan horse -- possibly explaining its high infection and transmission rates; perhaps even why it is that people with cardiovascular disease experienced more severe cases of infection given that sodium can impact blood pressure. "It was a surprise, completely accidental. The machine just spotted that", he pointed out.

The pandemic generates millions of gigabytes of data day in and day out in medical records, along with other information relating to infected patients; bloodwork results, age, race, genetic testing, attempted interventions; outcomes. Ten months into the outbreak, scientists are now beginning to make connections in the mass of letters and numbers, with the aid of artificial intelligence, leading to new theories about the virus and potentially -- goes the hope --  how it can be stopped.

 

Machine learning can aid epidemiologists’ understanding of the coronavirus pandemic, but not all comparisons will be useful if standards differ between countries. Pitinan Piyavatin / Alamy Stock Photo.

 

The human brain is a natural work of art, complex and capable of processing much information, but it is taxed beyond its limit when faced with examining, interpreting, analyzing masses of data reflected by what is being collected in the hopes of discovering some key secrets to diminishing the prevalence and affect of this singular virus. Machines, on the other hand are capable of finding subtle patterns in huge data amounts, leading to their deployment against COVID-19 in ways beyond imagination. 

A modelling lab is running large-scale simulations on the effects of travel restrictions and social distancing on infection rates, through Northwestern University data technicians, with the use of AI. The Argonne National Laboratory of the U.S.Energy Department is making use of AI to hone in on the most promising molecules for lab testing as possible treatments. AI aids as well, in countering coronavirus misinformation in Arabic, in Egypt.

Had the novel coronavirus surfaced twenty years earlier -- according to Jason Moore, director of the Penn Institute for Biomedical Information at the University of Pennsylvania, helping to amass an international COVID-19 data consortium -- the world might have been doomed. "But I think we have a fighting chance today because of AI and machine learning", he now states. 

 

A computer sorting through medical records in April confirmed a lack of smell and taste, earlier anecdotally reported, as one of the earliest symptoms of infections, leading the Centers for Disease Control and Prevention to add anosmia to its symptoms list. It was discovered in June through a deep dive into the records of close to eight thousand patients that while a small fraction only had obvious and catastrophic blood clots, almost all of the patients showed worrying changes in their blood coagulation. 

 

Researchers piggy-backed on Dr.Venkatakrishnan's findings of the aberrant genetic sequence n an effort to understand how the virus binds to cells and to make use of that knowledge in the development of drugs aiming to reduce transmission. 

 

Dr.Venkatakrishnan and colleagues in a follow-up paper published in September reported a computer analysis showed this evolutionary "tinkering" by coronavirus, appearing to have made it identified as a 'friend' rather than the 'foe' it truly is, to the human immune system, targeting the lungs and blood vessels, constituting a finding providing new insights about clinical symptoms seen by doctors at hospitals. 

 

Early progress in AI has been promising even as critics are concerned that efforts to harness COVID data have been disjointed and frustratingly slow.

One of the largest challenges to overcome in the medical-investigative-scientific field has been the reluctance among business interests and academic researchers, medical societies and private companies to pool and share information to benefit everyone. A number of efforts have been launched to see whether such barriers to sharing can be overcome in the interests of establishing a giant database of health records and other data. Among them, a $20-million, four-year project by the National Institutes of Health led by data scientist Bill Kapogiannis who is optimistic the pace of science will accelerate based on computing power.

Coronavirus cases for most affected countries, (except China) since day when cases>=90, as of March 21, 2020.   KDNuggetsNews