The sudden arrival of the coronavirus across the world from its source in China in a matter of a few months was a surprise to everybody. But it was no surprise to those you could term as being “in the know” since they had long predicted such an event. So whilst the world was focussed upon the subject of climate change, just as significant but less speedy, this suddenly caught the world off guard. Back in 2018 at Davos, Sylvie Briand, an infectious disease specialist at the World Health Organisation, said that the next great pandemic was coming, that we were more vulnerable than ever and we had no way to stop it. In 2019 at a health conference Bill Gates declared that a virus could easily appear that could kill 30 million people. But at the time these predictions fell on deaf ears with the world focussed upon climate change.
So if were are a digital human and the coronavirus is a digital virus why were we not ready for it in terms of using Artificial Intelligence (AI). The vision I have is why as soon as the virus genome arrived in my email inbox from China could I not just copy it into my AI based vaccine designer and out the other side popped the genome for the required vaccine. Send this off the vaccine factory and within hours we could have million shots of vaccine. Job done. Pandemic stopped before it started.
Well the fact is in 2020 we did not have this vaccine designer. But I would bet money on us having one by 2030 if not earlier. The costs both in terms of human life and economic loss are such that the investments will now be made to build this relatively simple program to process from a virus to a vaccine. But it is not just the writing of a new program. We need to go further and invest in a whole new more powerful computer architecture called Quantum Computing. We cannot afford not to do so. But could we have been better prepared now in 2020. Possibly since the seeds of artificial intelligence have already been sown within the existing digital architecture and we could have dedicated all the world's computers to the task to acquire the power needed. So what needed to be done.
Well I am not in the vaccine making world either academic or pharmaceutical but I do know if Google can decide which paint I should purchase to paint my fences then deciding on which vaccine to treat a virus should be straightforward. Google have my painting buying history over the last few years. They can see when I have changed brands. They can see which brand I have settled on buying that best suits my fences. They can predict what I will purchase next. They are likely to be right. They have masses of historical data and my purchasing outcomes. In fact they can do this for all my purchases not just fence paint. They have applied artificial intelligence to retailing. So why not apply it to vaccine creation?
So this is artificial intelligence (AI) stepping back over masses of previous stored historical data and at the same time being able to see what the outcomes of this data were at different times. So you are recording the data before and behind the final result. But you have the opportunity to see what the result was in the real world based upon this stored data. Having established this link between the data and the result you have established what has been termed an artificial intelligence algorithm. In fact the algorithm is just representing in computer code how you may do it manually. Stepping through the data deciding where certain patterns in the data are say more common so more likely to result in the outcome than other patterns that are less common. So how could we have made more use of this capability?
So the use of vaccines has been around for some time. Edward Jenner (1749 – 1823) developed the smallpox vaccine. Jenner noted that milkmaids were generally immune to smallpox so he scraped pus from the cowpox blisters of a milkmaid who had caught cowpox and inoculated an 8 year old boy with it. He then injected the boy with some virus, but only a little bit, taken from someone with smallpox. No smallpox disease triggered. He then tested 23 additional subjects. The vaccine based upon cowpox worked. Good practical non digital chemical experimentation only involving putting an 8 year old boy at risk. No Health and Safety rules to worry about in those days.
Now it’s a simple step to digitally establish the digital genome for smallpox, although only a few labs retain this dangerous virus, and then to digitally establish the genome for the cowpox vaccine used. Do not get confused by the term genome since all it means is a complete digital map of something. Similar to a Google map but of biological things. Both the Google map and the genome are just millions and millions of bits, like light bulbs, either switched on or off. You can create an artificial intelligence (AI) algorithm that runs through the smallpox genome to see how the patterns in the bits (on or off) within it link to the patterns within the cowpox vaccine bits. But this is just one disease and one vaccine.
Now share a vision of taking every disease we know and every vaccine we have ever created to successfully stop the spread of the disease. So thousands of diseases and thousands of vaccines. Now we all know that diseases and their viruses come in different flavours. These flavours have been categorised. By categorising them together you can establish in more detail the patterns that link the virus to the vaccine. We know coronavirus is of one particular category, call it spikey, so we have to group all these related types together since they are related in the way they have evolved by mutation. So if it is all so easy why has it not been done?
Well it is not that easy. Think of it down at the atom level within these viruses and their possible vaccines. Firstly there is a lot of data just for one little virus. There is also a lot of data for one little vaccine. But it is never that simple because we have a third dataset to consider. The human genome. This human genome introduces the body’s own defence system with the anti- bodies as another dataset. These are the body’s own soldiers to fight off the virus. Unfortunately they can also attack your own working body parts in what are called auto-immune diseases. A sort of human own goal situation.
So we have a virus genome, a vaccine genome, various anti - body genomes and then this enormous human genome that they all live within at the same time. They are all interdependent on each other all doing things within the body. Its one enormous chemical process which put simply is your life. Now we are taking about the digital representation of this whole chemical process.
But it goes further than just considering one life. To support defining an artificial intelligence algorithm linking virus to vaccine will require thousands of life examples. These being different human genomes. Now conventional computers based upon digital technologies, that’s binary on and off, are unlikely to be powerful enough to process this type of multi-dimensional artificial intelligence. The next level of computing hardware is based upon the principles of Quantum Computers. So we may need to fund the move towards Quantum Computers before we can finance their use for the digital vaccine generator. But it something we need to start doing now.
No comments:
Post a Comment