In the year 2020, there was a prolonged ‘incidence’ on The Earth that threatened to wipe out human civilization caused by a virus. With very limited scientific knowledge at hand, People on the Earth fought the virus with great tenacity and in the process, after the involuntary and untimely death of many human beings, the virus was contained…
Now, what the hell is this? Oh, yes. This is from a would-be chapter of a history lesson in the year 2110! A time when diseases will be termed as ‘conditions’, death will be ‘voluntary’, present scientific knowledge will be considered ‘archaic’, and so on. But, history will remember our time – this time as a Great Revelation. Of everything else, our heroism will be remembered and revered.
Well, those will be good new days for the generation unborn. In reality, we are still in April 2020 and the fight is going on all across the globe with glimpses of slowing down here and there. While things are scaling up in the United States at a breakneck speed and in India, in sudden spikes, new cases are showing declination in Spain, Italy, and Iran; mighty China and gutsy South Korea are already ahead in the COVID-19 containment race. Things will improve, well at least for some major parts of the world, and in another four months, people will start looking back at the utter carnage with despair and fear.
Thus, as nature dictates, four months down the line, people are expected to talk more about medical preparedness and precautionary processes along with COVID-19 medicines, vaccines, and ventilators. Medical science and technology will be subjected to an unprecedented demand for drugs and devices. The branch of science that takes its own time as governed by Nature, will seek alternative means to make up the speed and improve precision. Artificial Intelligence and Machine Learning, already in use in Medical research and development, will be applied in more practical and consumer-facing sides.
So, how may AI become our savior? Of all practical use of AI and ML, healthcare alone has the best applicability of intelligence-induced algorithms. Let’s find out the four major use of AI and ML in medical science today:
To diagnose diseases
Diagnosis of a disease is both an experience and training-oriented skill that one acquires gradually over time. In certain areas of medical science, there is a heavy disparity in supply and demand when it comes to experts. As a solution to this, Deep Learning Algorithm is getting used to diagnose in the same way an expert analyses the observations and findings before pinpointing the disease. With enough digitized data in place, machine learning can give near-accurate results. In the latest scenarios, Deep Learning is being applied to a collection of sources of data including CT and MRI scans, genomics, proteomics, patient’s history, and prescriptions to assess the extent of a disease. The result is somewhat like a hundred different experts analyzing, corroborating, and finally mapping everything to generate the most accurate diagnostic report.
Cheap and faster production of medicines
Medicine production is an expensive and time-consuming process. The implementation of AI in the four major drug manufacturing stages – identification of target molecules, discovering the effective drug candidate, fast-tracking clinical trials, and finding the biomarkers for diagnostics, have now become less time-consuming and more precise. This is the reason the experts predict that COVID-19 drugs and vaccines will be ready in less than 10-months as against years needed to produce an effective drug in the normal way. If there is a cure for COVID-19 which, we are very much sure of, Artificial Intelligence can help us find it quickly.
Personalized treatment & care
Not every patient responds to medicine in the same way. The patient’s response depends upon various factors and complexities in the whole process. This usually makes it extremely difficult for the doctor to accurately ascertain the outcome of a certain drug on a certain patient. Now, AI is there to help the good old doctor! Using machine learning, a very large volume of statistical data can be compared to derive the knowledge which can further be used to personalize medication and care. This is indeed a revolutionary breakthrough in an area that otherwise still sometimes depends on speculations and often subjected to hit and trial methods.
Improvement in gene editing
The CRISPR (Clustered Regularly Inter-spaced Short Palindromic Repeats) system is the most effective gene-editing technique available today. However, the CRISPR system makes use of short guide-RNAs which, sometimes cause off-target editing resulting in dangerous side-effects. The use of ML has drastically reduced the risks by correctly predicting off-target effects and guide-target interactions. AI can radically improve the development process of better guide-RNAs for CRISPR processes.
How Artificial Intelligence will be used in the post-COVID19 era?
Preliminary detection and assistance in isolation
Beijing-based Infervision (a Sequoia Capital-backed company) recently developed an application that can detect initial signs of COVID-19 pneumonia from CT scans. Although the process is in the rudimentary stage, it can drastically reduce the time spent by over-worked lab workers in the early detection process. As per a report published in Wired magazine, the application has been deployed in 34 hospitals across China and so far used in 32,000 cases. It is another AI company, BlueDot (a Canadian start-up venture) that first detected an unusual surge of pneumonia syndromes in Wuhan as early as December 30, 2019. Designed in the SaaS (software as a service) model to track, locate and conceptualize the extent of the spread of infectious diseases, BlueDot uses big data models to source information and then employ smartly written AI algorithms for detection. Services from companies like Infervision and BlueDot will be in high demand in this COVID-19 period. Detection of Coronavirus-affected patients must be swift and Artificial Intelligence can help us immensely.
Treatment and care
An AI-based application can go through an ultra-large volume of medical information, medical research papers, and all other forms of data and take highly logical decisions. These decisions are primarily focused on finding the right compound or in COVID-19’s case a set of most likely compounds that can block SARS-CoV-2 from attacking human cells. French AI service provider Iktos is doing this in collaboration with US-based SRI International. Iktos’ role in this venture is to develop virtual novel molecules while SRI’s automated synthetic chemistry platform (known as SynFini) will be tasked with creating the actual molecule. In fact, in IEEE’s prestigious Spectrum magazine, Boston-based Journalist Megan Scudellari also enlisted four other similar on-going ventures which, are using machine learning towards producing drugs for COVID-19.
The ultimate goal: A vaccine for COVID-19
Finding an effective vaccine against COVID-19 has become the primary goal of medical and biotechnology labs, research institutes, pharmaceutical companies, governments, and expert individuals. In collaboration with these entities and individuals, the Allen Institute of AI has come out with an open resource dataset for COVID-19 research. The datasets, thus accessible for anyone across the world, provides an enormous amount of information for the research and development of an effective vaccine. Similarly, the online community of data scientists and machine learning experts, Kaggle (a Google subsidiary) is also contributing to the same effort. Another Google-backed venture DeepMind has already made significant progress in helping pharma-research entities in the race to produce the COVID-19 vaccine. As Ben Dickson points out in a recent article on TheNextWeb, the most difficult hurdles to make a vaccine for COVID-19 are time and artificial intelligence can help reduce both.
As I am concluding this post, I checked Worldometer once again a few minutes ago. Except for the United States where the ‘new cases’ graph is menacingly going up every day, Spain, Italy, Iran, and Germany are all showing a slow but steady decline. By the time I write my next post, I sincerely hope to add some more countries to this list. By then, I also hope that the data scientists and AI developers together with bio-science experts will advance a few more steps towards COVID-19 drugs.