As a child, I was immensely fascinated to know about the surreal power of science to impact on the human race and the humongous potential of “Artificial Intelligence (AI)” to revolutionize our imagination beyond the normal curve. We, the humans are now living in the age of AI, the signal to start a new Industrial Age which claims that machines with the deep learning algorithms have the capability to surpass human intelligence.1
Figure 1: EDGE, an artificial intelligence platform, able to improve the identification of tumour-specific neoantigens and to analyze large human leukocyte antigen peptide datasets.2
What do you think: up to which extent the AI can challenge human cognitive skill or intelligence? Let me start by telling you an enigmatic story of an AI application of DeepMind!
“Go” is the world’s most complex board game. It is thought that there are more possible moves in the game of Go than there are atoms in the universe. Go is just not only a simple game to win: instead, it is an educational journey to understand individual’s personality, communication skills, bravery, competitiveness, discipline, concentration, balance, smartness, resilience power, and patience. To sum up, the prodigious depth and high skill cap strategic moves of this head or mind battle have the power to influence the character of a human, to strengthen the integrity and determination.3
A historic Go game challenge took place in March 2016 in Seoul, South Korea where Lee Sedol, world’s greatest Go player and a national hero of South Korea, 18 times world champion competed against the “AlphaGo”, which is a computer Go program developed by Google DeepMind, a database of historical games which is designed to learn by itself as well as to figure out its own moves by studying the available datasets at its disposal. All Lee Sedol’s die-hard fans and Go lovers were quite confident that human intuition was undoubtedly much more strategic and complex for AI to catch up and that Lee Sedol would beat AlphaGo hands down.1 This game was watched by millions of peoples as a sports spectacle with national pride at stake. Instead, the result of the Game showed the human that the AI programs could come up with moves that humans have never thought of before, revealing a plausible form of Al called Deep Learning, which can mimic the neural network of the human brain. It was a clear victory for AI which taught us about the enormous ability of AI to marshal a vast quantity of data beyond anything any human could ever handle.4
What is actually AI, Machine Learning (ML) or Deep Learning (DL)?
AI, a smart computer program, able to develop data-based decision and perform the tasks requiring human intelligence. The final goal of AI is to enable the machine to think.
ML takes artificial intelligence a step further where algorithms are programmed to learn and improve by themselves, without the need for human data input or reprogramming. It provides some statistical tools to explore and analyze the data.
DL is the next generation of ML and introduces multiple layers of learning from massive datasets of special interest where decisions and data classifications are refined at each layer producing accurate insights. It enables the machine to mimic the way human brain learns things by creating Multi- neural network architecture. It has several techniques like: artificial neural networks, ANN (use numerical data inputs); convolutional neural networks, CNN (use inputs in the form of images); recurrent neural network, RNN (inputs as the time series kind of data).5,6
Can you predict what can happen, if we combine the know-how of health science regarding the drug design and discovery and artificial intelligence with advanced machine learning?
A crucial obstacle in the field of breast cancer screening is the dependence on “human qualitative assessment of breast density”, which is inherently limited by our cognitive capacity and requires more accurate tool. Research has shown huge discrepancy among radiologists in making breast density determinations. This results in cancer masking effect of breast density which is one of the risk factors of developing breast cancer and one of the major causes of the miserable death of women worldwide. But now AI is able to change the face of breast cancer diagnosis. Dr Lehman, from Massachusetts General Hospital (MGH), Boston in collaboration with AI expert Regina Barzilay, professor of computer science and electrical engineering at the Massachusetts Institute of Technology in Cambridge have developed an algorithm through enough data of mammographic images and diagnostic information using deep learning and machine learning: which can automatically measure breast density on mammography, leading to the early-stage diagnosis of breast cancer preventing the metastasis of cancer and cure it without having painful chemotherapeutic ravages to the human body.7,8 Moreover, AI can diagnose Speech recognition patterns in depression and cognitive decline of Alzheimer patient. A small company “Benevolent AI”, has found new possible therapeutic for Amyotrophic lateral sclerosis (ALS), a rare neurodegenerative disease which had no treatment before.9,10
The revenue of AI health market is expected to reach up to 6.6 billion dollars in 2021. The Drug development sectors have massive amounts of data making it particularly suitable for use of AI application. For example, Human Genome Project scientists sequence a person’s entire genome and gather immense data to develop the genomic image with the help of Goggle's deep variant AI. Later on, from those images, the scientists find the specific areas within the genetic code which is druggable and offers new targets for the pharma industry to revamp the drug development pipeline in a better, faster and cheaper way. Machine Learning is used for understanding genomics, neuroscience, biologically feasible computing, medical diagnosis. AI is implemented to cut cost and speed up clinical development of medicine by aiding in the process of proper recruitment of eligible participants for clinical trials, by helping in monitoring the chronic clinical trial patients and the effect of the drug, with an Algorithmic application of AI or by being the digital guide of the participants of the clinical trial, through a simple mobile app.10
Figure 2: Different forms of Artificial Intelligence (AI) and diverse applications of AI .5,11
There are further diverse and plenty of AI applications in healthcare sectors as well as in each step of drug-related research and development: from the early-stage detection of new target molecule till the post-marketing surveillance of the new marketed medicine, personalized/tailored medicine, targeted therapy to patient's genetic makeup, molecular imaging, surgical robots, intelligent pharmacy, maintaining and processing huge medical record data, bionic machines, biochemical, and medicinal chemistry research.11
Needless to say, Artificial intelligence is a powerful gift of science, since the best and brightest among us just simply do not have enough computing power inside their heads due to cognitive power limitations. AI is Janus faced, with both positive and negative impacts but the last thing we want to do is to stop progressing and dreaming to go beyond our limit, don’t we? However, it is the human choice and ethics which are substantial to properly use AI enforcing proper regulatory laws, so that this technology cannot encroach on individuals’ rights and privacies. On the other hand, AI should be perceived as the serendipity to liberate us from routine tedious works, to allow us to be more human and to push us to do what we love and thrive for.
Bibliography
(1) (1) In the Age of AI (full film) | FRONTLINE - YouTube https://www.youtube.com/watch?v=5dZ_lvDgevk&t=331s (accessed Mar 20, 2020).
(2) Adding Artificial Intelligence to Drug Discovery https://www.genengnews.com/insights/adding-artificial-intelligence-to-drug-discovery/ (accessed Mar 21, 2020).
(3) No Human Can Beat AlphaGo, and It’s a Good Thing - Towards Data Science https://towardsdatascience.com/no-human-can-beat-alphago-so-what-3401b40fa0f0 (accessed Mar 20, 2020).
(4) Silver, D.; Huang, A.; Maddison, C. J.; Guez, A.; Sifre, L.; Van Den Driessche, G.; Schrittwieser, J.; Antonoglou, I.; Panneershelvam, V.; Lanctot, M.; et al. Mastering the Game of Go with Deep Neural Networks and Tree Search. Nature 2016, 529 (7587), 484–489. https://doi.org/10.1038/nature16961.
(5) Chan, H. C. S.; Shan, H.; Dahoun, T.; Vogel, H.; Yuan, S. Advancing Drug Discovery via Artificial Intelligence. Trends in Pharmacological Sciences. Elsevier Ltd August 1, 2019, pp 592–604. https://doi.org/10.1016/j.tips.2019.06.004.
(6) (1) AI VS ML VS DL VS Data Science - YouTube https://www.youtube.com/watch?v=k2P_pHQDlp0 (accessed Mar 22, 2020).
(7) Artificial Intelligence Used in Clinical Practice to Measure Breast Density https://press.rsna.org/timssnet/media/pressreleases/14_pr_target.cfm?ID=2036 (accessed Mar 21, 2020).
(8) A Promising future for AI in breast cancer screening .: UPO Finder - University of Eastern Piedmont http://eds.b.ebscohost.com/eds/pdfviewer/pdfviewer?vid=1&sid=5a515629-fef1-4575-90ed-3f09a377d5d4%40sessionmgr101 (accessed Mar 21, 2020).
(9) Bakkar, N.; Kovalik, T.; Lorenzini, I.; Spangler, S.; Lacoste, A.; Sponaugle, K.; Ferrante, P.; Argentinis, E.; Sattler, R.; Bowser, R. Artificial Intelligence in Neurodegenerative Disease Research: Use of IBM Watson to Identify Additional RNA-Binding Proteins Altered in Amyotrophic Lateral Sclerosis. Acta Neuropathol. 2018, 135 (2), 227–247. https://doi.org/10.1007/s00401-017-1785-8.
(10) (1) Will Artificial Intelligence Revolutionize Pharma? - YouTube https://www.youtube.com/watch?v=7zRndl3wZj0 (accessed Mar 23, 2020).
(11) (1) Better, faster, and cheaper drug discovery with machine learning by AstraZeneca - YouTube https://www.youtube.com/watch?v=qH3z5GwccxE&t=64s (accessed Mar 22, 2020).