When the cure is worse than the medicine – AI based solutions for Antibiotic Resistance
The great serendipitous discovery of antibiotics by Fleming in 1928 changed the course of medical science. It took care of many dreadful infections and saved several lives. According to reports, only 400 million units of penicillin was available in the initial months of 1943. Still, as the World War II ended, US pharma companies were making 650 billion units per month.
Too much of nothing is right.
But the overuse of this wonder drug has proven the cure to be worse than the medicine. It has caused enormous problems for all in terms of health and economy. We call it antibiotic resistance and its everyone’s’ problem! We may ask, is the problem so gigantic?
Well, yes, because antibiotic resistance has become a public health emergency that has been acknowledged worldwide. Antibiotic-resistant bacteria reportedly caused around 30,000 deaths annually in the US and a slightly higher number in Europe, with Spain being one of the European countries with most cases and around 1,800 deaths every year.
What’s even worse?
In addition to a rise in antibiotic resistance, these resistances can be spread without global borders!! Bacteria along with the resistant ‘genes’ is known to be transported from one point to another along with food, our body, and even through animals because they too are given antibiotics.
Once antibiotic resistance is developed, the resistant bacteria grows through poorly prepared food, proximity and poor hygiene to infect us, humans. But scientists have now successfully found a solution to this daunting problem with the advent of artificial intelligence.
Ray of hope
Artificial Intelligence is not a new phemonenon. AI and machine learning has been used in a lot of areas. However AI and machine learning algorithms in drug discovery is changing the way how R&D is done within the pharmaceutical industry. AI has helps companies scan through huge datasets to identify new targets, find new molecules and even re-purpose drugs.
Earlier this year, Artificial Intelligence (AI) was used by the Massachusetts Institute of Technology (MIT) researchers to identify molecules that kill even the ‘untreatable’ strains of bacteria. A potent new antibiotic compound was detected from a pool of more than 100 million molecules using a machine-learning algorithm by Professor James Collins, a bioengineer and Regina Barzilay, an artificial-intelligence researcher at MIT. This was made possible by developing a neural network and allowing the computer to develop its expertise.
A deep neural network, in which the nodes and ties of its learning framework are influenced by the interconnected neurons in the brain, is behind this groundbreaking discovery. Neural networks, which are adept at identifying patterns, are used for image and speech recognition across diverse industries and consumer technologies. In order to find certain identified chemical structures, while traditional computer programs may screen a library of molecules, but neural networks can be trained to learn for themselves the structural signatures that may be unique and then identify them.
James Collins says, “We wanted to develop a platform that would allow us to harness the power of artificial intelligence to usher in a new age of antibiotic drug discovery.” This study used a machine learning system that was made to work through a data set of molecular features of nearly thousands of drugs and natural substances. AI successfully detected how well or not the substance blocked the growth of the bug E coli, and this work was published in Cell early this year.
Meet Halicin – World’s first AI Discovered antibiotic molecule.
Halicin initially entered the scene as a possible remedy for diabetes and failed the test. But it wins hugely when the scientists at MIT found these Halicin molecules to be promising in the initial laboratory experiments. Interestingly, “Halicin” gets its name after Hal, the AI in the sci-fi film 2001: A Space Odyssey.
These molecules were validated by doing experiments on mice by destroying many of the world’s most troublesome disease-causing bacteria, including some strains that are immune to all known antibiotics. Who knew computer algorithm would find what humans didn’t know and were not looking for!!
The scientific fraternity is hopeful to come up with some more promising antibiotics using artificial intelligence. Not only are sequencing-based AI applications employed to tackle the issue of antimicrobial resistance but also can be used to gather clinical data to help with clinical decision support systems. It can be helpful for the medical practitioner to monitor the trends in antimicrobial resistance. What is more, AI applications are also gaining popularity in designing new antibiotics.
This miraculous discovery comes as relief while the global issue of Antimicrobial Resistance grappled the world. According to a report from the UN in 2019, if the issue of antibiotic resistance was left unattended without any significant breakthrough, we could witness up to 10 million deaths a year by the time we reach 2050.
Miles to go…
Many scientists are unwilling to share data at this stage regarding antibiotic-related information. This could lead to increasing demand for AI-based algorithms which can predict accurately using a small training dataset. There are also experiments going on to use AI to explore the best combination of biomarker combinations that can be used for clinical decision making. The algorithms can be programmed with elements similar to known drugs to have a preference for novel structures.
In the current scenario of businesses resuming after the Covid-19 lockdown, computer algorithm and AI can provide many healthcare solutions to battle various issues of the post-pandemic world. This could range from AI algorithms for risk assessment predicting Covid-19 patients in real-time, enabling cloud hardware tracking, using CCTV surveillance for monitoring violation of Covid-19 safety protocols etc.
Insilico Medicine has already used the AI-based generative chemistry method to engineer six new molecules that possibly could control the 2019-nCoV virus from replicating inside the body. Insilico is a Hong Kong based start-up that provide AI-hypothesis generation engine for drug discovery, drug target identification and biomarker development. The company.
Another example is of BenevolentAI, which has successfully used AI for drug repurposing. The Artificial Intelligence solution provider has successfully used machine learning to identify baricitinib as a treatment for COVID-19. The London based company integrates technology into every step of the drug discovery process, from hypothesis generation to early-stage clinical development.
Analysis from WhatNext, show that there are 79 start-ups focused on developing AI based drug discovery solutions. The total funding that these start-ups have received is over $2.8 billion.
With the advancement in AI, neural networks, computer vision, NLP and robotics, we may see many more new antibiotics remain being discovered, and also could potentially witness AI screening being implemented as a real-time solution in our healthcare scenarios?
There is great potential to connect molecular structure data to biomedical information about relevant receptors and diseases to find potential drug targets. Let us hope and contribute to this beginning of a new journey, for the greater good of humanity.
To deep dive and stay continuously updated about the most recent global innovations and learn more about exponential technology applications in your industry. Register free for Light Access to the WhatNext Platform.