Share
Artificial Intelligence, Machine Learning and The Future of Biotech

Artificial Intelligence, Machine Learning and The Future of Biotech

Artificial intelligence and machine learning are the core buzz words in tech-world and have been for some time now. More recently, AI and machine learning software technology has started to crop up in the field biotech. And it’s becoming integral to the pace at which major biotech discoveries are pouring out of the pipeline. In this analysis, we’ll look at why machine learning and AI are super-charging biotech and the key areas the technology is having a growing influence.

What Exactly are AI and Machine Learning?
AI is really just a more complex set of the coded rules that together form a piece of software. In essence, there is no formal delineation between ‘software’ and ‘AI’. AI is simply the term used to describe software that performs a task in a way that makes an impression as ‘smart’ and an automation of a task that would previously have been expected to require human input.

Machine learning can be best described as advanced AI, or non-applied AI. Whereas ‘standard’ AI software is created for the execution of a specific task along defined, programmed rules, ML software is designed to create its own rules based on data input. It marries algorithms with data and, mimicking to an extent the human mind works, classifies data, looks for patterns in it and then draws conclusions based on those patterns. It will then operate in accordance with those conclusions, as though it has ‘learned’.

A tech start-up, or even new product or functionality of a well-established tech company, that does not include the term AI or ML in its description is rarer than the Siberian tiger these days. While there are certainly occasions where companies use the term AI as almost a simile for ‘software’, the advances in the complexity of software solutions do mean that the functionalities created are often impressive and can justify the tag. Machine learning is, of course, more specific and software referred to as being machine learning should meet minimum standards of being able to set its own rules as a result of the data it is fed.

AI & Machine Learning in Biotech
Diagnostics – bio-assays, the quantitative or qualitative analysis for the presence or functionality of a target entity, were previously periodically updated when data indicated a paradigm shift. Machine learning algorithms now immediately add all newly authenticated test results to an assay, improving it on an ongoing basis. This means with every new diagnostic test, the next becomes more accurate.

The newest ML-based genetic analysis can produce results from a blood sample within days, rather than the several months previous standards required. Each iteration of the analysis also makes the diagnostics process ‘smarter’. ML algorithms can be applied to any diagnostics process that can be digitalised. Faster, more effective and multiple diagnostics will help identify conditions in their early stages. This is often key to effective treatment, cure or management.

Robo-Lab Assistants – a significant part of biotech research involves tediously repetitive tests, recording their results and other time consuming administrative work. The human element required to execute this extensive and precision but repetitive work has always greatly added to the expense and time involved in biotech research. More and more of the work previously carried out by lab assistants is now being automated through AI-powered robotics and algorithms then recording and assimilating results.

Much of the bottleneck in biotech, particularly in gene editing techniques and maps, is analysing the huge volumes of data that both already exist and is being increasingly generated. ML algorithms are easing this bottleneck and facilitating big data compilation and analysis.

Drug Discovery – the average cost of identifying a new drug and putting it through clinical trials is $2.5 billion (£1.88 billion). It also takes 12 years to go through the rounds of clinical trials needed for a new drug to be approved. Those costs are increasing as pharmaceuticals companies have already developed the drugs that represent the ‘lower hanging fruit’.

ML technology will hopefully also solve that bottleneck in biotech. Dramatically cutting the time and expense involved in compiling the test data required for clinical proof will allow pharma companies to explore more leads. That will make R&D into drugs for common conditions with a limited commercial market less of a business risk.

ML increasing the efficiency of genetic sequencing and processing genetic data is also opening up doors for drug discovery starting from genetics. Image recognition, a cornerstone of machine learning, can also now be used to analyse cells treated with drug compounds. This used to require PhD-educated scientist analysing one cell at a time.

Personalised Treatment – AI and ML’s ability to screen huge volumes of cloud-based data, as well as quickly and cheaply map and analyse an individual’s genetics, is expected to lead to greater personalisation of drugs and treatment. Big data patterns may show that an individual’s genetic disposition means that their condition is likely to respond better to different combinations and quantities of drugs. Matching individual genetics to patterns across the response to different drugs of thousands to millions of previous patients would not have been a practical possibility before the advent of ML in biotech. Over time this personalisation would be expected to become incrementally more effective as the volumes of data build up.

Bio-mechanical Prosthetics – among the most interesting biotech developments based around machine learning technology is neural nets that understand signals from the brain. Research currently being conducted in this area will hopefully eventually lead to biotech scientists creating what will essentially be like a biological USB port which will connect our body to an external computer.

Our brains send neural signals, which are basically voltages, to our limbs, telling them what to do. Developing neural nets requires huge volumes of data. Sensors placed on the body can pick up these voltage signals being sent from the brain to nerves. Biotech scientists are well on the way to using machine learning to analyse enough of these voltage readings and the resultant body movements to build a neural net of the intricate web of brain signals that control our movements. It is believed that this will mean bionic prosthetics whose movement will be controlled by the wearer’s brain in the same way a biological limb is controlled will soon become a reality.

Conclusion
AI and machine learning are leading biotech into a Golden Age of research and discovery. Much of the work in this area is still being conducted by biotech start-ups on much more limited budgets than big pharma has. However, as major pharmaceuticals companies acquire biotech start-ups, or are sold their IP once it has reached a certain stage of development, the influence of machine learning is growing within the moneyed echelons of the health industry. The big companies are also now starting to build their own in-house computer science teams.

It is forecast that by 2025 the application of machine learning will have become mainstream within the pharmaceuticals industry. The hope is that, combined with the lightning pace that machine and deep learning technology is itself developing at, biotech will revolutionise medicine and our understanding of the intricacies of human biology and genetics over the coming decades.

Leave a Comment

19 − 17 =