Discovering a new future through Artificial Intelligence and Machine Learning

Discovering a new future through Artificial Intelligence and Machine Learning

A number of industry sectors including aviation, retail and financial services have long applied data and utilised analytics to augment human capabilities, optimise their customer engagement and maximise returns.

Indeed the term Artificial Intelligence (AI), incorporating machine learning approaches, is something of an heir apparent to what was once commonly referred to as the utilisation of ‘Big Data’.

The Engineering and Physical Science Research Council (EPSRC) has usefully defined AI to reflect the evolution and innovation in Big Data deployments: “Artificial Intelligence technologies aim to reproduce or surpass abilities (in computational systems) that would require ‘intelligence’ if humans were to perform them. These include: learning and adaptation; sensory understanding and interaction; reasoning and planning; optimisation of procedures and parameters; autonomy; creativity; and extracting knowledge and predictions from large, diverse digital data.”

Early forms of AI have been in use for several decades, particularly in sectors which have a requirement for somewhat repetitive but critical decision-making, or for the efficient analysis of large volumes of data.

However such computing systems – and indeed their operators – did not have the capacity to interpret vast quantities of unstructured data. As computers have become more powerful, it has become possible to analyse those vast quantities of data. And Deep Learning, a subset of Machine Learning, has exploded in the last few years alongside more affordable access to compute capacity.

It is this exponential increase in computing hardware performance, together with the increased affordability, that is now driving the rapid growth in the application of AI across many more business sectors, including life sciences.

This evolving landscape has been recognised by the UK Government with the recent publication of its Artifical Intelligence Sector Deal to ensure that the UK is at the forefront of AI development. The Deal aims to take “immediate, tangible actions” to advance the UK’s ambitions in AI and the data-driven economy, in line with the Government’s new Industrial Strategy, which is also underpinned by an associated Life Science Sector Deal

And life sciences companies are now increasingly recognising the opportunities to apply data in order to optimise drug development activity and address treatment and patient care challenges. As we continue to understand more about our human biology and the biological processes involved in the progress of disease, the opportunities to apply data to progress new developments, treatments and indeed cures, continues to grow.

For R&D, the ability to analyse large volumes of data to produce deeper insights is enabling us to undertake advanced modelling to produce stronger hypotheses and therefore more targeted research. Other advances, including medical imaging interpretation, genomic profiling and personalised medicines are all set to be revolutionised by the application of Artificial Intelligence.

Speeding up drug discovery

Real life deployment of Artificial Intelligence in drug discovery is delivering the future right now.

My own company GSK, like other pharmaceutical companies, generates a vast amount of data – and we have recognised that this represents an immense wealth of valuable information. We are using advanced technology and analytics to transform the way we run drug discovery and make decisions.

Although we have used “machine learning” – where machines apply knowledge from vast data sets, recognise patterns for themselves and make predictions as a result – for a number of years, we recognised that these techniques could be applied more broadly. In particular, we saw that the type of pattern recognition used in the financial world or for image analysis could also work in drug development. The true value of machine learning in drug discovery will be realised by the integration of the drug discovery domain expert with the data scientist.

In response we formed a new Drug Discovery Unit dedicated to exploring and applying AI to drug discovery, under the leadership of John Baldoni, our Senior Vice President of In silico Drug Discovery. Using AI is set to help us and the wider pharmaceuticals industry significantly accelerate our drug development processes.

Drug development is highly resource-intensive and costly, timescales can vary – but typically take 12 years from the initial discovery stage to reach the market; estimates of costs also vary – with the Association of the British Pharmaceutical Industry (ABPI) putting the cost at £1.15bn per drug.

With AI, this process could be considerably shortened. In order to harness the power of AI, GSK has recently entered into a range of different collaborations.

Last year, we started working with Exscientia, a UK-based AI and machine-learning company. During this collaboration, Exscientia is applying its AI enabled platform and combining this with the expertise of GSK, in order to discover novel and selective small molecules for up to 10 disease-related targets, nominated by GSK across multiple therapeutic areas.

Exscientia’s objective is to use this approach to deliver candidate-quality molecules in roughly one-quarter of the time, and at one-quarter of the cost of traditional approaches.

GSK is also working with the Hartree Centre, the UK government-backed research centre, which has some of the most technically advanced computing, data analytics, machine-learning technologies expertise in the UK. GSK is working with Hartree on a supercomputing project that enables scientists to analyse millions of biomedical research publications to identify pattern correlations that might act as the starting point for new treatments.

And it is also part of the Accelerating Therapeutics for Opportunities in Medicine (Atom) consortium, a US public-private partnership with the Department of Energy and the National Cancer Institute. Atom is developing computing models capable of vetting millions of molecules for efficacy and structural relationships. These models will be able to adapt (without human supervision) as they learn from the huge data sets they go through – a process known as “deep learning”. The value is in reducing the search space for compounds as opposed to finding “the” compound. We still need human interpretation/design.

GSK is giving Atom access to more than a million compounds it has screened over the past 15 years, all of which have biological data associated with them.

Impact on skills

AI Technologies, Big Data Analytics, Robotic Process Automation and Machine Learning are set to bring a range of unimagined benefits to science-based industries – from drug development and diagnostics to medical devices and monitoring technologies. Practical benefits include new insights into disease, cost reductions, improved productivity and increased speed to market. But it is the benefits to society, including improved treatment outcomes, care management and quality of life that will continue to drive progress.

The Science Industry Partnership of employers is now seeking to establish how the continued development of AI-based technologies will impact the skills needs of science industry organisations, highlighting any potential skills gaps, risks to current roles, new opportunities and recommendations for education and training reform that may be required.

It began evidence gathering through its Skills Strategy 2025, which immediately identified those areas and occupations requiring urgent action on skills including Bioinformaticians, Cheminformaticians & Health Informaticians.

It has also evidenced a shortage of Computational Scientists – and a mismatch of skills in the current workforce, with large portions educated before computational sciences formed a significant part of curricula, and current graduates lacking the type of skills sought by employers.

The next step is to further identify the future skills required for the development and deployment of AI technology and secondly the implications for those working in an AI environment for sector-specific applications. Working in partnership with be key here – with organisations such as Innovate UK and The Alan Turing Institute.

However, the SIP is not waiting to take action on skills and has already made a number of clear recommendations to help the Sector prepare for a technology-enabled future.

An early move into formalising vocational skills has been the development of a Masters Level Apprenticeship Standard – Bioinformatics Scientist at Level 7. Bioinformaticians are life scientists who use computer and informational techniques, which are applied to a range of problems in the life sciences, for example, in pharmaceutical companies in the process of drug discovery.

This vocational training pathway will deliver highly skilled work-ready individuals who will be able to apply their knowledge of computing (including coding), biology, statistics and mathematics while fulfilling a job role in a host company. We’re now looking at other Standards such as Computational Scientist at Level 8/Masters.

This development recognises that the sector requires clear, highly specialised, vocational pathways into the jobs of the future.

The SIP is also behind moves to build mathematical & statistical skills throughout the education system, and has applauded recent Government investments in this critical area. It is essential that a good grounding in mathematics and statistics is embedded from school age. In addition, training in maths and stats must be available and accessible to address identified weaknesses in the existing scientific workforce.

We also need to build practical computing skills across stem education: data handling, computing, programming and software using skills are now touching roles at all levels. The need for these skills is growing and should be core elements of STEM related education. This would provide learners with the skills to work with company specific software and equipment as soon as possible.

Finally, we need to facilitate transferability between public and private scientific workforces: facilitating the transfer of staff between industry, NHS and academia has many benefits. These include developing multidisciplinary individuals and teams, increasing cross-sector collaboration, enhancing transfer of ideas and innovation, and creating a more flexible workforce.

There is no doubt AI has immense power to deliver improvements in cost, quality, speed, efficiency and outcomes – right across every function in our businesses. The SIP will be seeking collaborations in order to build the skills and capabilities to give our sector the combination or core science and computational skills that will allow to transition from utilising Big Data to Deep Data – high quality, actionable information that will take us to another level of understanding and in turn discovery.

For further information on the SIP contact

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.