Drop off your CV
We'd love to hear from you. Send us your CV and one of our specialist consultants will be in touch.
Consulting firm Frost & Sullivan predicts that artificial intelligence systems will generate global revenues of $6.7bn from the healthcare sector by 2021. Artificial intelligence (AI), is the capability of computer software to imitate human intelligence and carry out complex tasks. It can be used to find hidden patterns in vast amounts of data that would be highly time consuming, and sometimes impossible, for a human to achieve. With the large quantities of medical and patient data available in the scientific sector, AI has the potential to transform the industry resulting in new ways of diagnosing and treating diseases.
But what does the future of this industry look like under AI, and what are the biggest threats to its success? In this article we explore how the application of AI is advancing the scientific sector in the areas of genetics, oncology and drug discovery and what could hinder the industry from being able to reap the full benefits of the technology.
Many illnesses are determined or influenced by a person’s genetics. This has led to an interest, and realistically, a necessity, for those in the industry to better understand our genetic makeup in the hope of being able to develop more successful medicines and treatments. Genome sequencing – the process of identifying the order of DNA building blocks in a living organism – is crucial to understanding how genes work together and, therefore, understanding an individual’s risk of diseases and the appropriate treatment options. This process involves sequencing millions of genomes, which would have traditionally been done through Sanger sequencing, but this method is slow and isn’t ideally suited for large-scale genome sequencing.
New advances in AI have revolutionised the study of genomes giving the possibility of sequencing on a large scale. High-throughput sequencing (HTS) became commercially available in the 2000s allowing this process to be done quicker, cheaper and on a larger scale. HTS has its downfalls: random errors can be generated throughout the process, which need to be identified. Google’s Brain team launched an artificial intelligence tool, called DeepVariant, to counteract this. By using AI to analyse the data, DeepVariant is able to produce a more accurate picture of the genome.
The opportunity to explore the human genome on an individual level means personalised medicine based on genetic differences is now becoming a possibility, with the industry expected to reach $87 billion by 2023. The success rate of a medication varies from person to person and can be influenced by an individual’s genetics. By being able to prescribe medication and treatment based on a person’s genetic makeup, the amount of people being put through unsuccessful invasive and expensive treatments will be reduced and, instead, they will be prescribed the most effective treatment – saving lives, stress, time and money.
One particular focus in the industry is gene editing as a treatment option for genetic diseases. This involves DNA being either inserted, deleted or modified in the genome. This technology could be used to inhibit genetic diseases, such as cystic fibrosis, or to potentially treat illnesses, such as cancer and HIV. It also has potential applications for other industries, such as the agricultural sector, enabling farmers to produce high-yielding and disease-resistant crops. However, gene editing gives rise to its own challenges in terms of ethical, moral and legal implications. How far is too far? Do we look to eradicate diseases, or would this technology be used to modify characteristics such as height or intelligence?
Even with the best healthcare system and medical staff, cancer can sometimes be difficult to diagnose. In the UK, four out of ten people with cancer are misdiagnosed at least once before their cancer is finally detected, with 21% of patients waiting more than six months to be diagnosed correctly. AI is now paving the way for new methods of diagnosis, and potentially at an earlier stage.
Recent studies have proven that AI can detect several types of cancer with a higher level of accuracy than standard diagnostic methods. A 2018 study found an artificial intelligence machine correctly diagnosed 95% of skin cancers, whilst dermatologists averaged at an 86.6% success rate. The AI software had learnt to distinguish between benign and cancerous lesions after being shown more than 100,000 images. These kinds of advancement could result in earlier diagnosis, allowing treatment to be more successful and leading to an improvement in survival rates from various types of cancer.
AI could also help improve the success of cancer treatment. The advances in genome sequencing discussed earlier look promising for enabling personalised cancer treatment plans in the future. Intel launched the Collaborative Cancer Cloud in 2015 allowing institutions to securely share patient genomic, imaging and clinical data to be analysed. By 2020, it hopes to deliver treatment plans to patients based on their genome within 24 hours, which could lead to more successful treatment and higher survival rates.
One of the biggest threats to the success of AI in cancer detection and treatment is a lack of data, which the software needs to learn from and to be able to identify patterns. Fiona Nielsen, Founder and CEO of genomic data company Repositive.io backs the need for more data to help in the fight against cancer: “To ‘really find a cure’ it is important to unlock the data needed to make new discoveries. There is already plenty of clever algorithms, plenty of clever researchers and clinicians; the speed of discovery will depend on the speed of data availability."
Could all these developments in AI ever be a substitute for human decision making when it comes to cancer diagnosis and treatment? It is hard to see that happening any time soon. IBM’s AI platform, Watson, was introduced into hospitals with the aim of revolutionising cancer diagnosis and treatment. It made the headlines in 2018 after it was found to be recommending “unsafe and incorrect” cancer treatments. Whilst the use of AI in oncology is promising, at this moment in time AI is not sophisticated enough to be used in isolation - it can only be used in conjunction with human medical judgement. The focus is very much on helping to improve and speed up human decision making, rather than replacing it.
The average cost of bringing a new drug to market is $2.6bn with the process taking 10 years on average. The introduction of artificial intelligence is promising to revolutionise the industry making the process both cheaper and quicker. A study conducted by researchers at Carnegie Mellon University and Albert Ludwig University in Germany, estimates that AI could cut the cost of drug discovery by about 70%. Many pharma companies already have some type of AI program in place with the software currently being used at various points within the drug discovery process.
Firstly, AI is being used to help design new medications. For example, Atomwise has built its own AI system to create potential new drugs for various diseases, such as Ebola. The software starts with a 3D model of a molecule and then generates a series of synthetic compounds. It then predicts how likely it would be for the two molecules to interact. If a drug is likely to interact with the specific molecule, it can be synthesized and tested. As with all proposed medications, these drugs would then need to go through a clinical trial process to ensure their safety and effectiveness. So far, this technology has led to a drug for multiple sclerosis being licensed to an undisclosed UK pharmacology company.
With its ability to process huge amounts of data, AI is also having a big impact on clinical trials. Due to the importance of selecting the right patients when trialling potential new medication, the process can be timely and costly. By using AI to analyse complex genomic data, researchers are able to find candidates with similar genetic profiles more quickly. For example, Deep 6 AI can find eligible patients for medical trials in a matter of minutes by sifting through large amounts of unstructured and fragmented medical data in free-text form that can be difficult to search through manually. The technology can also match patients with similar symptoms and characteristics who may have not yet been diagnosed. This kind of advancement could potentially reduce clinical trial recruitment time by many months increasing the speed to market of successful treatments.
There are also benefits to using AI during the clinical trial process. Traditionally, participants have been expected to keep a log of all symptoms and to attend regular medical appointments. This process is open to error as participants may forget to attend appointments or fail to document symptoms correctly. Intel has launched an AI-based system called Pharma Analytics Platform. Patients wear a device to keep track of stats such as their heart rate, with all the data fed to the cloud. The data can be analysed in real time speeding up the process of analysing a drug's effectiveness.
Companies like GNS Healthcare have also found a way to use AI to predict the way a patient will react to a proposed treatment. Reverse Engineering and Forward Simulation (REFS) is a machine learning software that uses a vast amount of information - from genomic data to electronic health records - to find possible combinations of elements that may affect a patient’s response to drug treatment.
With its current applications in the drug discovery process, it is easy to see how AI has the potential to radically transform the industry, but will this impact on the skills the sector needs? Many industry leaders feel that drug-discovery jobs - and the skills needed to do them - are already beginning to change.
Whilst the prospect of AI transforming the scientific industry is still at an early stage, the technology has the potential to improve and save many lives. Having access to vast amounts of reliable data is key to AI becoming more accurate, and this could pose a problem for some areas of healthcare. For example, with rare illnesses, identifying genetic patterns could be difficult.
Trust and public perception could also limit the impact AI has on the scientific sector. For those suspecting they may have a serious illness, such as cancer, allowing a ‘machine’ to diagnose you could be quite a daunting prospect. However, if the accuracy of such technology can be consistently proved beyond that of humans, many may prefer this option. The same applies to areas such as gene editing, where the potential outcome of this has yet to be explored and the possibility of making ‘designer’ humans may face strong objection.
Lastly, for AI to be successfully implemented on a large scale there needs to be a change in skill sets within the scientific industry, as well as mindsets, to fully embrace the change it may mean for certain job roles. It is likely AI will take over certain tasks and parts of job roles, which could either be embraced or face objection. For example, tasks usually carried out by scientists, such as measuring cell characteristics, can be completed more quickly utilising AI software. Whilst it is unlikely that full jobs will be made redundant any time soon, the reduction of many timely tasks could mean less demand for specific roles. We are also likely to see an increased need for certain skilled roles, such as computational biologists, as more and more companies seek to apply AI technology to different processes.
To fully embrace the opportunities AI offers, retraining for those in the industry is essential. The uses for, and application of, AI are changing constantly. Staying abreast of the latest advances and how these can theoretically be applied is particularly important for those in the industry that would like to stay ahead of the curve. For senior leaders in this sector, an understanding of how to apply the latest AI technology to existing processes to increase efficiency and accuracy, as well as reduce costs, is going to be crucial.
As the industry continues to adapt to new technology, companies need to have a plan in place to attract and retain those with the crucial skills sets they need. Pipelining future talent can ensure companies are ready to implement technology and new processes to allow them to stay ahead of competitors.
If you would like to discuss this topic and how the growth in AI may have altered your company's talent needs, I can be reached at +44 (0) 113 258 5150 or email@example.com.
Wilson Center, Your DNA Avatar - What Happens When Artificial Intelligence Meets Cutting-Edge Genetics?
Forbes, The Wonderful Ways Artificial Intelligence Is Transforming Genomics and Gene Editing
Genetics Home Reference, What are genome editing and CRISPR-Cas9?
New Medical Life Sciences, Artificial Intelligence and Life Science
The Guardian, Computer Learns to Detect Skin Cancer More Accurately Than Doctors
Intel IT Peer Network, Intel & OHSU Announce Collaborative Cancer Cloud at Intel Developer Forum
Bloomberg, IBM’s Watson Hasn’t Beaten Cancer, But A.I. Still Has Promise
Nature International Journal of Science, How Artificial Intelligence is Changing Drug Discovery
Clinical Research News, The Intelligent Trial: AI Comes to Clinical Trials
Forbes, How AI Is Revolutionizing Drug Discovery
Emerj, Artificial Intelligence in the Pharmaceutical Industry – An Overview of Innovations
Quartz, Artificial Intelligence Could Build New Drugs Faster Than Any Human Team