There is a lot of potential for Artificial Intelligence (AI) in life sciences to revolutionise things and make key roles easier, ultimately giving patients better care.
Diagnostic tools that incorporate AI have already started to appear, helping to detect things like fractures and eye problems at a faster rate.
The management of data is one area where AI can make a significant impact, making it easier for organisations to communicate key pieces of information when it’s needed. Of course, there are limitations for what AI in life sciences can realistically do, but if used well, AI technology will supplement the work of doctors, nurses and other key players to make their jobs easier and simplify the processing of data.
What is artificial intelligence?
The term ‘artificial intelligence’ was coined by John McCarthy in 1956 who defined it as: “the science and engineering of making intelligent machines." [1] In essence, it refers to computer systems that have been developed to perform with aspects of human-like intelligence.
There are four types of artificial intelligence:
• Reactive - this is the most basic type of AI system which is built to perform a specific function. It doesn’t have the ability to interpret data and respond to it, rather it will perform in the exact same manner every time it’s used. A good example of this is IBM’s chess-playing computer Deep Blue. It can react to a player’s chess moves and perform the game, but it cannot learn from past experiences or apply its functions to any other situation.
• Limited memory - a self-driving car is an example of a limited memory AI system. It has pre-programmed information relating to roads and traffic rules which it can draw from but it cannot take in the context of a situation and learn from it.
• Theory of mind - this is a much more advanced type of AI that has an understanding of human thoughts and emotions. This type of AI is where we might be headed in the future, with research currently focusing on it.
• Self-aware - a self-aware form of AI is rather far-fetched at the moment and is more likely to feature in the plot of a science fiction movie. To reach this stage, computer scientists would need to recreate human consciousness in AI form.
While these four types represent the different levels of sophistication within AI, there are also different domains that represent certain functionality. Machine learning is probably the most well-known term, where a computerised system can take its data to produce a solution to the problem. Deep learning is a more advanced form of this, and Natural Language Processing takes in information from both spoken and written speech.
These types of AI are present in a lot of everyday situations. For example, Amazon Alexa, or the Echo Dot is a voice-activated device that can play music, turn on your games console, control the lights and thermostat and place orders on Amazon at your request. There are also the algorithms that show advertisements based on your Google search history, or even things you’ve talked about recently via Messenger or IRL conversations in the vicinity of your smartphone.
AI and data in life sciences
One of the ways that digital technology has revolutionised the life sciences industry is through moving on from paper-based to software-based systems. This means that records are stored on a central system and can be accessed and shared much easier from multiple locations. Of course, there is a whole host of other functionality like managing risk and improving existing processes and systems with solutions like quality management software. AI systems can build on this to analyse the data collected and act as a diagnostic tool, identify links relating to disease progression and provide insights into how a patient might react to certain treatments.
From a research point of view, AI can help to identify and analyse a researcher’s findings at a faster rate which could speed up the process of discovering new drugs or uses for existing medicine. This also has the advantage of reducing the risk of oversight, as well as reducing the costs involved. Ultimately, this gives medical laboratories greater resource to strengthen their response to disease and the development of treatments.
Where are we now?
Despite the advances in technology in recent years, there are relatively few examples of machine learning systems operating in medical labs. For example, CellaVision DM96 which performs automated digital cell morphology [2] and the Accelerate Pheno system which is a diagnostic tool [3]. Like anything else, these systems need to be tested, validated and approved by the FDA/MHRA for use in medical labs which is a time-consuming process.
It seems we are just at the beginning in the world of AI within medical labs, and as these types of technologies develop, we could be seeing them used more over the coming years. This is going to require a leap of faith from organisations and a willingness to try something new, but this will be worth it in their goal to improve patient treatment.
[1] Science Daily, Artificial Intelligence: https://www.sciencedaily.com/terms/artificial_intelligence.htm
[2] Sysmex, CellaVision DM96: https://www.sysmex.com/us/en/Products/Hematology/CellImageAnalysis/Documents/Brochure_DM96.pdf
[3] Accelerate Diagnostics: https://acceleratediagnostics.com/products/accelerate-pheno-system/#features
Deborah Nicholson
New business and Enterprise sales for governance, risk and compliance software solutions with over 10 years working in the global Life Science sector. Deborah specialises in the pharmaceutical sector as well as contract manufacturing, CROs, university research, biotechnology, vets and rural sectors.