Machine learning offers considerable potential for streamlining and transforming routine information processes in life sciences — but could more creative applications of artificial intelligence take things further still? Amplexor's Siniša Belina assesses the opportunities.
AI
The scope for artificial intelligence in life sciences is potentially significant — enabling accelerated scientific breakthrough, thanks to new potential to identify the minutest anomalies in unwieldy global data masses, and facilitating greater drug personalisation.
Far from hovering on some futuristic horizon, the technology is already available too — which means pharma companies need to build it into their agendas and plans now, if they don’t want to risk sacrificing competitive advantage. From intelligent internet and content searches that adapt to user preferences, to automated personal assistants like Alexa and Siri, and customer care channels such as web chat, AI is already in common use as part of people’s routine activities.
One of the great appeals of AI is the scope for process automation and acceleration, using clever algorithms that can complete complex tasks previously limited to human capacity. The technology is also highly adaptive — machine learning tools can be guided to respond to the conditions they are exposed to and the results they find, so that they get better and better at the job they’ve been tasked with. So over time they are able to absorb more and more human tasks, to the point that the only intervention needed is in the form of supervisory quality checks.
The human-machine continuum
Although countless articles have been written about the threat to people’s jobs presented by AI, if machines can get to grips with routine knowledge work and do it more rapidly than humans ever could, why wouldn’t companies want to take advantage — especially if it allows them to free up experts for more advanced, value-added work?
AI technology offers to transform patient outcomes — as is already being seen in frontline patient diagnostics. UK researchers in Oxford recently announced the availability of AI technology that can diagnose heart disease and lung cancer at a much earlier stage from analysis of patient scans.1 Meanwhile, connected devices are being used increasingly to transmit patients’ data to those managing their care — to enable earlier interventions.
In pharma, machine intelligence has substantial potential for enhancing R&D, through the ability to analyse large volumes of data leading to richer insights. To this end, applications, systems, and platforms have already been developed to transform clinical trial innovation. This isn’t just about distilling subtler patterns from once unmanageable volumes of disparate data either. It is also about modelling and extrapolating from such findings to arrive at bolder hypotheses and deeper and more targeted work, to accelerate discoveries and the development of treatments.
Scaling data mountains
Machine intelligence at scale also offers a viable means to track global patient trends, concerns, experiences, behaviour and needs, enabling the life sciences industry to understand what is happening in the real world — to a degree that hasn’t been possible previously. This offers potential not only for more proactive and thorough monitoring of adverse events and other safety signals as drugs move into markets, but also for identifying emerging requirements, triggering new innovation.
Where the life sciences industry has traditionally been one step removed from patients, public internet forums and social networks offer an opportunity to understand evolving demands and engage with patients in new ways. AI is already proven for social media monitoring in other markets, whose best practices could be carried across to life sciences with a few adjustments.
For all the excitement around AI, though, the life sciences industry is not exactly known as an early adopter of new technology. So, there are a number of things that need to happen if companies are to adapt to and exploit the potential ahead of them.
Testing AI’s potential
Once firms have accepted that change is coming, the next step is to prepare an IT and data environment which allows for new experimentation and insights — within the restrictions of regulatory control and privacy protection.
This isn’t just about developing ‘big data’ strategies, but rather preparing that data so it can be analysed efficiently, accurately and holistically using AI platforms — to spot emerging trends, anomalies, concerns and opportunities in a very efficient and granular way.
For now, regulatory pressures are driving most data-related initiatives in life sciences. So, this is a good place to start with AI — even if just for taking over some of the more repetitive or preparatory stages of submission creation, or content checking, to accelerate speed to market. (Using machine learning, systems could ‘learn’ how to produce better output, or the conditions most likely to result in a new marketing submission being accepted first time.)
If data preparation work has to be done to fulfil regulatory demands, why not exploit this — laying the foundations for future innovation, and testing just how powerful AI can be in transforming everyday processes?
The critical enabler for maximising the potential of data is the creation of a comprehensive master data model — one that also includes inter-dependencies between the data, in a way that can drive new efficiencies and increased impact through proactive process automation, boosted by AI/machine learning. Today, just about every life sciences business is striving to be more agile, responsive and patient centric. But this relies on a strong sense of purpose and a foundation of rich, ready-to-exploit data. So, if companies are going to start anywhere, this is as good a place as any.