Monitoring scientific studies and brand-owned or sponsored web channels for the impact of drugs is already mandatory as part of pharmacovigilance. Extending the same rigour across the web as a whole is the obvious next step, with a view to ensuring that nothing is missed once products move beyond trials and onto the market. It is thought that 10- 17% of adverse events currently go undetected because companies are not ‘listening’ in the right digital places. Effective web and social listening could also pave the way to better clinical trials.

But how can companies achieve complete internet vigilance reliably and efficiently? That includes Twitter and public Facebook posts, independent patient forums, blog articles on WordPress, and experiences shared via YouTube and Pinterest – anywhere in the world, and in any language.

This is big data on an unthinkable scale. Even the best-resourced safety team could not hope to scan diligently and report on all of those data feeds using existing tools. Especially when new data is being added minute by minute, 24/7.

So far the life sciences industry hasn’t found a definitive way to overcome this challenge, despite high levels of concern (web monitoring was a hot topic at the recent DIA EuroMeeting in Glasgow, UK). The main options have been to buy into very expensive proprietary turnkey analytics solutions such as IBM Watson Analytics, or patch something together from existing, generic tools. Both approaches fall short of what life sciences firms need.

Extraordinary data volumes demand a smarter approach

With so many channels to keep track of, a more intelligent and focused approach is required. This is where artificial intelligence (AI) comes in.

Using natural language processing algorithms and artificial intelligence, it’s now possible to sift and clean data, reducing irrelevant or false positive content by ensuring signals match specific criteria. The ability to interpret natural human language and semantics mean the technology can identify mentions in context, and read into the subtext to determine how relevant they are. As teams interact with and classify data, machine-learning algorithms adapt to their preferences, honing subsequent findings.

Filtered, meaningful data is pushed to users who easily can share findings and feed important adverse event findings into regulatory processes for urgent action – in near real time. The latest technology is capable of returning accurately-filtered findings from entire global web and social media searches within just 90 seconds – and red-flag events can be escalated to supervisors just as quickly. Humans could never match that!

AI-based web and social listening can also help ensure that nothing critical slips under the radar. Where solutions come with pre-loaded algorithms for life sciences, levels of accuracy start high (above 80%) – even before the software has been trained in what’s of interest to a particular team. That’s a massive head start.

Tighter controls allow the freedom to innovate

AI can also make sure that firms adhere to strict rules on patient privacy – tracking when relevant mentions are subsequently deleted by the poster, for example, so that compliant processes can be applied.

Once companies have seen the potential, they start to realise the scope for additional applications for AI. With consent, they could start to be more proactive in monitoring patient user groups, for instance, about diseases they are interested in – to get a clearer picture of where they stand in the market and where needs are not being met.

Feedback could also help design more rounded clinical trials, using patient anecdotes and measurements from connected devices to look out for early signals, so that in time, companies are able to predict adverse events and intervene before they happen.

More immediately, intelligent web and social monitoring could help companies keep track of regulatory changes, and shifting deadlines. Or monitor how a brand and product is perceived in the market – priceless insight for product development and marketing teams.

There is no reason why companies can’t get started with all of this now. The technology is there, and AI-based data analytics is becoming increasingly accessible and affordable, thanks to open source solutions delivered via the cloud – and which can be run as a managed service on companies’ behalf.

To maximise the impact, life sciences firms need to ensure they track everything holistically, rather than in silos. There should be efficient workflow, feeding straight into established systems and process, i.e. existing pharmacovigilance systems and recognised regulatory procedures. Again, all of this is possible today.

The medical profession has already accepted AI’s potential – in accelerating diagnosis, for instance, so that complex cases and rare diseases can be spotted and treated much earlier. AI’s potential to transform insight and proactivity in life sciences is much the same. It could be just the breakthrough the industry needs at a time when data demands are peaking and specialist teams are feeling the strain.