Honing the sharp-shooter approach: how a legacy drug discovery pathway can be redefined through digital methods

For over a century, drug discovery has relied on a simple logic: take enough shots and eventually one will hit the target. This high-throughput, legacy mentality originates from the wet-lab techniques that discovery pathways have traditionally relied upon. 

But the ground is shifting: AI-designed vaccine components even commenced human testing while the iterative reviews of this very article took place [1]. This reflects a wider question facing pharma R&D: whether the current business model remains fit for purpose in an environment shaped by technology progress, cost pressure and changing expectations of evidence. 

Within discovery, this raises a more specific challenge: do the same processes remain the best way to deliver both scientific and commercial impact? 

Technological innovation has long supported drug discovery: from X-ray crystallography and flow cytometry to AlphaFold protein structures and neural network processing. But implementing these techniques as digital ‘add-ons’ continues the ‘bio-tech’ mentality - its continued relevance in an increasingly digital age is questionable.  

Each successful new drug is accompanied in its journey to market by failed cousins through this ‘scatter-gun’ approach: the existing challenge is in the number of failed candidates and the late stage at which they fail. Improving discovery accuracy will provide a cost-efficient and strategic advantage to companies who are willing to re-evaluate and ‘digitally enhance’ their operating model and decision-making processes.  

For example, given the cost of failed development programmes, bringing a single new medicine to market can cost approximately $2.6 billion [2]. However, this figure is likely much higher. The pharmaceutical industry is cited to annually invest approximately $250 billion in development, with only 60 new drugs getting approved - suggesting that costs encompassing failure rate are double that figure [3,4]. A re-evaluation of the drug discovery process could reduce the escalating costs of these high-throughput processes, while minimising the risk of late-stage failure in a new balance between in silico and in vitro techniques. 

What this means for the industry: the new operating model

Digital advancements can provide the ‘autobahn’ of highways for the future of drug discovery. Maximising the quality of predictions with fewer but better-informed candidates would create a validation-first approach, where accuracy is prioritised over scale and through-put. This would move the discovery process away from a probability-based approach, instead honing predictive accuracy with digital methods.  

So far, most companies continue to rely on familiar late-validation models, adding digital capabilities as compensation mechanisms rather than fully re-designing their approach. Digital bolt-ons provide limited benefits when compared with fundamental integration. In fact, it can amplify persisting late-stage failure: taking the same shots towards the target faster, means faster off-target hits, too. 

Access to complex patient datasets, automated and predictive analysis, and simulated disease modelling will continue to enable work in minutes that previously took years. But to fully embrace the digital shift, discovery operating models would need to fully adjust their focus to computer-generated predictive discovery, supported by in silico validation and later followed by in vitro, pre-clinical, and clinical exploration. The addition of earlier decision-gates throughout this process would allow for a flexible approach through the digitised workflow. 

Automation, enhanced workflows, advanced analytics, and remote operations could all support this future-state. However, the value comes not from digital tools alone, but from embedding them into the rhythm of scientific decision-making. Oaklin’s experience in digital transformation, AI-driven assets, and operating model design can help organisations move away from the traditional waterfall R&D and siloed handovers towards agile ways of working; with earlier decision gates and faster test-learn cycles, without compromising the integrity of the underlying science.