Digitising Drug Discovery: How AI/ML Is Transforming Target Identification and Screening

The life sciences industry is undergoing a major evolution, driven by artificial intelligence (AI) and machine learning (ML). Nowhere is this more apparent than in drug discovery, particularly in target identification and screening, where these technologies are helping to cut time, cost, and attrition.

Why Now? The Industry Context

Developing a new drug traditionally takes 10–12 years and can cost upwards of $2.6 billion. From early-stage target validation to Phase III trials, the journey is slow, expensive, and often ends in failure. The introduction of AI and ML into this pipeline is a game-changer, offering new ways to interpret data, accelerate decision-making, and increase the odds of success.

The Problem with the Old Way

Historically, drug discovery has been reliant on manual processes and siloed systems. Screening potential compounds has often been laborious, with research teams navigating fragmented data and time-consuming experiments. This approach leads to long development cycles, high attrition rates, and costly late-stage failures, particularly frustrating after years of investment.

Real-World Examples of AI in Action

Several AI-driven biotech companies are already demonstrating what's possible:

• Insilico Medicine used AI to design a novel drug in just 46 days, reaching Phase I clinical trials in under 30 months, an unprecedented timeline in the industry.

• Exscientia cut early-stage development timelines and, in a 2023 clinical trial, achieved a 54% increase in tumour control duration using AI-optimised molecules.

• XtalPi developed predictive chemistry platforms for partners like Pfizer, delivering accurate molecule structure predictions and significantly reducing preclinical guesswork.

A World of Opportunity

AI brings two compelling pathways for innovation:

• Off-the-shelf AI models offer a fast and scalable route, ideal for teams looking to enhance early-stage discovery.

• Bespoke platforms provide tailored, domain-specific models with greater IP defensibility and precision performance.

For organisations with the right expertise and infrastructure, these options open the door to smarter, faster, and more commercially viable R&D.

The Risks to Watch

Despite the promise, AI is not a silver bullet. Poor implementation can result in:

• Opaque, non-interpretable models

• Poor reproducibility of results

• Regulatory challenges

• Misallocated investment in unsuitable tech

A clear strategy, robust data governance, and scientific oversight are essential to avoid these pitfalls.

The Road Ahead, and How XPS Helps

At XPS, we work closely with biotech and pharma teams to integrate AI and ML in a way that is both scientifically sound and commercially scalable. Whether you’re exploring cloud-based AI pipelines, implementing open-source ML tools, or building proprietary models, we help ensure:

• Proper governance and data hygiene

• Cross-functional collaboration between science, tech, and compliance

• Flexible, future-proof integration strategies

We believe the next wave of blockbuster drugs won’t be found, they’ll be designed, and AI will be central to that transformation.