The Potential Of AI In Drug Discovery

The Potential Of AI In Drug Discovery

Sharing Knowledge

The Potential Of AI In Drug Discovery

Artificial intelligence is dominating discussions due to its potential to revolutionise certain areas of the pharma industry Paul O'Reilly at Sonrai Analytics answers our questions on if the drug discovery and development field will see any of these potential benefits?
Subscribe to stay up to date with the latest Sonrai content.
Paul O'Reilly

Paul O'Reilly

Head of Innovation at Sonrai

As a leader in the field of artificial intelligence (AI) and machine learning (ML), Paul is responsible for developing cutting-edge novel AI applications as proof of concept that add real value to Sonrai's customers. In his previous roles, Paul has significantly impacted the field of AI and ML. He led a team at Philips DCP, Belfast, where they innovated and developed AI-based solutions for tumour detection and quantification, and biomarker scoring in digital pathology – resulting in the creation of the TissueMark product, one of the earliest applications of deep learning in the field. He also worked as a Bioinformatics researcher at Queen's University Belfast, developing ML applications for gene expression analysis.

Some of the key benefits of integrating AI into drug discovery processes

EBR: How is artificial intelligence (AI) revolutionising drug development and addressing its key challenges, such as accelerating the process and increasing efficiency?

Paul O’Reilly (POR): AI refers to a set of data-driven approaches to recognising patterns, finding correlations and ultimately deriving insights from datasets, across a range of modalities, from numerical and time series data, through language models and image analysis and beyond. Building an AI model of a particular dataset (or datasets) allows a variety of novel approaches to realise value from those data. AI can allow automation of tasks which are commonly carried out manually, and can increase the throughput and efficiencies of those tasks.

Generalizable AI can allow predictive models to be developed (e.g. predicting the response to a particular treatment) for efficacy or safety, optimize clinical trial design (1) or enable mining of large databases such as publication databases or compound libraries.

"AI tools allowed a pharma company to reduce the time needed to identify three preclinical candidates to between 12 and 18 months, compared with the three to five years typically required by traditional players who do not use AI."

BCG found that AI tools allowed a pharma company to reduce the time needed to identify three preclinical candidates to between 12 and 18 months, compared with the three to five years typically required by traditional players who do not use AI (2).

According to a 2024 report by McKinsey & Company, AI applications in drug discovery significantly enhance the efficiency and speed of R&D processes. For example, generative AI can rapidly screen and optimize chemical compounds, leading to a more than fourfold increase in the speed of identifying new leads (3)​.

The Difference Sonrai Makes

Every organization is unique, and generic solutions fall short in addressing their distinct challenges. Sonrai is purpose-built to tackle the complexities of precision medicine, offering tailor-made solutions that align with your specific needs.

Discover Predictive Biomarkers 52% Faster

Sonrai’s AI technology supports rapid biomarker identification, empowering our partner to predict optimal therapy responses through an innovative toolkit.

57% Acceleration in MoA Studies

See how Sonrai unified diverse data sources to rapidly understand complex molecular pathways, helping identify predictive biomarkers and drug mechanisms of action.

14% Reduction in Preclinical Phase

Discover how Sonrai helps pharmaceutical companies overcome challenges in Invivo Drug Efficacy with AI-powered analytics.

EBR: What types of data sources and data sets does AI utilise for drug discovery, and how does this aid in identifying potential drug candidates and disease targets?

POR: There are a wide variety of data types which are typically collected during the drug discovery process. These may include genomic data, proteomic data, Electronic Health Records (EHRs) and imaging data such as MRI/CT scans, pathology images (brightfield and fluorescence) or High Content Screening (HCS) images. Increasingly, studies are collecting data such as single cell RNA-Seq and spatial transcriptomics/proteomics which are amenable to analysis using AI. There are various sources for such data, from cell lines, animal models, patient derived xenografts, and real patient data, from preclinical and clinical studies. On the other side, there are a large number of molecular, chemical and publication databases, which may be used to identify patterns in the specific data collected by the drug developer.

"AI’s ability to handle and integrate multiple modes of data (clinical data, sequencing, images) means that more ‘holistic’ biomarkers can be developed in the future, with increased sensitivity and specificity."

EBR: In what ways does AI contribute to precision medicine, enabling tailored drug treatments and patient-centric approaches?

POR: Precision medicine is enabled by the ability to identify subgroups of the patient population for a disease (or patient stratification) using particular biomarkers to identify the patients who can most benefit from a particular treatment. AI is increasingly being used in the identification of such biomarkers, and in their cost-effective delivery at scale within the clinic. For example, AI approaches to quantification of molecular status from pathology images, allows consistent, accurate and fast stratification of patients with cancer, and the determination of the treatment which has the best chance of success for any particular patient. AI’s ability to handle and integrate multiple modes of data (clinical data, sequencing, images) means that more ‘holistic’ biomarkers can be developed in the future, with increased sensitivity and specificity.

EBR: What kind of technological infrastructure is required to support AI-driven drug discovery and what is the financial impact of integrating AI into current processes?

"The use of cloud data management platforms and deploying AI services in a ‘cloud native’ manner can allow organisations to leverage AI in a cost-effective, secure and flexible manner."

POR: The infrastructure requirements for applying AI to drug discovery are broad and inherently related to the particular problem being solved. For complex molecular or protein modelling, this will often require access to a high performance computing (HPC) environment, whereas for other tasks the more scalable approach of utilising cloud computing will be appropriate. With modern cloud architectures enabled by microservices, systems can be scaled flexibly, and storage and compute costs can be optimised for a particular user’s needs. This all comes at a cost, particularly HPC, but the use of cloud data management platforms and deploying AI services in a ‘cloud native’ manner can allow organisations to leverage AI in a cost-effective, secure and flexible manner.

If you'd like to talk to Paul or another expert from our team, reach out to us

Contact our friendly team for expert guidance and transformative insights.

EBR: What are the major challenges faced by the pharmaceutical industry in adopting AI, and how can these obstacles be mitigated? What role do AI firms play in mitigating these?

POR: Given that the basis for most AI is data, often highly confidential proprietary and/or patient data, one major challenge is how to store and manage the large amounts of data in a secure manner, allowing access to approved stakeholders while maintaining data integrity. Another challenge is that data within organisations is often disaggregated in data silos, stored in multiple locations, accessed by disparate groups of stakeholders, often unversioned and held in a variety of different formats. Since the foundation of AI approaches is highly dependent on data, stakeholders need to standardise and formalise their data management processes, and control access to that data in a role and project-based way.

Apart from the data, there are budgetary and organisational challenges to an organisation wanting to adopt AI. There is undoubtedly an element of investment (in technical/infrastructure capability and people) required, so companies should mitigate against the risks associated with AI projects. This may entail hiring expertise with particular experience in developing and deploying AI solutions within their particular domain, and/or engaging with external companies specialising in AI and its delivery. It is also important to define the particular scope of AI projects and quantify the cost/benefit ratio, in order to identify where AI can have the biggest impact.

"According to McKinsey & Company, biopharma companies should focus on balancing between internal capability building and partnerships with AI drug discovery companies."

Finally, given the ultimate goal of drug discovery is to deliver benefit to patients, it is most important that adopters of AI are aware of the regulatory landscape for their use of AI. Depending on the ultimate use of the AI system, this may be in the form of a clinical decision support tool or an in vitro diagnostic (IVD). There are multiple regulatory contexts, from the generally applicable EU AI Act, to the specific requirements for AI devices as IVDs, and these are continually evolving as the FDA and other regulatory bodies come to terms with the fast-evolving field of AI in medicine. This is a challenge for regulatory experts within those organisations and their partners/suppliers, and engaging with other regulatory experts and bodies can help an organisation stay abreast of latest developments.

According to McKinsey & Company, biopharma companies should focus on balancing between internal capability building and partnerships with AI drug discovery companies. Biopharma can benefit from the partnerships through accessing AI technology, algorithms and infrastructure, data, teams of experts such as data scientists, and data protection to ensure regulatory compliance (4).

EBR: How does AI facilitate interdisciplinary collaboration among researchers, data scientists and clinicians in drug development?

POR: As mentioned above, AI requires an organisation to have systems and processes in place to manage their valuable data. One consequence of this is that such systems and processes also enable better sharing of the data among stakeholders, with the consequent enhancement of collaboration – with a single source of information, visualisations and results from the AI systems, the risks of miscommunication between stakeholders can be mitigated.

With the multimodal capabilities of AI, the AI systems can act as a means of integrating insights from multiple disciplines and experts, providing a single point of access to those insights and mapping onto the problem domain, in this case drug discovery.

EBR: What is the outlook for the future of AI in drug discovery, including potential innovations, ethical considerations, regulatory compliance and industry partnerships?

POR: Pharma increased their AI investment from less than $1 billion in 2015 to over $7 billion in 2021 (4). Generative AI is expected to produce $60 billion to $110 billion in annual value across the pharmaceutical industry value chain (3). AI is a continually evolving domain, as the relatively recent surge in Large Language Models such as ChatGPT has shown. There are a large number of researchers continually looking to extend the state of the art in AI, particularly in language processing, image analysis and multimodal integration. Nothing stays still in AI for very long! As a result, the regulatory bodies are continually refining their views of AI, but as they start to come to terms with the field, we can see the regulatory frameworks firming up and becoming more mature. There are of course ongoing considerations of patient consent for their data (in its many forms) to be used to build AI models, and interesting questions about how much the ‘encoding’ of patient data within AI models conflicts with the ‘right to be forgotten’ or the right to withdraw consent.

"There are of course ongoing considerations of patient consent for their data (in its many forms) to be used to build AI models, and interesting questions about how much the ‘encoding’ of patient data within AI models conflicts with the ‘right to be forgotten’ or the right to withdraw consent."

One thing is certain is that while drug developers can build the AI capabilities internally, the quickest, safest, most cost-effective means of deploying AI is often in partnership with AI experts.

Build Versus Buy: Multi-Omic Platform

Managing the abundance of multi-omic data in precision medicine is challenging and complex. Click the button to access the white paper on building or buying a platform. The white paper also includes an interactive guide to help you understand your specific needs and which option may be best for you.

References

  1. Aliper, A., et al (2023). ‘Prediction of Clinical Trials Outcomes Based on Target Choice and Clinical Trial Design with Multi‐Modal Artificial Intelligence’, Clinical Pharmacology & Therapeutics,114(5):972-980.
  2. BCG (2022). Adopting AI in Drug Discovery. https://www.bcg.com/publications/2022/adopting-ai-in-pharmaceutical-discovery Accessed: 24/04/2024
  3. McKinsey & Company (2024). Generative AI in the pharmaceutical industry: Moving from hype to reality. https://www.mckinsey.com/industries/life-sciences/our-insights/generative-ai-in-the-pharmaceutical-industry-moving-from-hype-to-reality Accessed: 24/04/2024
  4. McKinsey & Company (2022). AI in biopharma research: A time to focus and scale. https://www.mckinsey.com/industries/life-sciences/our-insights/ai-in-biopharma-research-a-time-to-focus-and-scale Accessed: 24/04/2024

Discover how companies worldwide grow with Sonrai. Explore all our client use-cases and case studies.

Flow Cytometry Analysis

Explore the cutting-edge world of biomarker discovery using flow cytometry data with Sonrai Discovery technology.

AI Biomarker Discovery

Explore an end-to-end AI biomarker discovery workflow using Sonrai Discovery. Discover just how easy it can be.

IVBH Early Cancer Detection

Discover how IVBH collaborated with Sonrai Analytics to revolutionize early lung cancer detection.

Read More
Get in touch

Like What You See? Let's Talk

No hard sales conversations
We Listen to your problems
We give you confidence to make your decision
Related Posts
Leave a Reply