PreScouter recently organized a webinar that provided a comprehensive exploration of the transformative role of artificial intelligence (AI) in pharma and drug development.
Led by PreScouter’s Technical Directors, Dr. Maikel Boot and Dr. João Guerreiro, the discussion highlighted the rapid adoption of artificial intelligence in pharma for drug discovery, clinical trials, and decision-making processes. Esteemed thought leaders including Amir Emadzadeh, Kyle Tretina, Punitee Garyali, and Veysel Kocamam joined the conversation.
The experts unanimously agreed that AI has diverse applications in the pharmaceutical field. These include computer vision for image analysis, machine learning models for clinical trial feasibility analysis, and natural language processing for text data extraction and analysis.
These advancements have immense potential to improve efficiency, reduce costs, accelerate drug development timelines, and enhance patient outcomes in the field of artificial intelligence in pharma.
The experts also delved into topics such as measuring Return on Investment (ROI), effective implementation strategies, and the vital importance of collaboration between pharmaceutical companies and AI specialists.
Here is the complete webinar if you would like to listen to the full discussion.
Applications of AI in the Pharma Industry
Three key categories of AI applications in the pharmaceutical industry have demonstrated significant promise:
1- AI-powered Computer Vision Applications:
AI-powered computer vision applications show significant potential in image classification and segmentation, especially in oncology and ophthalmology. They enable the analysis of medical images for tumor identification and characterization, as well as the diagnosis and monitoring of eye conditions.
2- Machine Learning and Bayesian Models:
Machine learning and Bayesian models provide feasibility analysis for clinical trials, encompassing tasks such as site selection, investigator ranking, and patient enrollment forecasting. By leveraging operational and patient-level data, these models optimize study focus, identify suitable trial sites, improve patient enrollment, and address recruitment challenges that lead to significant financial losses. For example, clinical trials can cost US $10 m a day in the U.S.
3- Natural Language Processing Applications:
Natural Language Processing applications, or NLP applications, using LLMs can extract insights, summarize text, and analyze sentiment in the context of artificial intelligence in pharma. In pharmaceutical domains involving text data, such as protocols and standard operating procedures, NLP enhances data analysis, decision-making, and information extraction from large volumes of textual information.
Advancements and promising areas:
The application of AI in the pharmaceutical industry goes beyond the mentioned categories. Significant progress has been achieved in understanding biological networks, encompassing data integration analysis, network reconstruction, visualization, and target identification.
Moreover, AI techniques such as reinforcement learning and generative models have aided in de novo drug design, allowing the generation of molecular structures with precise attributes like efficacy and safety. Drug repurposing is also gaining attention as a valuable area of investigation, further expanding the potential of AI in drug development.
Impact of AI/LLMs on cost and time savings
Despite technological advancements, the process of drug discovery and bringing innovative drugs to the market has become slower and costlier over time. While there have been significant technological improvements such as high throughput screenings and computational drug designs in the past decade, it still takes approximately 15 years and $2.8 billion to bring an innovative drug to market. Areas with untapped potential include:
- Accelerating Clinical Trials: AI applications, like AI-powered patient enrollment tools, help overcome challenges related to inadequate patient participation in clinical trials. By reducing the time required to enroll patients, aiding biomarker discovery, and developing companion diagnostics, AI improves the efficiency of oncology trials.
- Enhancing Diversity in Clinical Trials: AI analytics and proactive planning can improve patient recruitment strategies, ensuring a more diverse patient population. By diversifying patient cohorts and recruiting investigators from different backgrounds, efficacy, and safety can be enhanced while ensuring broader representation.
Measuring ROI and convincing stakeholders
Measuring ROI is a crucial aspect to consider. Given the long-term nature of internal investments in the pharmaceutical industry, it becomes challenging to demonstrate short-term returns and convince stakeholders. A positive observed trend is that key decision makers, who previously lacked technical backgrounds, now have a basic understanding of the capabilities of LLMs. This understanding simplifies the communication between technical experts and decision-makers at the executive level.
- Building Trust through Proof of Concepts (POCs): Creating rapid POCs using LLMs showcases immediate results, piquing curiosity and enthusiasm among stakeholders. User-friendly interfaces and open-source community support make AI implementation more accessible, highlighting significant cost savings and improving decision-making processes.
- Identifying Primary Business Outcomes: Clear objective definitions and the establishment of measurement methods for ROI metrics are crucial. Data quality, governance, and privacy considerations are necessary steps. Cross-functional collaboration and communication of benefits help build trust and demonstrate a well-defined plan for maximizing the value of AI implementation.
Successful implementation of AI in pharma and biotech
Three key factors should be considered when implementing AI capabilities in-house in the pharmaceutical and biotech industry to ensure success:
- Establishing In-house Software Teams: Building in-house software teams ensures data privacy and security. Software product development practices enhance the efficient utilization of AI capabilities, allowing organizations to maintain control and ownership over their data.
- Addressing Stakeholder Concerns and Managing Change: Engaging with stakeholders and highlighting that AI enhances efficiency instead of replacing human roles helps overcome resistance to change. Communicating benefits and seamless integration with current practices encourages stakeholders to gradually embrace AI technologies.
- Collaborative Approaches and Future Outlook: Collaboration between big pharma and AI specialists is crucial for leveraging AI expertise while ensuring a deep understanding of clinical and regulatory perspectives.
The pharmaceutical and healthcare landscape has undergone rapid changes in recent years. On a broader scale, certain companies have been actively developing in-house expertise in this field. Novartis and Microsoft collaborated in 2019 to establish their AI capabilities, followed by various companies acquiring biotech firms.
For example, Genentech Roche’s acquisition of Prescient Design in 2021 and BioNTech acquiring InstaDeep. Silicon Medicine’s recent collaborations with Sanofi and Novo Nordisk, as well as Recursion’s partnership with Sanofi, are notable instances.
These examples indicate a significant increase in partnerships, suggesting a potential rise in merger and acquisition activities in the near future, and highlighting the ongoing evolution in the pharmaceutical and healthcare landscape.
The integration of AI in the pharma industry is having a transformative impact on drug development and decision-making processes, showcasing the power of artificial intelligence in pharma. With advancements in computer vision, machine learning, and natural language processing, there are abundant opportunities to enhance efficiency, reduce costs, and improve patient outcomes.
