ORIGINAL RESEARCH

Mapping the AI Landscape in Life Sciences: A Framework for Evaluation and Adoption

Jennifer Hinkel, MSc^1,2 ; Fraser Peck, MBBCh³ and Cory Kidd, PhD¹

¹Sigla Sciences, Incline Village, NV, USA; ²Kellogg College, University of Oxford, Oxford, UK; ³Pharmacy Consulting Limited, Farnborough, UK

Keywords: artificial intelligence, clinical development, commercialization, ethical challenges, life sciences, regulatory, strategic model

Abstract

Background: The adoption of Artificial Intelligence (AI) is accelerating in the life sciences sector, offering opportunities to enhance biopharmaceutical research, development, and commercialization. However, the sector lacks structured tools to prioritize AI initiatives across varied business domains.

Objective: This study presents a framework to categorize and evaluate potential AI applications in the life sciences industry, organizing use cases along two critical dimensions: Phase of Product Lifecycle and Operational Domain.

Methods: A structured mixed-methods approach was employed, including a modified Delphi consensus process with industry experts, a qualitative case study review, and iterative framework refinement between August 2023 and August 2024.

Results: The resulting matrix framework enables life sciences professionals to assess AI opportunities across research, clinical development, commercialization, and post-marketing activities. Key findings highlight the pervasive nature of AI impact, the emphasis on data-driven strategies, and the regulatory and ethical challenges facing biopharma firms.

Conclusions: This framework provides a practical model for strategic AI adoption decisions within the life sciences sector and lays the groundwork for future research, policy development, and enterprise transformation efforts.

Plain Language Summary

Artificial Intelligence (AI) is starting to reshape how pharmaceutical companies discover new drugs, test them in clinical trials, and bring them to patients. However, with so many possible uses for AI, it can be hard for companies to decide where to begin and how to prioritize. This article presents a simple framework that organizes potential AI projects by the stage of drug development, such as early research or post-market monitoring, and by how much the project involves people outside the company, like doctors or patients. The goal is to help life sciences companies think more clearly about where AI could add value and what risks to monitor. With thoughtful planning, AI can make biopharma companies and the healthcare system they support, smarter, and more focused on patient needs.

 

 

Recent advances in Artificial Intelligence (AI) are prompting discussions across various scientific and business sectors regarding where, how, and why AI technologies might be deployed, and what the implications will be. The medical and life sciences fields are exploring how AI might be incorporated throughout the drug development process, from discovery to patient care, including through commercialization, marketing, and market access strategies.¹

As of mid-2024, the AI adoption in the life sciences industry has progressed from theoretical discussions to practical implementations, albeit many in the ‘pilot phase’. stage. Clinicians, pharmaceutical manufacturers, and policymakers are currently evaluating whether these technologies offer substantive opportunities for innovation, enhanced efficiency, and improved patient outcomes or if they are subject to inflated expectations characteristic of a ‘hype cycle’.²

AI represents a range of technologies, including large language models, machine learning, natural language processing, and computer vision.³ Some researchers propose that AI and related technologies represent significant shifts in how researchers and clinicians approach complex problems in life sciences and healthcare.⁴

Some potential AI applications hold particular interest to biopharmaceutical manufacturers because they promise to reduce the time and/or investment involved in bringing new therapies to market. In the earliest stages of drug discovery, AI algorithms aim to speed the identification of new therapeutic targets and potential drug candidates.⁵ In clinical development, AI might optimize study designs, improve patient recruitment, or enhance data analysis.⁶ Post-commercialization, AI-driven tools may be deployed to enhance patient engagement or medication adherence or to automate the collection of real-world evidence (RWE).⁷

While AI technologies might be promising, they also present several challenges to the industry, ranging from ethical, legal, regulatory, and data security issues to more extensive concerns that organizations might need to change long-standing operating procedures as technologies advance.⁸ Compounding these issues is the pace of advancement. Examples include the rapidity of AI development far outpacing the ability of regulatory and legal frameworks to keep up.⁹ This asymmetry adds significant complexities in a highly regulated industry that is notably averse to legal risk.

Despite the risks, the potential benefits of ‘leading the pack’ on AI adoption might be significant. Companies that successfully integrate AI into their operations might gain competitive advantages, from accelerated research and development to more optimized market access strategies. Conversely, those who fail to bring AI on board may increasingly miss out on growth opportunities.¹⁰

In the light of these challenges and the rapid pace of advancement, the authors aim to provide a framework that professionals across various functions within life sciences companies can use to understand and evaluate the place of AI technologies in their workflows and firms. By aligning AI applications along two key axes that we refer to as Operational Domain and Phase of Product Lifecycle, we offer a structured approach to evaluating and prioritizing AI initiatives. This framework is designed to assist those in positions of leadership with making informed decisions about AI adoption, balancing potential benefits against risks and challenges.

Methods for Framework Development

This study employs a structured mixed-methods approach, including a modified Delphi Method. It draws from the authors’ industry experience, consultations with external biopharmaceutical experts, a qualitative review of case studies from literature and business reports, and an iterative process of elicitation and consensus-building. The research was conducted between August 2023 and August 2024, allowing for the incorporation of recent developments. It is expected to continue iteratively, incorporating elements of realist synthesis to further expand the framework and refine example use cases.

An initial framework was developed based on the collective expertise of the research team and additional consulting experts. These included professionals with extensive experience in pharmaceutical commercialization, clinical development, and AI applications in research and patient-facing settings. This sector expertise guided the creation of the two axes that form the foundation of the framework: Phase of Product Lifecycle and Operational Domain.

Once the axes were established, examples of AI applied to typical tasks within the life sciences industry were identified and mapped onto the framework based on expert discussion and consensus. The identification process involved (1) consulting industry case studies, publications, and expert knowledge, (2) selecting biopharma industry activities indicative of various phases and operational domains with existing or proposed AI use cases, and (3) highlighting examples that could highlight potential benefits and challenges of AI adoption in each section of the framework matrix, given that these examples are most likely to spark discussions on the requirements, risks, benefits, and objectives of AI as applied to specific industry use cases.

The initial framework and use case examples underwent iterative refinement based on the incorporation of additional feedback from industry experts. In summary, we followed these steps (Table 1).

Table 1. Initial framework and use cases based on incorporation of additional feedback from industry experts

Use case	Framework
Step 1	Conceptualization and axis development
Step 2	Expert consultation
Step 3	Framework refinement
Step 4	Identify business activities and map AI use cases to the framework.
Step 5	Iterative improvement as the field advances

This methodology allowed us to create a framework that is grounded in real-world industry understanding and reflects the current state and potential future directions of AI applications in life sciences.

Ethical Considerations

This study did not require formal ethics review, as it did not involve human subjects research, clinical trials, or the collection of personal data. The research was based on publicly available information, the expertise of the core research team, and consultations with field experts. All experts consulted in this study provided consent to share their professional insights and expertise in an anonymous fashion as a contribution to the framework development.

Results

Our research resulted in the development of a proposed framework for understanding and implementing AI technologies in the life sciences sector. This framework aligns AI applications along two key axes: Phase of Product Lifecycle and Operational Domain.

Phase of Product Lifecycle (Horizontal Axis)

This axis aligns approximately with the traditional stages of drug development and commercialization in the life sciences industry. It recognizes that AI might have differential impact at various stages of the drug product journey from concept to market.

Research & Discovery

Research and discovery represent the early stages of drug development before human testing and encompass a range of computational and biological research methods.

Clinical Trials and Regulatory Approval

Clinical trials and regulatory approval represent the clinical development phase of activities that typically start with first-in-human testing and culminate with the completion of registrational clinical trials and regulatory/marketing approval.

Commercial Launch

This phase overlaps here as a parallel process, with planning starting well prior to Regulatory Approval and continuing through that launch date and beyond, and including a range of commercial, medical, and access-related activities.

Post-Market Activities

Post-marketing commercial and scientific activities support pharmacovigilance, the collection of RWE, ongoing communication with healthcare professionals, and continued patient support.

Operational Domain (Vertical Axis)

This axis represents a continuum from internal operations to increasingly external interactions, reflecting not only the diverse operational areas where AI technologies are applied in the life sciences sector but also the growing complexity of regulatory, data, and ethical considerations as we move from internal operations toward patient-facing applications. Internal Operations include activities within the company’s boundaries, including those that deal with proprietary data and processes. While still subject to regulations, these activities carry the least external exposure and thus lower regulatory and ethical risks. B2B Engagements represent the first step from internal activities into external interactions, involving partnerships, suppliers, and other businesses, either scientific or commercial in nature. Here, collaborative processes and data sharing introduce additional regulatory considerations and data protection needs, but with less regulatory overhead than engaging with healthcare professionals or patients.

Healthcare Professional Engagements

These cover interactions with healthcare professionals, introducing significant regulatory oversight due to data protection, transparency laws, ethics, and the impact to patient care.

Patient Engagements

As the last operational domain, patient operational domains represent direct patient interactions and the highest level of external engagement. Patient-facing activities face the most stringent regulatory requirements, data privacy concerns, and ethical considerations given the impact on patient care and individual health outcomes.

Table 2 illustrates this matrix and provides representative examples of AI applications at each intersection of Lifecycle Phase and Operational Domain, showcasing potential use cases for AI technologies across common biopharma business areas.

Table 2. Matrix framework for understanding AI use cases and applications in life sciences

Phase of product lifecycle	Operational domain
	Increasing levels of regulation and risk with more external-facing activities towards the right
	Internal operations	B2B engagements	HCP engagements	Patient engagements
Phases of drug discovery, development, and launch
Discovery and Pre-Clinical Development	Al-driven target identification and validation Machine learning for in silico drug design Al-powered predictive toxicology models	Federated learning platforms for collaborative research Al-optimized partnering selection for CRO, investment, and BD Al-driven compound sourcing and synthesis prediction	Al-curated research insights for KOL engagement Natural language processing for clinical expertise mapping Al-powered scientific conference and KOL matchmaking	Al analysis of patient-generated health data for biomarker discovery Machine learning models for patient stratification in rare diseases Al-driven patient advocacy network analysis
Clinical Development and Regulatory Submission	Al for adaptive trial design and protocol optimization Machine learning for patient recruitment forecasting NLP-powered automated regulatory submission preparation	Al-driven site selection and performance prediction Machine learning for CRO performance optimization Al for optimizing multi-regional clinical trial designs	Al-powered investigator relationship management Virtual assistants for protocol clarification and support Machine learning for investigator site feasibility assessment	Al chatbots for patient screening and enrollment Machine learning for personalized patient retention strategies Al for real-time patient monitoring and adverse event prediction in trials
Commercial Launch and Market Access	Al for dynamic pricing models and market access strategies Machine learning for sales force effectiveness optimization NLP for competitive intelligence and market sentiment analysis	Al-driven payer negotiation simulators Machine learning for distribution channel optimization Al for automating and optimizing RFI and tender processes	Al-powered personalized HCP engagement platforms NLP for real-time HCP sentiment analysis during interactions Al for predicting and optimizing HCP prescription behaviors	Al for personalized patient support program design Machine learning for medication adherence prediction and intervention Al-driven chatbots for patient education on complex therapies
Lifecycle Management and Post-Marketing Activities	Al for signal detection in pharmacovigilance Machine learning for product lifecycle optimization strategies NLP for automated medical literature monitoring	Al for real-world evidence study design and partner selection Machine learning for supply chain risk prediction and mitigation Al-powered forecasting for long-term product performance	Al-driven personalized medical information services Machine learning for predicting and addressing HCP information needs NLP for processing and analyzing HCP feedback at scale	Al for detecting and predicting medication non-adherence Machine learning for personalizing patient support interventions NLP for analyzing patient-reported outcomes and social media data
AI: Artificial Intelligence; B2B: business-to-business; CRO: contract research organization; HCP: healthcare professional; KOL: key opinion leader; NLP: natural language processing.

Key Findings

In parallel with our development of this framework, we identified the following themes as applicable to biopharmaceutical manufacturers considering AI adoption:

Pervasive Impact

AI use cases span across all phases of the product lifecycle and all addressed operational domains. The breadth of potential applications suggests a fundamental shift in how the industry will operate in the future.

Efficiency and Innovation

Many AI applications focus on improving efficiency (e.g. automation of regulatory submissions) or driving innovation (e.g. AI-powered drug discovery). AI applications are frequently framed as tools that will reduce effort, time, and/or cost.

Enhanced Customer Engagement

AI technologies may be leveraged by both commercial and medical field teams, particularly in digital or ‘omnichannel’ marketing initiatives or in medical communications.

Data-Driven Decision Making

Across the matrix, there is a strong emphasis on data sources and data access as inputs to AI models. These data may be internally or externally derived, speaking to the value of high-quality data assets. Access to data assets is likely a competitive advantage or differentiator.

Patient-Centricity

The framework reveals numerous AI applications focused on improving patient experiences and outcomes, aligning with a general shift toward more patient-centric approaches and the need to communicate patient-centric value, especially in the commercial phase. Patient-centricity aligns with value-based care models and evolving regulatory expectations around the incorporation of patient reported outcomes (PROs) in drug development and post-marketing phases.

Interconnectedness

Several AI applications bridge multiple domains or phases, hinting at the potential for these technologies to change industry structures or processes over time. This interconnectedness suggests the need for integrated AI strategies that consider the full spectrum of pharmaceutical operations and stakeholder interactions.

Ethical and Regulatory Considerations

AI uses cases in biopharma that raise numerous ethical and regulatory questions. These include questions around data privacy and security, algorithmic bias, data ownership and consent, and the interpretability or generalizability of AI-driven decisions.

Limitations

Our methodology has limitations, including an acknowledgment that the rapid pace of AI development is likely to outpace even expert knowledge. Another limitation in expert consensus methods is the introduction of potential bias based on the team’s collective experiences and perspectives. We mitigated these limitations to the extent possible through engaging diverse experts and iterating through the framework development process.

Furthermore, this framework is not likely to be exhaustive, given the challenge of comprehensively capturing all possible AI applications in a rapidly evolving field. For example, biopharmaceutical manufacturing companies also perform a number of corporate functions in finance, legal, human resources, training, and similar operational domains considered out of scope in this project. However, there may be areas where AI innovation can also be applied.

Conclusion

This article presents a new framework for understanding and categorizing AI technologies across the life sciences product lifecycle and operational domain. By mapping along two axes representing paradigms well-understood by professionals in the biopharmaceutical industry, we provide a structured approach for such professionals to evaluate and prioritize AI technology initiatives.

Our findings demonstrate the pervasive potential of AI to transform various aspects of the life sciences sector, but not without risks and challenges to navigate, including data quality and availability, algorithmic biases, privacy and security, ethical and regulatory considerations, and the need for organizational change management.

As the field continues to evolve rapidly, ongoing research and industry collaboration will be crucial to fully harness the potential of AI while addressing its associated challenges. The framework presented in this article provides a foundation for these future efforts, offering a common language and structure for discussing and analyzing AI applications in the sector.

Authors’ Contributions

Data Availability Statement (DAS), Data Sharing, Reproducibility, and Data Repositories

Application of AI-Generated Text or Related Technology

All data produced in this study are available upon reasonable request to the corresponding author.

References

1. Mak KK, Pichika MR. Artificial intelligence in drug development: present status and future prospects. Drug Discov Today. 2019;24(3):773–80. https://doi.org/10.1016/j.drudis.2018.11.014
2. Gartner. Understanding Gartner’s hype cycles [Internet]. [cited 2024 Sep 19]. Available from: https://www.gartner.com/en/documents/3887767
3. Yu KH, Beam AL, Kohane IS. Artificial intelligence in healthcare. Nat Biomed Eng. 2018;2(10):719–31. https://doi.org/10.1038/s41551-018-0305-z
4. Topol EJ. High-performance medicine: the convergence of human and artificial intelligence. Nat Med. 2019;25(1):44–56. https://doi.org/10.1038/s41591-018-0300-7
5. Zhavoronkov A, Ivanenkov YA, Aliper A, Veselov MS, Aladinskiy VA, Aladinskaya AV, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors. Nat Biotechnol. 2019;37(9):1038–40. https://doi.org/10.1038/s41587-019-0224-x
6. Harrer S, Shah P, Antony B, Hu J. Artificial intelligence for clinical trial design. Trends Pharmacol Sci. 2019;40(8):577–91. https://doi.org/10.1016/j.tips.2019.05.005
7. Rajkomar A, Dean J, Kohane I. Machine learning in medicine. N Engl J Med. 2019;380(14):1347–58. https://doi.org/10.1056/NEJMra1814259
8. Vayena E, Blasimme A, Cohen IG. Machine learning in medicine: addressing ethical challenges. PLoS Med. 2018;15(11):e1002689. https://doi.org/10.1371/journal.pmed.1002689
9. He J, Baxter SL, Xu J, Xu J, Zhou X, Zhang K. The practical implementation of artificial intelligence technologies in medicine. Nat Med. 2019;25(1):30–6. https://doi.org/10.1038/s41591-018-0307-0
10. Lamarre E, Singla A, Sukharevsky A, Zemmel R. A generative AI reset: rewiring to turn potential into value in 2024 [Internet]. McKinsey & Company; 2024 [cited 2024 Sep 19]. Available from: https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/a-generative-ai-reset-rewiring-to-turn-potential-into-value-in-2024

Copyright Ownership: This is an open-access article distributed in accordance with the Creative Commons Attribution Non-Commercial (CC BY-NC 4.0) license, which permits others to distribute, adapt, enhance this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0. The authors of this article own the copyright.