­­­Quantitative Operating Model Design for Life Sciences Organizations in the Era of AI

In recent years, the Life Sciences industry has witnessed a significant transformation due to the convergence of digital innovation and artificial intelligence (AI). This technological breakthrough has led to the convergence of previously disconnected digital technologies, resulting in an unprecedented acceleration of change across the healthcare ecosystems. With the power to augment and potentially automate the work we do, it is imperative that we think critically about how advances in AI will redefine and reorient relationships spanning the Life Sciences industry.

The Life Science companies that are best at deploying AI and machine learning to expedite drug discovery, personalize medicine, and make patient-centric decisions will come out significantly ahead over those that do not. With these transformative new capabilities, these companies must also evolve their operating models to keep pace. 

A Brief History on Operating Model Evolution in Life Sciences

Historically, operating model design methodology evolved in tandem with the four industrial revolutions. Each era brought significant changes to technology and organizational practices. While many recognize these industrial revolutions in terms of energy sources and modes of production, they also created tectonic shifts in ways of working and organizing human capital, which had a larger impact on the pharmaceutical industry. 

• The first industrial revolution (1760-1840) gave rise to hierarchical structures and centralized decision-making, shaping pharmaceutical companies’ focus on centralized R&D (e.g., the emergence of large, vertically integrated groups with specialized roles, such as chemists, biologists, and pharmacists.
• The second industrial revolution (1870-1940) introduced functional departments and mass production, leading to increased efficiency and product diversification (e.g., Pfizer and Merck develop new classes of drugs, such as antibiotics). 
• The third industrial revolution (1970-2000) emphasized collaboration, innovation, and patient-centricity, prompting partnerships and open innovation platforms (e.g., Novartis and Sanofi’s collaborative platforms).
• The fourth industrial revolution (2000-Present) is currently underway. It is characterized by the integration of digital technologies and the use of data analytics (e.g., pharmaceutical companies a leveraging AI and machine learning to accelerate drug discovery and personalize patient care).

Traditional Outsourced Operating Model Design

Historically, companies have successfully outsourced billions in operational resources from high-cost to lower-cost geographies. In the last decade, pharma companies’ regulatory operations have successfully outsourced repeatable regulatory maintenance activities while maintaining quality standards and reducing costs. This process is function agnostic and has remained largely unchanged:

1. Map the organization’s guiding principles, constraints, and desired outcomes 

2. Conduct a current state assessment 

3. Assess the model and search for cost savings without compromising quality or compliance 

4. Design the future state operating model and identify outsourcing partners 

5. Implement the new model and improve it as needed 

This approach enables companies to optimize their operations and resource allocation, capitalizing on the advantages offered by different regions while achieving organizational goals.

👉 Recommended reading: Driving Diversity, Equity, and Inclusion in Clinical Trials

Disrupting the Traditional Operating Model Design Methodology

Depending on the specificity of the design, traditional operating model work does not always fully take workload “demand management / demand forecasting” into account. Traditional demand forecasting models vary widely based on industry, job type, and workflow (i.e. units of output, human labor hours, Kanban board tickets, FTEs, etc.) but do a poor of accurately managing demand. 

Adding AI to the methodology also adds complexity.

Designing operating models in the AI age requires fostering a culture that embraces technology as a positive transformative force. This entails establishing cross-functional collaboration and interdisciplinary teams that can effectively leverage AI capabilities. Moreover, it requires a redefinition of roles and responsibilities to ensure seamless integration of AI into existing processes. 

Quantitative Operating Model Design

Acquis’ Quantitative Operating Model Design methodology addresses parts of this challenge by focusing demand forecasting and operating model design on “Outcomes”. These outcomes are comprised of “Requirements” that must achieve “Fulfillment” through enhanced human effort (e.g. AI / automation). 

Requirements “R” are defined as one or several sub-outcome types: 

• Repeating Actions (RA) – which have higher potential for automation 
• Non-repeating Actions (NRA) – which have lower potential for automation
• Repeating Questions (RQ) – which have higher potential for AI & ML platforms
• Non-repeating Questions (NRQ) – which have lower potential for AI & ML platforms

Depending on these sub-categories, there is a threshold of automation potential or AI-enhancement potential. The future of human work and the operating models must be a function of the nature of the Requirements “R” multiplied by Fulfillment “F”.

In other words, Fulfillment “F” can be understood as: 

• Fulfillment (F) = (E*Ai + Aut), where 

• E = Enhanced Human Effort 
• Ai = Artificial Intelligence 
• Aut = Automation

In mathematical terms, Acquis designs quantitative AI-enabled operating models with the following equation: Outcomes = Enhanced Human Effort to Fulfill Requirements 

1. Outcomes (O) = Requirements (R) x Fulfillment (F)* 

2. O = (RA+NRA+RQ+NRQ) x (E*Ai + Aut) 

With this framework, one can build operating models to achieve specific outcomes by defining aggregate requirements against human effort (enhanced by AI and automation). These models then iteratively scale the effort distribution between human effort, AI, and automation to “fulfill” the organization’s specific demand for capability.

These effort and requirement tradeoffs will create crucial decisions, as life sciences companies decide how to invest into AI. Impact of these technologies are hard to predict and companies must operate carefully to gain efficiencies without harming the core capabilities of value to the company.

Where Do We Go from Here?

In 2011, a16z famously observed that “software is eating the world”. Today, this trend is not only accelerating, but reaching an inflection point. Companies will soon need to integrate AI and automation capabilities to remain competitive in their operating models. 

However, navigating the rapidly evolving tech landscape is high stakes, and the most successful life science companies have dedicated innovation teams to do due diligence for core focus areas e.g. early research, clin dev, clin ops, and regulatory affairs. But even these companies suffer from limitations of resourcing and internal bias.

Solving these challenges involve bringing together three distinct types of expertise: 

• Knowledgeable strategic advisors who can understand both broad (health) tech landscape and life science company specific requirements 
• Internal / external functional area experts 
• Operating & resourcing model strategists. These partnerships tend to create the most leverage for AI and automation-enabled operating models to maximize value to life science enterprises. 

Conclusion

While there is no magic eight-ball that will accurately predict how AI will disrupt the industry in 5, 10, or 25 years, innovative companies can mitigate the risk of being disrupted by creating mechanisms to actively “listen”, learn, and pilot for potentially disruptive capabilities. The companies that take these types of calculated risks will see significant dividends in the coming decade.