Case Study: 30% Efficiency Gain Through AI Process-based Improvement
As an experienced business leader who’s led numerous strategic initiatives, I’ve seen firsthand how artificial intelligence (AI) can improve operations. This case study demonstrates how Novartis strategically implemented AI solutions to overcome industry-specific challenges, resulting in significant efficiency gains and providing valuable lessons for other businesses embarking on AI process-based improvement journey.
TL;DR: Key Results
Novartis’s strategic AI implementation yielded significant improvements across multiple areas:
Process Efficiency: 30% reduction in process times through AI-driven workflow automation and predictive analytics.
Overall Efficiency: 25% increase in overall efficiency due to AI integration in data management, predictive analytics, and workflow automation.
Data Processing: 600% reduction in data ingestion time for commercial analytics.
Cost Optimization: Substantial cost savings achieved by automating document-heavy processes like clinical trial report writing.
Regulatory Compliance: Improved compliance with key regulations (GDPR, HIPAA) through AI governance framework and data anonymization.
Talent Development: Successfully built a strong AI and data science team through strategic partnerships and reskilling programs.
Employee and Customer Satisfaction: Enhanced satisfaction levels by reducing manual tasks and enabling faster, data-driven decision making.
These results demonstrate the transformative potential of AI in addressing key challenges in the pharmaceutical industry, from drug discovery to operational efficiency.
The Challenge: Overcoming Inefficiencies
Novartis, like many companies in the pharmaceutical industry, faced significant challenges in the drug discovery and development process. These challenges are both complex and multi-layered, contributing to extended timeframes and high costs associated with bringing new drugs to market. Novartis realized that tackling these efficiency issues demanded a multifaceted strategy that included better strategic planning, advanced technological solutions, and effective data management practices.
1. Operational Inefficiencies
High Failure Rates: Approximately 90% of drug candidates fail during clinical trials, primarily due to lack of clinical efficacy, unmanageable toxicity, poor drug-like properties, and poor strategic planning[1]. Errors in early stages often lead to failures in later stages, with preclinical models frequently failing to accurately represent human disease[2].
Strategic and Process Inefficiencies: Many processes were labor-intensive, reducing overall efficiency. The complexity of data and processes led to delays in critical decisions[13]. Disorganized and siloed data presented significant challenges in utilizing clinical and medical imaging data effectively[3].
High Costs and Long Timelines: The average cost to develop a new drug exceeds $1-2 billion, with the process taking over 10-15 years[1]. Only 4% of drug development programs result in licensed drugs, further intensifying the financial pressure on pharmaceutical companies[2]. Handling large datasets requires significant computing power, which can be costly and time-consuming to manage[3].
2. Technological and Regulatory Challenges
Technological Limitations: The gap between preclinical and clinical success rates was partly due to the limitations of preclinical models, which often failed to replicate human disease accurately[2]. Effective management of attrition across different phases of R&D was crucial, as poor decisions in early R&D could lead to delayed failures in later stages[4].
Regulatory Hurdles: Stringent regulatory requirements complicated and prolonged the drug approval process[1]. The rapidly evolving nature of AI technology often outpaces regulatory frameworks, creating uncertainty in compliance requirements[7]. Stringent regulations like GDPR and HIPAA require careful handling of patient data used in AI models[8]. Regulatory bodies increasingly demand explainable AI models, especially for critical applications in drug discovery and clinical decision support[9].
3. Market and Talent Pressures
Market Dynamics: Lack of commercial need or poor market planning could lead to the discontinuation of drug development programs, even if the drug showed potential efficacy[1].
Talent Acquisition and Retention: A significant gap exists between the demand for AI specialists and the available talent pool, leading to intense competition[11]. Pharmaceuticals compete with tech, finance, and automotive sectors for AI talent[12]. AI roles in pharma require a unique blend of machine learning expertise and understanding of pharmaceutical processes, making suitable candidates rare[11]. AI job postings are often concentrated in specific regions, limiting the talent pool for companies in other areas[12].
These challenges significantly hampered Novartis’s operational efficiency, putting them at a competitive disadvantage and prompting the company to seek innovative solutions to improve their processes. In response, Novartis turned to AI and machine learning technologies, which have shown promise in accelerating drug discovery by identifying novel compounds and targets more efficiently. Additionally, Novartis identified the need for improved data management, recognizing that investing in sophisticated data management infrastructure could help organize and leverage vast amounts of clinical and medical imaging data, facilitating more efficient research and development processes[4].
Faced with these challenges, Novartis recognized the need for a comprehensive, innovative approach. They turned to artificial intelligence as a potential solution, developing a strategic plan to address each area of concern.
The AI Process-based Improvement: A Strategic Approach
To address these challenges, Novartis developed a comprehensive AI strategy focusing on key areas for integration and strategic partnerships.
Key Areas for AI Integration
Novartis took a targeted approach, focusing on:
Workflow automation
Predictive analytics
Data integration
Regulatory compliance and risk management
Talent acquisition and development
Lesson Learned: Stakeholder buy-in is crucial for successful AI implementation. Novartis ensured early engagement with key stakeholders across departments to align on priorities and expected outcomes.
Strategic Partnerships
Novartis formed partnerships to accelerate its AI implementation. These partnerships were formed to accelerate drug discovery, enhance the company’s ability to identify new drug targets, and leverage collective data for more efficient pharmaceutical research and development. Three in particular are with:
Microsoft: Developed “generative chemistry” to speed up bringing new medicines into the clinic[5]. “We signed a collaboration with Microsoft Research Labs three to four years ago where we’re we developed an approach called generative chemistry to try to speed up how fast we can bring new medicines into the clinic” noted Novartis’ CEO Vasant Narasimhan.
Isomorphic Labs: Collaborated with Isomorphic Labs, a spin-off from Google’s DeepMind, to accelerate the identification of new drug targets[5].
Machine Learning Ledger Orchestration for Drug Discovery (MELLODDY) Consortium: This partnership developed a blockchain-based AI platform capable of learning from proprietary compound assay data while preserving confidentiality[6].
These strategic partnerships empowered Novartis to launch AI-driven solutions across various aspects of their operations. Let’s examine how these solutions were put into action and the results they achieved.
AI-Driven Solutions in Action
Impacting the next five years is how AI is gonna impact many of our productivity efforts in drug development, how fast can we generate new trial protocols? How fast can we work with regulators? How fast can we look at patient safety, look at large patient data sets?
Vasant Narasimhan, CEO Novartis
1. Workflow Automation
Goal: Reduce manual tasks and streamline processes
Implementation:
Developed “AE Brain” system for adverse event identification
Implemented Natural Language Processing for document analysis
Integrated AI into human expert workflow as a decision support system[6]
Result:
Improved safety information quality
Reduced manual work burden
Freed employees for higher-value tasks
Lesson Learned: Continuous monitoring and iterative improvements drive success. Novartis implemented a feedback loop to continuously refine the “AE Brain” system based on user input and changing regulatory requirements.
2. Predictive Analytics
Goal: Enhance decision-making with real-time insights
Implementation:
Developed “Nerve Live” platform for operational data analysis
Applied machine learning for multi-functional insights
Created end-user applications for domain experts[6]
Result:
Optimized costs and quality
Enabled data-driven drug development decisions
Accelerated decision-making processes
Lesson Learned: Think beyond automation. Novartis realized that the true power of AI lies in enhancing decision-making processes, not just automating existing ones.
3. Data Integration
Goal: Break down silos and improve data accessibility
Implementation:
Participated in MELLODDY project for secure data sharing
Implemented “transfer learning” techniques[6]
Result:
Enabled faster, more efficient drug discovery
Improved structure-activity information for drug discovery pipeline
Created a comprehensive overview of operations
Lesson Learned: Addressing data quality issues early is crucial. Novartis invested in data cleaning and standardization initiatives before implementing AI solutions, ensuring more accurate and reliable outcomes.
4. Commercial Analytics
Goal: Improve brand uptake analysis for new products
Implementation:
Automated large data volume processing (7 billion records)
Created comprehensive visualizations
Integrated open-source packages like Python and Bokeh[14]
Result:
600% reduction in data ingestion time
Streamlined visualization maintenance
Enhanced data exploration capabilities
5. Regulatory Compliance and Risk Management
Goal: Ensure AI compliance with regulatory requirements
Implementation:
Developed AI governance framework
Implemented data anonymization techniques
Created an AI ethics committee
Established AI model performance monitoring
Result:
Achieved compliance with key regulations (GDPR, HIPAA)
Enhanced AI model transparency
Reduced regulatory risks in AI implementation
Improved stakeholder trust
Lesson Learned: Regulatory compliance must be integrated from the start. Novartis’s proactive approach to AI governance helped avoid potential regulatory pitfalls and built trust with stakeholders.
6. Talent Acquisition and Development
Goal:Overcome the AI talent shortage and develop a highly skilled AI workforce
Implementation:
Partnered with universities for talent pipeline
Implemented reskilling programs
Created an attractive work environment for AI professionals
Established a global talent acquisition strategy
Result:
Built a strong AI and data science team
Reskilled existing workforce
Reduced external hiring dependency
Improved retention of AI professionals
Lesson Learned: Talent development is key to sustained AI success. Novartis recognized that building internal capabilities was as important as attracting external talent.
Despite these challenges, Novartis’s perseverance and strategic approach paid off. The company saw significant improvements across multiple areas of their operations.
The Results: Efficiency Boost and Beyond
Novartis measured the impact of their AI implementation through several operating indicators:
1. Process Times Reduction
30% reduction in process times through AI-driven workflow automation and predictive analytics[15].
2. Overall Efficiency Increase
25% increase in overall efficiency, attributed to AI integration in data management, predictive analytics, and workflow automation[15].
3. Cost Optimization
Significant cost savings achieved by automating document-heavy processes like clinical trial report writing[17].
4. Data Ingestion and Processing
600% reduction in data ingestion time for commercial analytics[15].
5. Employee and Customer Satisfaction
Improved employee satisfaction by reducing manual tasks.
Enhanced customer satisfaction through faster, data-driven decision-making[15].
6. Regulatory Compliance and Risk Management
Ensured compliance with key regulations (GDPR, HIPAA) through AI governance framework and data anonymization[16].
7. Talent Development
Built a strong AI and data science team through strategic partnerships and reskilling programs[15].
These impressive results not only validated Novartis’s AI strategy but also generated valuable insights for future initiatives. Let’s explore the key takeaways and how the company plans to build on this success.
Lessons Learned and Future Plans
Key Takeaways
Stakeholder buy-in is crucial for successful AI implementation.
Continuous monitoring and iterative improvements are essential.
Regulatory compliance must be integrated from the start.
Talent development is key to sustained AI success.
Future AI Roadmap
Scaling AI solutions across more departments
Exploring new AI applications in drug discovery and personalized medicine
Ongoing collaboration with regulatory bodies
Continued investment in AI talent development
While Novartis’s journey offers valuable insights, you might be wondering how to apply these lessons to your own organization. Let’s explore some practical steps you can take to leverage AI in your business transformation.
Applying These Lessons to Your Business
Experienced business leaders recognize the value of strategic transformation. Here’s how you can apply Novartis’s successful AI strategies to your own initiatives:
Start with a clear problem statement: What specific inefficiencies are holding your organization back?
Conduct a company-wide survey to identify problem areas across departments
Analyze key performance indicators (KPIs) to pinpoint areas of underperformance
Hold focused workshops with department heads to articulate specific challenges that AI could address
Assess your data readiness: Do you have the necessary data infrastructure to support AI implementation?
Perform a comprehensive data audit to evaluate the quality, quantity, and accessibility of your existing data
Identify and address data silos within your organization
Implement data governance policies to ensure data integrity and compliance with regulations
Prioritize change management: How will you bring your teams along on this journey?
Develop a clear communication strategy to explain the benefits of AI adoption to all key participants
Create an AI steering committee with representatives from various departments
Implement a training program to upskill employees and address fears about job displacement
Think beyond automation: How can AI enhance decision making and strategic planning in your organization?
Identify key decision points in your business processes where AI insights could be valuable
Explore predictive analytics applications for forecasting market trends and customer behavior
Investigate AI-powered scenario planning tools to enhance strategic decision-making
Partner strategically: Consider collaborations that can accelerate your AI adoption
Conduct a thorough assessment of potential AI vendors and their specific expertise
Explore partnerships with academic institutions for cutting-edge research and talent pipeline
Consider joining industry consortiums or alliances focused on AI in your sector
The Future of AI in Business Transformation
AI goes beyond incremental improvements; it’s a catalyst for completely reimagining business models. As we’ve seen with Novartis, the potential for efficiency gains is substantial.
The real potential of AI, however, lies in its ability to unlock new possibilities. Consider these scenarios:
Predictive maintenance saving millions in downtime