Scientific Method in Digital Age
Scientific Method in the Digital Age#
The core principles of the scientific method—asking questions, forming hypotheses, designing experiments, collecting data, and sharing results—remain unchanged. However, in the digital era, how we document, analyze, and communicate our work must evolve to ensure research is transparent, reproducible, and collaborative.
Inspired by The Turing Way, this section explores methodologies and best practices for reproducible research in pharmaceutical science. The Turing Way emphasizes that reproducibility is not just a technical challenge, but a cultural one, requiring open tools, clear documentation, and inclusive collaboration.
Just as you wouldn’t store your money in a bank that makes it hard to withdraw, you shouldn’t use digital tools that lock away your data. Choose platforms and software that support open standards, easy export, and sharing—ensuring your research remains accessible and reusable.
The Problem#
Traditionally, scientists have relied on paper notebooks for discovery and documentation. In the digital age, however, research data and workflows are often managed as “IT problems,” disconnected from the scientific process. This can lead to fragmented records, poor reproducibility, and barriers to collaboration.
Mitigation#
We aim to educate scientists and engineers on digital methods and trends that support the entire research lifecycle:
- Use open-source, community-driven tools for data collection, analysis, and sharing.
- Document every step of your process, from hypothesis to publication, in a way that others can understand and reproduce.
- Adopt version control, automation, and collaborative platforms to streamline teamwork and transparency.
Design Decisions#
To ensure global scalability and long-term accessibility:
- Every recommended tool must meet open-source criteria.
- Workflows should be designed for interoperability, transparency, and ease of use.
- Documentation and code should be shared in public repositories whenever possible.
By aligning with the principles of The Turing Way, we can foster a culture of reproducible, open, and collaborative science—accelerating discovery and ensuring that our work benefits the broader scientific community.
1 - Critical Question
Identifying and addressing the key scientific and operational questions that drive pharmaceutical development, from safety and efficacy to manufacturability and patient delivery.
The Role of Critical Questions in Pharmaceutical Development#
Every successful pharmaceutical product begins with a set of critical questions. These questions guide the scientific method and shape each phase of development, from early discovery through to commercial manufacturing and patient delivery. Identifying and answering these questions is essential to ensure that medicines are not only safe and effective, but also manufacturable at scale and accessible to patients.
Phase 1: Is It Safe and Efficacious?#
The first and most fundamental questions in drug development focus on patient safety and therapeutic benefit:
- Is the drug candidate safe for human use?
- Does it demonstrate the intended efficacy in preclinical and clinical studies?
- What are the potential risks, side effects, and contraindications?
- How do process parameters and material attributes impact the critical quality attributes (CQAs) of the final product?
- How does the scale of my process impact the performance and reproducibility of the product?
- What is the stability of my form and formulations over time and under various conditions?
Phase 2: Can We Manufacture and Deliver It Efficiently?#
Once safety and efficacy are established, the focus shifts to practical and economic considerations:
- Can the product be manufactured consistently and at high quality?
- What are the key cost drivers in the manufacturing process?
- How can we optimize the process to reduce costs without compromising quality or compliance?
- Are there supply chain or distribution challenges that could impact patient access?
- How do we ensure scalability from clinical to commercial production?
- What regulatory requirements must be met for global distribution?
The Ongoing Cycle#
Critical questions do not end once a product reaches the market. Continuous improvement, post-market surveillance, and lifecycle management all depend on asking and answering new questions as knowledge, technology, and patient needs evolve.
Examples of Critical Questions#
- How do process parameters and material attributes impact my final product quality attributes?
- How does the scale of my process impact the performance of the process?
- What is the stability of my form and formulations?
- What are the most significant risks to patient safety and how can they be mitigated?
- How can manufacturing be optimized to ensure both quality and cost-effectiveness?
- What strategies will ensure reliable delivery and patient access worldwide?
By systematically addressing these critical questions at each stage, pharmaceutical teams can develop products that are not only innovative and effective, but also practical to manufacture and deliver—ultimately ensuring that patients receive the therapies they need.
2 - Design
Applying design thinking and contextual knowledge to shape effective, innovative, and value-driven pharmaceutical research and manufacturing projects.
Design Thinking#
A design thinking approach is essential for clearly defining the outcomes you aim to achieve in pharmaceutical research and manufacturing. This phase is about being prescriptive and intentional—identifying the business value, reviewing historical knowledge, and determining what gaps must be filled to move forward.
Scoping projects during the design phase is critical. It allows teams to “dare to dream” and innovate, while ensuring that every experiment or process improvement is grounded in a clear understanding of context. In manufacturing, this context is often captured through material genealogy—the detailed record of how materials and processes are linked throughout production.
Understanding and documenting this genealogy is foundational for reproducibility, traceability, and process optimization. For a deeper dive into the importance of material genealogy in manufacturing, see this background paper: 2012-batch-genealogy-justin-neway.pdf
3 - Data Collection
Effective data collection is the backbone of scientific progress in pharmaceutical research and manufacturing. To answer the critical question, teams must gather both historical and new data, building a robust foundation for process knowledge and understanding (PKU).
Data collection is not just about amassing information—it’s about ensuring that the right data is captured, in the right format, and with the right context. This enables meaningful analysis, supports regulatory compliance, and drives continuous improvement.
Key considerations for data collection include:
- Defining Data Needs: What specific data is required to address your scientific or operational question? This may include process parameters, material attributes, environmental conditions, and outcomes.
- Historical Data Mining: Leveraging existing datasets can reveal trends, variability, and potential risks. However, legacy data may be incomplete or inconsistent, requiring careful curation and validation.
- Designing New Experiments: When gaps exist, targeted experiments should be designed to generate high-quality, reproducible data. This includes clear protocols, standardized measurements, and rigorous documentation.
- Data Integrity and Traceability: Ensuring data is accurate, complete, and attributable is essential for both scientific validity and regulatory compliance (see ALCOA+ principles).
- Integration and Accessibility: Data should be stored in interoperable formats and systems that facilitate sharing, analysis, and long-term access.
For a deeper exploration of these concepts and practical strategies for building process knowledge and understanding, see this paper by Fermier & Higgins (2018), which discusses the challenges and best practices in pharmaceutical data collection and PKU.
4 - Aggregation
Strategies and best practices for aggregating diverse pharmaceutical data sources into unified, analyzable, and reproducible datasets.
Aggregation Strategies Come in Many Shapes and Sizes#
Once data has been collected, it often exists in a variety of formats—PDFs, CSVs, Excel files, databases, and more. The challenge is to bring this disparate information together in a way that enables meaningful analysis, supports decision-making, and maintains data integrity.
Aggregation is not a one-size-fits-all process. The optimal strategy depends on your goals, available resources, and the complexity of your data landscape. The focus should always be on fit-for-purpose approaches that optimize speed, cost, and quality.
Key Considerations for Data Aggregation#
- Conformance to a Common Data Model: Standardizing data into a unified structure is essential for downstream analytics, regulatory submissions, and reproducibility.
- Data Cleaning and Transformation: Raw data often requires cleaning, normalization, and transformation to ensure consistency and usability.
- Automation vs. Manual Integration: Automated pipelines can accelerate aggregation and reduce errors, but manual curation may be necessary for complex or legacy datasets.
- Documentation and Traceability: Every step in the aggregation process should be well-documented to ensure transparency and reproducibility.
Recipe-Based Approach#
Our recommended strategy is to follow a recipe-based approach, which provides a structured, repeatable method for aggregating and transforming data. This approach emphasizes modularity, transparency, and adaptability—key principles for robust data management in pharmaceutical research and manufacturing.
For a detailed discussion and practical examples, see the following paper:
5 - Analysis
Exploring the spectrum of analytical methods in pharmaceutical research and manufacturing, from foundational statistical process control to advanced machine learning and AI.
Analysis is the phase where data becomes actionable insight. In pharmaceutical research and manufacturing, analytical methods span a broad spectrum—from foundational statistical techniques to cutting-edge machine learning and artificial intelligence.
Statistical Process Control (SPC)#
At its core, analysis in pharma often begins with Statistical Process Control (SPC). SPC uses statistical methods to monitor and control processes, ensuring that manufacturing remains consistent and within defined quality limits. Key tools include:
- Control charts for tracking process stability
- Capability analysis to assess whether a process meets specifications
- Trend analysis for early detection of deviations
SPC is essential for maintaining product quality, meeting regulatory requirements, and driving continuous improvement.
Multivariate Analysis and Chemometrics#
As processes and data become more complex, multivariate analysis techniques—such as Principal Component Analysis (PCA) and Partial Least Squares (PLS)—help uncover relationships among multiple variables. These methods are widely used in:
- Process optimization
- Root cause analysis
- Real-time process monitoring (PAT)
Advanced Analytics and Machine Learning#
With the explosion of data in pharma, advanced analytics and machine learning (ML) are transforming how we extract value from information. Applications include:
- Predictive modeling for process outcomes and product quality
- Anomaly detection to identify rare events or out-of-spec results
- Automated image analysis in quality control and pathology
- Natural language processing for mining scientific literature and regulatory documents
Machine learning models can learn from historical and real-time data, enabling proactive decision-making and process automation. However, successful deployment requires careful attention to data quality, model validation, interpretability, and regulatory compliance.
The Future: AI-Driven Pharma#
Artificial intelligence (AI) is poised to further accelerate drug discovery, development, and manufacturing. From generative models for molecule design to digital twins of manufacturing processes, AI is opening new frontiers in pharmaceutical science.
By combining foundational statistical methods with advanced analytics and AI, the pharmaceutical industry can achieve greater efficiency, quality, and innovation—ultimately delivering better outcomes for patients and society.
6 - Reporting
Best practices for transparent, reproducible, and regulatory-compliant reporting in pharmaceutical research and manufacturing.
The Importance of Reporting#
Reporting is the final, critical phase of the scientific and manufacturing process. It transforms data, analysis, and insights into a structured narrative that communicates findings, supports decision-making, and ensures compliance with regulatory requirements.
Key Aspects of Effective Reporting#
- Transparency: Reports should clearly document methods, data sources, analyses, and results, enabling others to understand and reproduce the work.
- Regulatory Compliance: In pharmaceuticals, reporting must meet stringent standards set by regulatory agencies (e.g., FDA, EMA, ICH). This includes traceability, data integrity, and adherence to guidelines such as 21 CFR Part 11.
- Audience Awareness: Reports should be tailored to their audience—scientists, engineers, management, or regulators—highlighting relevant details and actionable insights.
- Data Visualization: Effective use of tables, charts, and graphics can make complex data more accessible and support better decision-making.
- Archiving and Accessibility: Reports must be stored securely and remain accessible for audits, inspections, and future reference.
Digital Reporting and Automation#
Modern reporting leverages digital tools to automate data extraction, analysis, and formatting. Automated reporting systems can:
- Reduce manual errors and save time
- Ensure consistency across reports
- Facilitate real-time dashboards and alerts
Reproducibility and Open Science#
Following principles from The Turing Way, reproducible reporting means sharing not just results, but also the underlying data, code, and workflows. This fosters trust, accelerates innovation, and enables collaboration across the scientific community.
Effective reporting closes the loop in the scientific method—turning research and manufacturing data into knowledge that drives progress, compliance, and better outcomes