Design of Experiments (DOE) Techniques

DoE, QbD and PAT 1. Introduction Evolution of pharmaceutical development: from empirical trial-and-error → risk-based scientific approaches. Regulatory drivers: ICH guidelines (Q8–Q14), FDA PAT initiative (2004). Importance of integrating design, knowledge, and real-time control. Positioning DoE, QbD, and PAT as a “triad” for robust, efficient, compliant development. 2. Historical Context and Regulatory Push Past reliance on end-product testing and its limitations. Shift to lifecycle management approaches. Role of FDA’s Critical Path Initiative. QbD introduced into regulatory lexicon in 2004; PAT guidance published. Global adoption: EMA, MHRA, WHO. 3. Understanding the Three Pillars 3.1 Quality by Design (QbD) – The Framework Definition & Philosophy: Proactive design vs reactive testing. Key Concepts: QTPP – Quality Target Product Profile. CQA – Critical Quality Attributes. CPP – Critical Process Parameters. CMA – Critical Material Attributes. Stages of Application: Early development → Technology transfer → Lifecycle management. Regulatory Basis: ICH Q8(R2), Q9, Q10, Q11, Q12, Q13, Q14. Tools: Risk assessments (FMEA, Ishikawa, Fault Tree Analysis), control strategy design. Case Study Example: QbD applied to controlled-release tablet development. 3.2 Design of Experiments (DoE) – The Optimizer Definition: Statistical framework for systematic factor–response exploration. Role in QbD: Tool to identify design space. Types of DoE: Screening designs (Plackett-Burman, Fractional Factorial). Optimization designs (Central Composite, Box-Behnken). Robustness studies. Benefits: Identifies interactions, reduces experiments, builds knowledge quantitatively. Case Example: Optimizing binder level, granulation time, and impeller speed. 3.3 Process Analytical Technology (PAT) – The Real-Time Guardian Definition: Real-time monitoring and control toolkit. Role: Ensures processes remain within validated design space. Techniques: NIR, Raman, FTIR, Particle size analyzers, Focused Beam Reflectance Measurement (FBRM). Applications: Blend uniformity. Moisture control. Coating thickness. Continuous manufacturing. Regulatory Context: FDA PAT Guidance (2004). Case Example: Inline NIR monitoring for RTRT (Real-Time Release Testing). 4. Interrelationship of the Three Pillars DoE as the engine of knowledge → defines design space. QbD as the overarching framework → integrates knowledge, risks, and control strategy. PAT as the execution safeguard → ensures adherence in manufacturing. Lifecycle integration (development → validation → continuous verification). 5. Benefits of Integrated Use Regulatory alignment & faster approvals. Cost savings through fewer failed batches. Increased robustness and reproducibility. Knowledge management & data-driven decision-making. Example: Continuous manufacturing systems where DoE defines design space, QbD integrates it, and PAT ensures execution.

“Never let one person design and conduct an experiment alone, 𝘱𝘢𝘳𝘵𝘪𝘤𝘶𝘭𝘢𝘳𝘭𝘺 if that person is a subject-matter expert in the field of study.” – Doug Montgomery This is one of my favorite quotes in the #DesignOfExperiments world as it beautifully illustrates the trade-off between making use of experience and suffering from bias. I first encountered it in the excellent DOE FAQ Alert newsletter by Mark Anderson from Stat-Ease. Today I want to show what the quote might refer to, using an example of a situation that can occur in phase three of the DOE workflow: Choosing which factors to vary. We know what we want to achieve (phase 1), we know how to reliably measure quality (phase 2), so now it’s time to gather the team to come up with ideas on what “handles” we should pull in our study. In less than an hour, a small group can easily think of 10-20 factors that could be important for quality (fishbone diagrams facilitate the process well). This list must now be cut down to something manageable that fits the budget and purpose at hand. One way to do this is the following: Have everyone privately rank the potential factors by expected effect size. Then compare across the team and get ready for surprises. Once the dust settles from the inevitable discussion, you should have a decent take on a prioritized list. You can then iteratively explore trade-offs between budget, risk, and final number of factors in phase 4. Now for the challenge that I sometimes encounter before we get this far: When a subject-matter expert with a long history at the company has “tried everything there is to try”. In that case, they may protest that “You can’t improve quality further with any of these factors. We have tried everything on this list many times before”. Since companies that don’t use DOE vastly outnumber those that do, I want to explain how this employee can be both right and wrong at the same time (DOE practitioners have probably guessed where I’m going). It may be that this employee has diligently tested every conceivable factor via one-factor-at-a-time (OFAT) experiments, around the present settings. The result is shown in the figure, where changing each factor *on its own* has little to no effect. This is why the expert has concluded that nothing works (which was correct in their experiments). However, without DOE they typically assume this generalizes to combinations of changes too, which is often wrong! You can replicate this in your kitchen: Imagine your “present setting” for baking a cake is 50 °C for 1 minute. Increasing either factor (temperature or time) a lot won’t make a difference – you must do both in combination to see a large improvement. This is one reason why experience isn't always a blessing and why we should be wary of ignoring factors "we've tried before" with OFAT. We must use other knowledge (of e.g. chemistry/physics) to guide our choice of factors. DOE will then often make us explore areas that give surprising results!

🤯 Unraveling the Design Of Experiments (DoE) mindsets. In the world of DoE, two distinct approaches emerge: Model-Based and Model-Agnostic, each with its unique charm and utility. Let's dive deeper into these fascinating concepts! 🔍 𝐌𝐨𝐝𝐞𝐥-𝐁𝐚𝐬𝐞𝐝 𝐃𝐨𝐄: 𝐄𝐦𝐛𝐫𝐚𝐜𝐞 𝐭𝐡𝐞 𝐩𝐨𝐰𝐞𝐫 𝐨𝐟 𝐩𝐫𝐢𝐨𝐫 𝐤𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 𝐚𝐧𝐝 𝐝𝐨𝐦𝐚𝐢𝐧 𝐞𝐱𝐩𝐞𝐫𝐭𝐢𝐬𝐞 ! In the Model-Based approach, we start by postulating a specific model a-priori (with main effects, interaction terms, and more...) to generate the most efficient and informative data points that optimize the estimation of the model's coefficients. This approach shines in three key scenarios: when the underlying model is well-known, when prior insights can guide the exploration, or when sifting through a multitude of factors to identify the truly significant ones. This type of design may require fewer data points, and also offers easily explainable and interpretable results, perfect for decision-making. 🎯 𝐌𝐨𝐝𝐞𝐥-𝐀𝐠𝐧𝐨𝐬𝐭𝐢𝐜 𝐃𝐨𝐄: 𝐄𝐦𝐛𝐫𝐚𝐜𝐞 𝐭𝐡𝐞 𝐛𝐞𝐚𝐮𝐭𝐲 𝐨𝐟 𝐟𝐥𝐞𝐱𝐢𝐛𝐥𝐞 𝐩𝐫𝐞𝐝𝐢𝐜𝐭𝐢𝐯𝐢𝐭𝐲 ! In the Model-Agnostic approach, we disperse points uniformly and randomly across the experimental space (thanks to Space-Filling designs type) and then fit different models to these points, seeking to obtain the best predictive model. This methodology truly thrives when the underlying model is complex or unknown, as it refrains from making any assumptions about the model form. However, it may demand a higher number of data points, depending on the model complexity (Regression, SVM, Neural Networks, Tree-based models...) and response surface non-linearity. Yet, it proves to be a versatile choice, resilient to model misspecifications. ⚖ 𝐏𝐫𝐨𝐬 𝐚𝐧𝐝 𝐂𝐨𝐧𝐬 Model-Agnostic designs offer flexibility and robustness, successfully exploring and modeling complex and unknown experimental spaces. On the other hand, Model-Based designs are extremely efficient, requiring less data but treading carefully in the face of model misspecification. Choosing or combining these two DoE mindsets depends on the specific research objectives, the amount of domain expertise and knowledge, and the supposed complexity of the response surface. 📊 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐧𝐠 𝐃𝐨𝐄 Model-Based approaches place more emphasis on statistical significance, especially with optimal designs (except for Response Surface Models and Mixture designs). Meanwhile, Model-Agnostic designs prioritize predictive accuracy, embracing metrics like RMSE and other predictive performance indicators. They strive to achieve the best predictions within the experimental space. 💡 Choose or combine the DoE path(s) that best aligns with your research goals. Empower your decisions with data-driven insights ! #DOE #DataScience #DesignOfExperiments

When #xylanases were first introduced to #flour, they were hailed as a miracle #enzyme, revolutionizing bread-making by improving -> dough handling, -> volume, and -> crumb structure. Their impact was so significant that they quickly became a standard ingredient in flour treatment and bread improvers. ▪️ But soon, the industry realized a key challenge -> not all xylanases perform the same way. ✔ A xylanase that works perfectly in one recipe may fail in another. ✔ The same enzyme at the same dosage can produce wildly different results depending on - flour type, - water absorption, - mixing conditions, and - proofing times. ✔ There is no universal xylanase solution—each application requires careful selection and fine-tuning. ▪️ Traditionally, bakers and mills rely on trial and error, but there’s a smarter way: Design of Experiments #DOE. So, how can bakeries and mills move beyond trial and error to optimize xylanase performance? 🔹 DOE: A Smarter Way to Optimize Xylanases Instead of testing one factor at a time—which is slow and overlooks key interactions—Design of Experiments (DOE) allows us to: ✅ Compare multiple xylanases and conditions at once ✅ Identify the best enzyme (or blend) for a specific application ✅ Map how xylanase type, dosage, water absorption, and mixing interact ✅ Reduce costly trial-and-error while achieving consistent results For example, a DOE study could compare: 🔹 Aspergillus vs. Bacillus vs. Trichoderma xylanases 🔹 Different enzyme dosages (e.g., 10, 20, 30 ppm) 🔹 Varying water absorption and mixing times -> By analyzing the data, bakeries can pinpoint the optimal xylanase and process conditions—without endless guesswork. ▪️ At GRAINAR -> We have seen firsthand that DOE is the most effective approach to optimizing xylanase performance. -> Rather than relying on traditional trial and error, we work together with our customers to systematically fine-tune their enzyme strategies—ensuring better dough performance, volume, and consistency every time. ▪️ Xylanases are a huge topic in milling and baking✨, and from time to time, I come back to them because the more we understand them, the better our flour and bakery products become. Stay tuned! 👀🚀 Thanks for reading✨📚 #bioscience #rheology ##breadmaking #Grainar