Experimental Design Methods

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.

Nearly all economists understand the basic structure of randomized controlled experiments, but John List has a new paper on the benefits of within-subject experimental design. Many RCTs exist leveraging a between-subject (BS) experimental design where treatment effects are measured by comparing two separate groups. However, while BS designs provide clear estimates of average treatment effects (ATE), they are limited in their ability to reveal distributional effects or individual-level variation. That is where the within-subject (WS) experimental design comes into play and John List does a good job explaining. In a WS design, each participant experiences both the treatment and control conditions at different times. This approach increases statistical power as each individual serves as their own control, reducing the variance associated with individual differences. When properly executed, WS designs allow researchers to observe the full joint distribution of treatment effects, uncovering insights beyond marginal averages. However, WS designs introduce three key assumptions that must be met for results to be internally valid: 1️⃣ Balanced Panel – Every participant must remain in the study for the full duration, ensuring complete data across all conditions. 2️⃣ Temporal Stability – The participant’s potential outcome must not be influenced by the passage of time alone. 3️⃣ Causal Transience – The effect of a treatment must not persist into subsequent conditions, meaning that exposure to the first condition does not alter the response in later conditions. When these assumptions hold, WS designs provide superior insights compared to BS designs. However, they can also introduce distortions when assumptions are violated, so it is helpful to think about three categories: 🔴 Overt WS Designs – When causal transience is violated, earlier treatments influence later responses, distorting results. 🟠 Epochal WS Designs – When temporal stability is violated, outcomes shift over time independent of treatment. 🟢 Stealth WS Designs – These designs successfully navigate the assumptions and maintain validity, making them the gold standard for WS experiments. Understanding the strengths and limitations of within and between experimental designs is essential for drawing reliable conclusions. While WS designs can extract richer insights, they demand careful methodological execution to avoid invalid comparisons. #ExperimentalDesign #ResearchMethods #DataScience #Statistics #CausalInference #ScientificMethod #Economics

Design of Experiments (DOE) is deeply entrenched in some R&D labs, and dismissed as overkill in others. A new paper shows you can use it both flexibly and frugally. DOE is widely used in ingredient screening, formulation development, process optimization, and beyond. The toolkit ranges from screening designs that separate active factors from noise, to factorial designs that quantify interactions, to response surface methods that model nonlinear behavior near an optimum. Each flavor makes a mathematically explicit tradeoff between resolution and experimental cost, suited to a different stage of development. In practice, I have seen teams pick a design without matching it to the question: full factorial "just to be safe" when a screening design would suffice. Further, even when the design type is right, it can often be further adjusted based on domain knowledge, for example weighting factors unequally or pooling dimensions known to matter less. The result is wasted effort and sometimes less clarity rather than more. A recent paper captures several practical DOE examples in catalyst screening and cross-coupling optimization that showcase flexible, frugal design shaped by both chemistry and instrumentation constraints. The authors reduced experiments by 75% compared to full factorial and still identified the most promising catalytic systems and conditions. Four lessons reinforced by this work: 🔹Start by ranking your variables: which factors drive outcomes, which interact, and which are secondary. That ranking is a bet. Making it explicit lets you invest experimental budget where it matters most and accept reduced coverage where a directional trend is sufficient. 🔹Match the design to that ranking. Some designs provide uniform coverage across all dimensions, ideal when factors are equally unknown. Others let you cut runs selectively on lower-impact dimensions. The right choice depends on what you must know precisely versus where a general trend is enough. 🔹Think in stages, not one big design. A preliminary screen does not need to find the optimum. It needs to eliminate dead ends and surface promising directions. Save the higher-resolution designs for the follow-up. It is being strategic to match the resolution and objective to each stage. 🔹Look beyond classical DOE when the problem calls for it. Approaches like Bayesian Optimization (BO) operate under different assumptions and yield different information. Understanding when each fits, and when to combine them, can unlock insights that no single method delivers alone. Check out the detailed use cases in the paper (including the integration of DOE and BO for cost-aware discovery), and see how you might adapt them to your own designs. 📄 Frugal Sampling Strategies for Navigating Complex Reaction Spaces, Organic Process Research & Development, April 10, 2026 🔗 https://lnkd.in/eQZjvzvc

As academics, we all want our research to be trusted, reproducible, and strong enough to withstand review. Yet most of the problems we face during publication come from one place: weak statistical foundations and unclear experimental design. This is why I want to give you a quick, practical guide you can use to strengthen any study you are planning or refining. These principles are simple, but they prevent the most common errors I see across manuscripts, reviews, and collaborations. 1. Statistics is not about numbers. It is about reasoning. Each test, each calculation, tells a story about your data and what it truly means. 2. Experimental design begins with purpose. Define your objective clearly before you begin collecting data. The design should flow naturally from the research question. 3. Randomization protects integrity. Assign treatments randomly to eliminate bias and ensure valid comparisons. 4. Replication increases confidence. Repeating experiments strengthens conclusions and helps distinguish real effects from noise. 5. Control groups matter. They provide the baseline that gives your results meaning. Without controls, interpretation becomes speculation. 6. Choose tests based on data, not habit. Understand whether your variables are categorical, continuous, or ordinal. Then select the statistical method that fits the data, not the one that feels familiar. 7. Interpret, do not just report. Numbers are not the end of the story. Explain what they mean, why they matter, and how they support or challenge your hypothesis. 8. Visuals clarify understanding. Use tables and graphs to reveal patterns and relationships, but keep them clean, accurate, and purposeful. 9. Ethical analysis is non-negotiable. Never manipulate data to fit a narrative. Transparency and honesty sustain the credibility of your research. 10. Statistics and design are partners. Good design minimizes errors. Good statistics reveal the truth within them. One without the other cannot stand. These principles are not theoretical. They are the difference between a study that moves quickly through review and a study that struggles with rejection, uncertainty, or inconsistent conclusions. Download the full PDF below. Do you think your current research would benefit from this guide? Reply and tell me. I would love to know. ______________________________ 📌 This is Prof. Samira Hosseini. I’ve helped 12,000+ ambitious academics go from struggling with publishing papers in Q1 journals, limited visibility, and poor citation records to building a solid research trajectory and high 𝘩-index. Book a free Strategy Call, and we can dive into your challenges in top-tier journal publication and citation and see how I can best assist you: https://lnkd.in/ezqV64dX

How to design PhD methodology that survives your viva, when your examiner asks "Why these methods?" Most PhD students design methodology backwards. They pick methods they're comfortable with. Then retrofit their research question to fit. When examiners ask "Why this approach?" they can't answer. Here's the 10-step framework for bulletproof methodology: Steps 1-3: Build Your Foundation Step 1: Define your research problem What gap are you addressing? What are your research questions? Step 2: Conduct your literature review What methods have others used? What worked? What didn't? Step 3: Establish your philosophical position Positivist? Interpretivist? Critical Realist? This decision guides everything that follows. Example: "How do knowledge workers make sense of mental health challenges in remote work?" → Interpretivist position (understanding subjective meanings) Steps 4-5: Make Design Decisions Step 4: Choose your research design Quantitative? Qualitative? Mixed methods? Must align with your research questions and philosophy. Step 5: Select specific methods Interviews? Surveys? Experiments? Consider triangulation for robustness. Example: Sequential mixed methods → Phase 1: Phenomenological interviews → Phase 2: Experience sampling via mobile app Steps 6-7: Plan Execution Step 6: Design sampling strategy Who will you study? How many? Why? Step 7: Plan data analysis Statistical tests? Thematic analysis? Choose software and frameworks. Steps 8-9: Ensure Rigour Step 8: Establish quality criteria Quantitative: Validity and reliability Qualitative: Trustworthiness Step 9: Address ethical considerations Informed consent, confidentiality, data protection. Get ethics approval BEFORE data collection. Step 10: Pilot and Refine Test with 3-5 participants. Iterate based on feedback. The non-negotiable principle: Every decision must trace back to: - Your research questions - Your philosophical position - Your practical constraints Common mistakes that fail vivas: Choosing methods before establishing philosophy. Weak justification for design decisions. No pilot testing. Misalignment between questions and methods. Keep a methods journal documenting every decision and rationale. Methodology isn't about picking the "right" method. It's about creating perfect alignment from problem → philosophy → design → methods → analysis → quality. When your examiner asks "Why this approach?" you'll have a clear answer at every step. What's your biggest methodology challenge?

📈 Econometric Corner: #18 📊 A/B tests have proven to be highly valuable for firms. However, the limitations of these simple A/B tests have been increasingly recognized, of which the two most prominent being addressing interference and estimating heterogeneous effects. In this citied paper, the authors address these two challenges by developing a theoretical framework for the optimal design and analysis of switchback experiments with minimal assumptions. In these experiments, a unit is sequentially exposed to random treatments, and its response is measured over a fixed period. Administering alternate treatments to the same unit enables direct estimation of individual-level causal effects and mitigates interference challenges. 𝗦𝗲𝘁𝘂𝗽: The authors use the variation from the random assignment path for inference (i.e., design-based inference). They focus on regular switchback experiments and exclude adaptive treatment assignments. 𝗧𝘄𝗼 𝗮𝘀𝘀𝘂𝗺𝗽𝘁𝗶𝗼𝗻𝘀 that limit the dependence of the potential outcomes on assignment paths: 𝗔𝘀𝘀𝘂𝗺𝗽𝘁𝗶𝗼𝗻 𝟭 (non-anticipating potential outcomes) states that the potential outcomes at time, 𝘵, do not depend on future treatments. 𝗔𝘀𝘀𝘂𝗺𝗽𝘁𝗶𝗼𝗻 𝟮 (m-carryover effects) restricts the order of the carryover effect. 𝗘𝘀𝘁𝗶𝗺𝗮𝗻𝗱: the average lag-𝘱 causal effect of consecutive treatments on the outcome, where 𝘱 reflects the experimental designer's knowledge of the order of the carryover effect. 𝗗𝗲𝘀𝗶𝗴𝗻 𝗼𝗳 𝗥𝗲𝗴𝘂𝗹𝗮𝗿 𝗦𝘄𝗶𝘁𝗰𝗵𝗯𝗮𝗰𝗸 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝘀:    • To find the optimal design of the regular switchback experiment, the authors use a minimax framework to derive the best possible design for the worst-case set of potential outcomes. 𝗞𝗲𝘆 𝗥𝗲𝘀𝘂𝗹𝘁𝘀: 1️⃣ 𝗢𝗽𝘁𝗶𝗺𝗮𝗹𝗶𝘁𝘆 𝗼𝗳 𝗙𝗮𝗶𝗿 𝗖𝗼𝗶𝗻 𝗙𝗹𝗶𝗽𝗽𝗶𝗻𝗴. Under p=m (i.e., perfect knowledge of carryover effects), the optimal randomization probabilities should be 1/2. 2️⃣ 𝗢𝗽𝘁𝗶𝗺𝗮𝗹 𝗗𝗲𝘀𝗶𝗴𝗻. Under p=m, when m=0, the optimal randomization frequency is {1, 2, 3,…,T}. When m>0, and if there exists , s.t. T=nm, then the optimal randomization frequency is {1, 2m+1, 3m+1, …, (n-2)m+1}. 3️⃣ 𝗧𝘄𝗼 𝗺𝗲𝘁𝗵𝗼𝗱𝘀 𝗳𝗼𝗿 𝗶𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲. Under p=m, Exact Inference (randomization-based test) and Asymptotic Inference (a finite population conservative test). The inference is still valid when p m, though. 4️⃣ 𝗧𝗵𝗲 𝗽𝗿𝗼𝗰𝗲𝗱𝘂𝗿𝗲 𝘁𝗼 𝗲𝘀𝘁𝗶𝗺𝗮𝘁𝗲 𝘁𝗵𝗲 𝗼𝗿𝗱𝗲𝗿 𝗼𝗳 𝘁𝗵𝗲 𝗰𝗮𝗿𝗿𝘆𝗼𝘃𝗲𝗿 𝗲𝗳𝗳𝗲𝗰𝘁. Define a hypothesis testing procedure that, when combined with a search method, provides an estimate of the magnitude of the carryover effect. Finally, the authors demonstrate their approach using a simulated study and conclude by discussing its practical implications and limitations. Check it out! 𝗥𝗲𝗳𝗲𝗿𝗲𝗻𝗰𝗲: Bojinov, I., Simchi-Levi, D., & Zhao, J. (2023). Design and analysis of switchback experiments. 𝘔𝘢𝘯𝘢𝘨𝘦𝘮𝘦𝘯𝘵 𝘚𝘤𝘪𝘦𝘯𝘤𝘦, 69(7), 3759-3777.

🚀 Why have I started Gateway.AI? Over the last few years I’ve been in a lot of rooms where teams try to adopt ML/automation in the real world - labs, factories, data-heavy groups. The biggest shift is to start with the workflow, not the model. Day-one questions I ask now: - Map the workflow: What are the true bottlenecks to throughput, cost, or quality? - Fit for AI/automation: Which bottlenecks can tech actually relieve—and which might it worsen? - Watch for negative ROI: Could AI create more dashboards/paperwork without new value? - For experimentalists: If today’s best theory/simulation were free/instant, how would you change experiments on the scale of seconds → weeks? - Benchmarks that matter: How will you measure productivity gains from AI internally? - Downstream value: Who benefits next - can we define benchmarks for downstream impact? - Rewards & objectives: What’s the objective function of the experiment? - For theory/ML folks: What experimental footprint (time/samples/$) is required to falsify the hypothesis? 🔧 Which AI/optimization method should you use? Pick methods by the shape of your problem, not by hype. A quick picker: - Small search, fast feedback, clear objective → start simple: design of experiments (DoE), gradient/coordinate search, rules. - Low–mid dimensional, moderate cost, noisy objective → Bayesian optimization (single/multi-objective; add constraints if needed). - Structured proxies available (cheap early readouts) → multi-fidelity BO or active learning with surrogate models (Gaussian Process, deep kernel learning). - Huge or discrete spaces, many viable recipes, rich constraints → Genetic algorithms / evolutionary strategies (keep operators “manufacturable”). - High-frequency control with a plant model → model-predictive control (MPC). - Sequential decisions under uncertainty, sparse rewards → contextual bandits (short horizon) → RL (only if you truly need it). - Hard planning with known costs/heuristics → tree search (A*, MCTS) beats RL in many cases. Choose with four dials in mind: parameter-space complexity, data dimensionality, proxy availability, and feedback latency (seconds vs hours vs weeks). Your algorithm should match your budget (samples/time), respect constraints, and exploit any physics priors you have. These questions and choices keep projects anchored to outcomes, not demos. It’s why I started Gateway.AI: to translate ML/AE enthusiasm into measurable productivity and downstream value for materials science! If you’re deciding where to start - or whether to- let’s talk! https://lnkd.in/eNeUiADP #AI #Automation #Optimization #ActiveLearning #BayesianOptimization #GeneticAlgorithms #MPC #RL #Bandits #RDM #LabAutomation #MLOps #ExperimentalDesign #GatewayAI

🎉 Continuing the 2025 series on the foundations of Design of Experiments (#DoE) and modern experimentation approaches, here’s Part 3: Optimization Methods in Experimentation (a more complete version will be published on Medium soon). 🔎 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧 lies at the heart of experimental design, helping researchers and practitioners refine processes, improve performance, and uncover the best experimental conditions while minimizing resources. The evolution of optimization methods reflects a balance between leveraging models to guide experimentation and exploring unknown spaces without assumptions. 🏛️ 𝐌𝐨𝐝𝐞𝐥-𝐁𝐚𝐬𝐞𝐝 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧 Model-based approaches rely on predefined mathematical models to guide experiments. These include classical designs like Central Composite Designs (CCD) and Box-Behnken Designs, which assume polynomial models for response surfaces, as well as Bayesian Optimization, which combines surrogate models and acquisition functions to propose new experiments iteratively. These methods excel when a prior understanding of the system exists or when computational efficiency is key. 🌌 𝐌𝐨𝐝𝐞𝐥-𝐀𝐠𝐧𝐨𝐬𝐭𝐢𝐜 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧 In contrast, model-agnostic methods avoid assumptions about the underlying system, focusing instead on geometric or distance-based considerations. Space-filling designs, such as Latin Hypercube or Maximin designs, ensure even exploration of the experimental space, making them ideal for nonlinear responses. Simplex optimization methods, on the other hand, employ geometric steps to converge on optimal conditions, relying purely on iterative distance-based logic and results ranking. ⏳ 𝐒𝐞𝐪𝐮𝐞𝐧𝐭𝐢𝐚𝐥 𝐯𝐬. 𝐏𝐚𝐫𝐚𝐥𝐥𝐞𝐥 𝐄𝐱𝐩𝐞𝐫𝐢𝐦𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 Sequential methods, such as Bayesian Optimization or Simplex, prioritize small batches of experiments. These allow iterative learning and adaptation, particularly useful when resources are limited or when experiments are costly. Parallel approaches, favored in space-filling designs and model-based optimization like DoE, enable larger experiment batches to be conducted simultaneously, providing a more comprehensive understanding of the experimental space at the expense of iterative refinement. 🎯 𝐓𝐚𝐤𝐞𝐚𝐰𝐚𝐲 The choice of optimization method hinges on the balance between prior knowledge, resource availability, and the need for exploration versus exploitation. Sequential methods align with adaptive learning, while parallel methods accelerate discovery in larger spaces. Similarly, the decision between model-based and model-agnostic approaches depends on the complexity of the system and the availability of prior information. 📢 What optimization approaches have you found most effective in your experimental work ? #Optimization #ExperimentalDesign #DataScience #Innovation

Here are tips to understand the nuances of cohort studies, randomized trials, and quasi-experiments. 1️⃣ Cohort Studies Observational designs where researchers track groups of individuals over time to observe outcomes based on natural exposures. Key Features: ✅ No Intervention: Researchers don’t assign people to interventions; they observe what naturally occurs. ✅ Grouping by Exposure: Participants are grouped based on characteristics (e.g., smokers vs. non-smokers). ✅ Time Frame: Can be prospective (forward in time) or retrospective (analyzing past data). 🎯 Example: Imagine tracking college students with a ChatGPT subscription at enrollment vs those without a subscription. By following these groups over time, we can assess academic performance (e.g., test scores). 💪 Strengths: ✔ Captures real-world conditions. ✔ Effective for studying rare exposures or long-term outcomes. 👎 Limitations: ✘ Prone to confounding (other factors may influence observed relationships). ✘ Cannot establish causation. 2️⃣ Randomized Trials RCTs are experimental designs where participants are randomly assigned to intervention or control groups. Key Features: ✅ Random Assignment: Reduces bias by ensuring comparable groups. ✅ Controlled Conditions: The researcher controls the intervention. 🎯 Example: To evaluate ChatGPT's impact on academic performance, half the students in a school could be randomized to use ChatGPT for homework, while the other half uses traditional methods (e.g., textbooks). Outcomes like scores can then be compared. 💪 Strengths: ✔ Establishes causality. ✔ Minimizes bias through randomization. 👎 Limitations: ✘ Costly and time-intensive. ✘ Ethical concerns when withholding interventions from certain groups. 3️⃣ Quasi-Experiments: Bridging the Gap These designs involve interventions but lack randomization, making them practical for real-world evaluations. Key Features: ✅ Non-Randomized Assignment: Groups are assigned based on existing conditions or convenience. ✅ Intervention: Researchers introduce an intervention to evaluate its effects. ✅ Comparison Groups: Pre-existing groups or natural events are often used. 🎯 Example: If one state in the country adopts ChatGPT in classrooms while a neighboring state does not, researchers can compare outcomes between the two to evaluate the policy's impact. 💪 Strengths: ✔ Practical for evaluating large-scale policies or programs. ✔ Reflects real-world settings. 👎 Limitations: ✘ Greater risk of bias and confounding. ✘ Weaker causal inferences compared to RCTs. 🛠️ When to Use Each Design Cohort Studies: Ideal for understanding associations, especially when interventions are impractical or unethical. Randomized Trials: Best for testing interventions when randomization is possible and ethical. Quasi-Experiments: Useful for real-world evaluations when randomization isn’t feasible, but causal insights are still needed. Please, reshare ♻️ #Chisquares #VillageSchool #StudyDesign

Experimentation lies at the core of effective digital product strategies. Vanguard’s recent tech blog explores how A/B testing and multi-armed bandit (MAB) algorithms each bring value to web optimization—and why choosing the right method matters for delivering fast, impactful results. The article presents a simulation study comparing three approaches: traditional A/B testing, Adaptive Allocation MAB, and Thompson Sampling MAB. For three or fewer variations, a properly powered A/B test often identifies the winner more quickly and is easier to implement and interpret. But once you move beyond four variations, bandit strategies like Thompson Sampling begin to outperform A/B testing—both in terms of speed and in minimizing “regret,” or lost opportunity cost. Thompson Sampling also tended to edge out Adaptive Allocation across most simulated scenarios, though the gap narrows when there’s significant performance uplift or a large number of variants. In short: use A/B when you want clarity and simplicity with a small set of variants; turn to MAB when you need efficiency at scale or rapid optimization. Of course, as this was a simulation-based study, some nuances and real-world dynamics may not be fully captured. Still, this analysis offers a practical rule of thumb for experimentation design—especially for teams looking to improve the efficiency and impact of their testing strategies. #DataScience #MachineLearning #Analytics #Experimentation #ABTest #MultiArmBandit #Measurement #SnacksWeeklyonDataScience  – – –  Check out the "Snacks Weekly on Data Science" podcast and subscribe, where I explain in more detail the concepts discussed in this and future posts:    -- Spotify: https://lnkd.in/gKgaMvbh   -- Apple Podcast: https://lnkd.in/gFYvfB8V    -- Youtube: https://lnkd.in/gcwPeBmR https://lnkd.in/gnfN4bGa