[Project] Clinical Trial Analysis: Learning to Balance Statistics and Real-World Decisions

Why I worked on this project

During my data analysis training, I worked on a simulated clinical trial project focused on evaluating a new diabetes medication.

At first, it felt very different from the business-focused projects I usually enjoy.

It involved statistical testing, medical metrics, and long-term patient data.

But it also raised an important question for me:

How do you turn complex statistical results into something decision-makers can actually use?

The problem I wanted to understand

The dataset represented a Phase III clinical trial comparing a new medication with a placebo.

From a business and regulatory perspective, the core questions were clear:

• Does the medication work better than placebo?

• Is the effect consistent over time?

• Does it work across different patient groups?

• Are there any safety concerns?

  
  

Behind those questions sat large amounts of data and complex statistical methods.

My goal was not just to run tests, but to understand what the results actually meant in practice.

How I approached the analysis

I started by cleaning and structuring the data to ensure consistency and reliability.

Using Python, I calculated changes in HbA1c levels, prepared demographic groupings, and validated key metrics.

I then applied statistical methods to compare treatment and placebo groups over time and across subgroups.

Throughout the process, I focused on two things:

• making sure the results were statistically sound

• making sure they were interpretable outside a technical context

  
  

The second part turned out to be just as important as the first.

What stood out during the analysis

Several patterns became clear.

The treatment group showed a significantly stronger reduction in HbA1c compared to placebo.

The effect was stable over time and did not vary meaningfully by age, gender, or disease duration.

At the same time, adverse event rates were similar in both groups, suggesting a balanced safety profile.

What interested me most was how easily these results could be misunderstood without proper context.

Small differences looked dramatic in isolation.

Some statistically significant results were less meaningful in practical terms.

This reinforced how important interpretation is in analytical work.

Why this matters from a business point of view

In regulated industries such as healthcare, decisions are rarely based on one metric.

They depend on:

• statistical evidence

• risk assessment

• regulatory expectations

• long-term impact

  
  

This project showed me how analysts support those decisions by translating complex analysis into clear, defensible conclusions.

That mindset is something I now carry into my business and reporting projects.

Tools used (kept intentionally focused)

• Python for data processing, statistical testing, and visualisation

• Jupyter Notebook for exploratory analysis and documentation

The emphasis was on analytical rigour and transparent interpretation.

Want to see the technical details?

The full notebook and dataset are available on GitHub.

They include the full analysis, visualisations, and statistical methods used in this project.

Final thought

This project helped me understand the difference between running analyses and supporting decisions.

It strengthened my analytical foundations and taught me how important interpretation is - especially when results influence high-stakes outcomes.

Today, I apply the same discipline in more business-focused projects using Excel and Power BI.

Different tools, same principle: clarity over complexity.

While this project was rooted in statistical analysis, it strengthened the analytical discipline I now apply in operational reporting and business-focused projects.