No Human Intelligence: Bio-Hub Resource Allocation Simulator

Inspiration

I was approached at 9 pm at night at Saturday on the 21st of March and was encouraged to join the challenge as they only had one competitor and the judge believed that it would be more fun if more people participated. Given my background in AI, Statistics and Consulting I thought that it would be fun to build something in a couple of hours.

What it does

My platform uses statistically robust modelling to evaluate biotech hubs across need, translational readiness, and sovereign value, then compares different funding strategies under clear assumptions. It helps decision-makers allocate capital more transparently, consistently, and defensibly.

How we built it

I built the platform by integrating heterogeneous biomedical and public sector datasets into a structured feature pipeline, then deriving engineered variables that capture scientific strength, translational readiness, commercial leverage, sovereign value retention, and health system relevance. These features are mapped into milestone specific scores and probabilities, which are then evaluated under alternative allocation regimes using constrained optimisation and risk-adjusted portfolio logic.

To model ecosystem level dependencies, we represent hubs as a graph and apply graph-based machine learning to capture relational structure, cluster effects, and inter-hub similarity that would be missed by hub level scoring alone. This allows the system to generate allocation recommendations that are not only rank based, but also sensitive to network position, structural risk, and portfolio interactions.

Challenges we ran into

The lack of time was a significant constraint due to me joining so late. I was also juggling this with another Team Challenge that I was doing with another team so was doing two challenges at the same time.

Accomplishments that we're proud of

I'm really proud of having put this together in only a couple of hours as I was encouraged to join at 9 pm on Saturday the 21st and the building closed at 11 pm. The next day we had to submit at noon.