We’ve all seen it, AI is making its way into health economics, whether we’re ready for it or not. Many of us are already exploring AI tools to make our work more efficient, while others are wondering what this means for the models we’ve relied on for years. Let’s break down what’s happening and talk about where we go from here.
What Can AI Do for Us?
AI is already changing the way we approach our work. Here are a few ways it’s proving useful:
- Automating Routine Tasks: How many hours have we spent on repetitive tasks, like sorting and analysing input data, when building models? AI tools are starting to take over that tedious work, speeding up timelines and cutting costs.
- Helping with Prototyping: Tools like ChatGPT can quickly draft code or generate initial model structures. Sure, it’s not perfect yet, but it gives us a head start so we can focus on refining the details.
- Minimising Errors: Let’s be honest—complex models can be error-prone. Machine learning algorithms can help catch mistakes or suggest improvements, which is a win for all of us.
- Wider Application: AI isn’t just for pharmacoeconomics anymore. Public health financing and healthcare policy-making are starting to use predictive models for budget allocation and simulations to evaluate the cost-effectiveness of prevention programmes.
What Do We Need to Watch Out For?
Of course, it’s not all smooth sailing. There are some real concerns we need to tackle:
- Privacy and Data Security: We’re dealing with sensitive data—patient outcomes, trial results—the kind of information that can’t fall into the wrong hands. Pharma companies are understandably cautious about sharing their data with AI tools they don’t control. We’d feel the same.
- Transparency: When it comes to HTA submissions, transparency is non-negotiable. But how do we explain a model built by AI or recommendations based on it? It’s something we’ll need to figure out together.
- Regulatory Lag: We all know how fast the industry moves compared to HTA agencies. It’s going to take time for them to build the capacity to evaluate AI-driven models. Until then, we might need to translate AI models back into Excel VBA while taking steps to make AI models more acceptable to HTA agencies.
What’s Next?
Generative AI—tools that create text, models, or predictions—is where things get really interesting. Imagine automating an entire cost-effectiveness model or running policy impact simulations in real time. It’s not so far-fetched anymore. But, like with any modelling, these tools are only as good as the data we feed them and the safeguards we put in place.
So, what can we, as health economists, do to keep up with AI development? Here are a few ideas:
- Learn the Tools: If you haven’t already, work with tools like GPTs or GitHub Copilot. They’re not perfect, but they’re great for handling repetitive tasks or giving you a starting point. Just don’t throw confidential data at them.
- Brush Up on Machine Learning: Techniques like feature selection or dimensionality reduction can make your models more robust and less complicated. It’s worth exploring.
- Learn Generative AI: Tools like ChatGPT and other generative AI platforms can create new content, such as text, predictions, or models. These tools are particularly helpful for automating repetitive tasks, enhancing policy simulations, and streamlining model creation. To get started, explore GEN AI training courses to build your expertise.
- Document and Collaborate: Transparency is key for AI model acceptance. We need to document assumptions, data sources, and algorithms thoroughly so that they’re clear to HTA reviewers. At the same time, engaging with regulators early helps align AI applications with their requirements and builds trust.
Let’s Keep the Conversation Going
AI is shaking things up, but it’s not about replacing what we do. It’s about finding ways to work smarter and open up new possibilities in health economics. We’ll need to share what works, what doesn’t, and how we can make these tools fit into our world.
