Bringing new drugs and compounds to market has always been expensive, but it’s become increasingly so over the last decade. Considering that up to 90% of drug candidates never make it to market, the biotech and pharma industries have made significant investments in new technologies to increase the chances of approval.
Enter artificial intelligence (AI) and machine learning (ML). Though often associated with number crunching and broad-scope data analysis, many decision-makers now use these tools to select trial sites and/or recruit patients.
Currently, ML and AI-based algorithms are being used to:
- Find healthcare providers (HCPs) within specific regions
- Find HCPs with affiliates to specific hospitals
- Identify communities in terms of shared sponsors
- Find potential inroads, such as referrals, community events, or other engagements
- Identify top sites for investigations and/or trial research
Using these technologies can reduce the number of trial sites and help you enroll the required sample size. But these results aren’t guaranteed. To truly reap the rewards, it’s crucial you avoid common pitfalls.
Let’s take a closer look at four of the most common mistakes sponsors make when implementing AI or ML into the clinical trial development process.
Mistake Number One – Trying to Do Everything On Your Own
Technologies, like AI and ML, are only as effective as the people using them. Without experienced professionals on staff, you won’t know where to start, or how to develop a roadmap your team can follow. Finding qualified workers with these technical skills can be challenging, but it’s getting easier.
If you can’t afford an in-house AI/ML expert, consider working with a technology partner. After an operations evaluation, they can identify skills gaps and help fill out your team.
Mistake Number Two – Failure to Accurately Interpret Your Data
The efficacy of ML and AI-based algorithms are entirely based on the accuracy of the data feeding them. Unfortunately, getting high-quality data isn’t always easy. Especially if you’re just using these technologies for the first time.
You can’t prevent bad data entirely, but there are things you can do to reduce mistakes, including:
- Only using data from reliable sources
- Creating a data profile before each trial that visualizes the information you already have
- Integrating various datasets from separate corporate silos
Many companies are starting to implement Common Data Models (CDMs). This process takes information from multiple databases and converts them into a uniform format, including coding schemes, vocabularies, and standard terminologies.
Mistake Number Three – No Administrative Framework
As we’ve previously mentioned, the successful use of AI and ML requires lots of data. Regrettably, most biotech and pharma companies aren’t able to house all of the information they collect in a single location. Often, it’s spread across multiple databases and stored in different formats.
Without a strong administrative framework in place, there’s no way to accurately gather and interpret all of those statistics and figures. To do that, industry experts say all administrative frameworks should include an plan for four key elements:
- Data integrity
- Data storage and integration
- Data visibility
- Data security
Building a data framework for your team to follow can help reduce human error and lower the risk of data anomalies.
Mistake Number Four – Failure to Implement New Practices and Make Them Habits
Successfully using new technologies isn’t something that happens overnight. It’s a committed process that requires focus and a willingness to adapt. Without a real blueprint to follow and regular check-ins to monitor your team’s evolution, it’s easy to fall back into old habits.
AI/ML in Clinical Trial Development – Takeaways
Tools like AI and ML are becoming the norm when it comes to clinical trial development and recruitment. If you want to stay relevant and agile, it’s worth adopting them in some form.
You don’t necessarily have to buy all in, but it’s good to think about how you and your team can get the most out of your data. Highlighting its importance in your organization encourages a shared purpose. It can improve the quality of your research, reduce the headaches you encounter, and increase your chances of success.
