AI IRL Podcast Episode 20: Can AI Save Billions in Clinical Research Costs?
When the cost of drug development has increased a few hundred fold, how do you find a way to cut expenses and time?
AI to the rescue.
AI is poised to make a major change in the economics of medical research, according to Crystal Black, Director of Marketing Programs for Saama Technologies, a unified, AI-enabled clinical data analytics platform.
“We’re geared toward the mission of helping pharma companies save lives faster. We believe that trends, data, and analytics can help with that.”
Saama’s primary purpose is to work with life sciences and pharma companies to deliver impactful outcomes from complex data as well as machine-augmented, AI-enabled analytics to drive new types of user experiences.
These experiences include conversational interfaces similar to Siri that can do a deep data dive. The Natural Language Understanding (NLU) engines allow users to quickly find information without having to go through mountains of data tables and dashboards.
This requires bringing in data from different areas and ensure it’s harmonized which can be extremely difficult. But AI can take the data in Excel files with pivot tables and produce actionable insights in 15 minutes instead of 15 days of data crunching in a spreadsheet.
Bringing Down Drug Costs
“Drug development costs are skyrocketing. Just a few decades ago, the cost to run a clinical trial was $30 million. Today it’s more like $3 billion. It’s crazy.”
Costs are rising in part due to inefficiencies in how the industry is processing the data.
“The data issues fall into a couple of areas,” said Black. “Too much data, as well as not enough data.”
The tonnage of data is mind-boggling:
- Patient monitoring devices produce over 86,000 readings per day.
- By 2025, 70% of clinical trial participants will be using wearable technology, creating even more data.
- There are 2 billion data points in a typical phase two trial that runs for six months with a hundred patients. That equals 200 billion data points.
On the “too little data” side, 80% of healthcare data is unstructured.
That unstructured data goes untapped and that creates delays within patient recruitment, site inefficiencies, issues with patient engagement, and retention. These can add years to a trial.
Saama is trying to help pharma companies tease out the wisdom that’s sitting in biomedical texts and medical notes. This is an area where AI is particularly helpful.
Using NLU, the AI parses all of that unstructured data and formats it into structured data that can be analyzed and centralized where researchers can start to find meaning.
Properly organizing the information can help researchers quickly find the right principal investigator and the right patients for clinical trials.
Finally, you can also use collective insights from historical trials to modify current and future protocol designs. These insights from past field trials can be useful to avoid repeating errors, omissions, or even safety concerns.
If you don’t use iTunes, you can use this link here.