Leveraging AI for Better Outcomes Across Drug Development - with Patricio La Rosa of Bayer
🎯 Summary
Podcast Summary: Leveraging AI for Better Outcomes Across Drug Development - with Patricio La Rosa of Bayer
This 33-minute episode of the AI and Business Podcast features Patricio La Rosa, Head of End-to-End Decision Science at Bayer Crop Science, discussing the transformative role of Machine Learning and AI across the drug development lifecycle, drawing parallels between agriculture and life sciences. The central narrative focuses on how AI is moving clinical trials from being purely experimental objects to more human-centered, efficient, and precise endeavors.
1. Focus Area
The discussion centered on the application of AI/ML in Drug Development and Clinical Trials. Specific topics included:
- Predicting patient outcomes and tailoring patient engagement strategies.
- The critical role of reliable biomarker identification using multi-modal sensing.
- Optimizing clinical trial design, power calculation, and patient recruitment/retention.
- The intersection of behavioral science (gamification, contextual intelligence) and data science in ensuring trial adherence.
2. Key Technical Insights
- Multi-Modal Sensing for Biomarkers: AI is crucial for integrating data from diverse sensing modalities (e.g., genetic sequencing like metagenomics, fMRI) to identify robust, time-persistent biomarkers, especially for rare diseases, overcoming the limitations of single-source data.
- LLMs in Patient Identification: Large Language Models (LLMs) are being leveraged to scrutinize clinical notes and data to identify and scrutinize potential candidates for trials, particularly for rare conditions, improving study power.
- Active Learning and Experimental Design: Techniques like active learning, borrowed from genetic marker identification, are being adapted to dynamically optimize clinical trial recruitment by sampling the patient space step-by-step to determine the most informative subjects needed to prove or disprove a hypothesis economically.
3. Business/Investment Angle
- Efficiency in Trial Design: The ability to define and trace clear effects via robust biomarkers is key to creating “powered and precise” trials, reducing Type I and Type II errors, and ensuring a viable path to drug approval.
- Patient Retention as ROI: Investment in AI-driven patient engagement (using sentiment analysis and gamification) is critical because poor adherence jeopardizes the significant financial investment made in recruitment and trial setup.
- The “Last Mile” of Analytics: Behavioral science applications, like gamification, address the “last mile” problem—ensuring the adoption and consistent execution of analytical insights or trial protocols by the human subjects involved.
4. Notable Companies/People
- Patricio La Rosa (Bayer Crop Science): The expert guest, bringing 20 years of experience from both agriculture and healthcare, emphasizing the connection between technical models and real-world workflows.
- Matthew Damello (Emerge AI Research): The host, guiding the conversation toward the intersection of technology, ethics, and human-centric design.
5. Future Implications
The industry is moving toward highly human-centered clinical trials where AI facilitates personalized communication and engagement, fostering a partnership between researchers and patients. There is a strong push for transparency and ethical use of AI, ensuring patients understand how technology is being used to support their participation and the overall search for a cure, rather than feeling manipulated. The challenge remains in ensuring that complex, AI-identified biomarkers (like those from fMRI) are consistently reliable across different sites and equipment.
6. Target Audience
This episode is highly valuable for R&D professionals, Clinical Operations leaders, Data Scientists in Pharma/Biotech, and Healthcare Technology Investors who need to understand the practical, strategic, and ethical deployment of advanced AI/ML within the highly regulated drug development pipeline.
🏢 Companies Mentioned
💬 Key Insights
"it's uncanny how especially from the data side straight through management, you start to hear more and more, you know, we do need to think of these folks like the way our friends in retail think of their experiences because we want to fulfill the Hippocratic oath because we want to do what's best for the patient because these institutions have it really figured out to get people through a checkout line as fast as possible, not just in the name of the bottom line, but for their own good and making sure that the emergency room is clear, et cetera."
"So we need to treat human clinical trials also from the perspective that we're working with the future customers."
"it's especially difficult given this point of AI adoption, both in these really complex industries and in the culture that AI of such an objective nature, AI of such a deterministic nature as we see in the actual development of these drugs, how they are chosen on the biomarkers that runs so differently than what we see on the generative side with more probabilistic technologies."
"what an AI with advanced machine learning, the same machine learning that's helping pick the proteins would use machine learning in a different way to generate text that would give recommendations to a patient. And we need different words for these..."
"But in order for me to do this in an active dynamical manner, I can also rely on techniques based on AI like active learning experimental design."
"The protein being able to re-engineer, you know, what we're seeing at that molecular level is so ingrained in kind of the STEM fields and, you know, needs such an objective approach to how you're going about these changes at literally a microscopic level. But the ability to bring somebody along a clinical trial is to tell them a story... That's the humanities. That's how do we tell that story?"