Google Cloud: Medical AI Held Back by Skills Shortage

The Impact of AI on Medical Development

Shweta Maniar, who leads the Healthcare & Life Sciences strategy at Google Cloud, recently shared her perspective on the impact of artificial intelligence in the healthcare sector. She emphasized that 92% of drugs fail during phase 1 clinical trials, meaning that only 8% manage to pass this crucial stage. AI has the potential to transform this phase by improving the early screening of candidate molecules. This would allow for the identification of those that deserve to advance to phase 1 and to predict potential failures even before they occur. While AI continues to progress in areas such as drug discovery and diagnostics, it is in this initial phase of drug development that its impact could be most significant.

Beyond Computing Power: A Need for Skills

According to Maniar, the true barrier to the adoption of AI in the medical field is no longer computing power, but rather the upskilling of teams. It is essential for researchers, marketers, and life sciences professionals to take ownership of these technologies. AI tools are no longer reserved for IT teams. The question of how long it takes for these teams to feel comfortable with these tools is crucial. Teams that succeed in integrating these technologies will gradually become more digitally and AI-native, in a fundamentally different way than what is currently being deployed.

Concrete Successes Thanks to AI

Maniar also mentioned successful collaborations with companies like Recursion, which specializes in AI-based drug discovery. Through the use of TPUs and biomapping, Recursion has been able to bring a drug into phase 3 clinical trials, a success partly attributed to their collaboration with Google Cloud. These examples illustrate what Google Cloud aims to make possible in the medical sector.

She expressed her enthusiasm for the life sciences industry, particularly due to the development of sector-specific capabilities. For instance, Gemini is a tool that can be fine-tuned on medical data and has the ability to read and understand genomic information, making it particularly valuable. A growing trend is emerging: clients are increasingly finding less need to fine-tune these models, and the learnings are gradually being integrated into Gemini.

AI Adoption: A Gap Between Europe and the United States

The differences in AI adoption between Europe and the United States are notable. In Europe, as well as in the Asia-Pacific region, issues of cloud sovereignty and regulatory compliance, such as GDPR, are crucial. Google Cloud is therefore investing a lot of time in collaborating with local regulators and hospital systems to meet the specific requirements of each market. This includes discussions with local equivalents of the FDA and major hospital systems to understand how they serve their markets and what their specific requirements are.