Generative AI in Healthcare

The National Academy of Medicine’s report, Generative Artificial Intelligence in Health and Medicine, paints a clear picture: AI is reshaping medicine—but without governance, it risks becoming a liability.

Key Takeaways:

Clinical Integration

GenerativeAI is already enhancing workflows through automated clinical documentation, note generation, and decision support. In real-world pilots, AI-generated summaries often outperform human-written ones in clarity, consistency, and speed.

Patient Communication & Engagement

AI tools are being deployed to draft empathetic patient responses, deliver multilingual health education, and offer personalized self-care recommendations. But human oversight remains critical to ensure relevance, accuracy, and trust.

Equity and Research

GenAI can identify disparities in treatment and outcomes, informing more equitable interventions. However, models trained on non-representative data risk perpetuating systemic bias and health inequalities.

AI in Clinical Trials & Discovery

From streamlining patient recruitment to synthesizing research data, GenAI is accelerating timelines across the pharmaceutical pipeline—especially in rare disease modeling and drug repurposing.

Risks

- Privacy vulnerabilities in GenAI data feedback loops

- Algorithmic hallucinations and fabricated clinical claims

- Embedded bias in training datasets

- Performance drift without algorithmic vigilance

- Lack of explainability in high-stakes AI decisions

The integration of GenAI into healthcare systems demands more than optimism—it requires strategic investment, regulatory readiness, and risk-based execution. In Europe, the clinical use of AI must comply with the AIAct, while the healthcare sector is already governed by NIS2 and GDPR, enforcing strict requirements around resilience, data protection, and operational risk management.

Based on ENISA reports and my own experience, the health sector remains one of the least mature industries in cybersecurity. This gap makes it a top target for ransomware and large-scale data leaks. But beyond digital risk, there’s a deeper issue: patient safety. Healthcare is not just a data-driven industry—it’s one that holds human lives in the balance. Failing to implement proper controls over AI systems or to secure digital operations could directly jeopardize patient outcomes, interrupt critical care delivery, or cascade into disruptions across the broader human safety supply chain —impacting pharma, emergency response, or national health infrastructures.

This is why it's imperative that healthcare leaders treat AI governance and cybersecurity as foundational pillars of care delivery, not just technical enhancements.

Link

Author: Sebastian Burgemejster