Russ Altman’s Testimony Before the U.S. Senate Committee on Health, Education, Labor, and Pensions

Executive Summary

AI is already improving care for patients and accelerating drug discovery. Properly deployed, these tools can make patient care more personalized, fair, and affordable, and sharpen drug discovery and development pipelines. However, healthcare systems need frameworks to evaluate and deploy safe AI tools while the drug discovery field needs stronger data sharing, adequate infrastructure, and an expert workforce. To realize the full potential of AI, policymakers should focus on several key areas.

AI to Improve Patient Care

• Augmenting diagnosis and treatment: The use of AI to analyze medical images is already well established. Beyond imaging, health systems are rolling out tools that read images and analyze records and notes to flag likely conditions for timely follow-up — prioritizing patients and surfacing risks earlier.

• Improving the physician-patient relationship: AI scribes that generate draft summaries of clinic visits within minutes can ease after-hours charting, reduce burnout, and help physicians focus their attention on patients.

• Increasing patients’ understanding and control of their care: Specialized medical LLMs that can explain lab results in plain language or estimate the likelihood of developing a disease enable patients to better understand diagnoses and treatments.

• Evaluating clinical effectiveness and safety: We will only fully realize the benefits of AI tools if healthcare systems thoroughly vet them. Interdisciplinary evaluation, such as Stanford Health Care’s multistakeholder process, should be standard.

AI to Accelerate Drug Discovery

• Improving privacy-preserving data collection and sharing: Incentivizing hospitals nationwide to contribute secure, de-identified datasets would improve generalizability while protecting patients.

• Investing in public computational infrastructure and research capacity: Ensuring that government and academic scientists have adequate resources to build biomedical AI tools and pursue curiosity-driven research is critical to maintaining America’s leadership in life-saving therapeutic innovation.

• Strengthening regulatory evaluation capacity: The FDA needs expertise to assess the novel ways drugs are developed with AI assistance. Academic partnerships offer one way for the FDA to obtain this specialized knowledge and expertise.

• Developing an interdisciplinary workforce: Training programs for biologists, clinicians, and computer scientists to validate and audit AI-generated predictions will be crucial to ensuring that AI augments human expertise rather than replaces it.