AI Transforming Medicine: Stanford Dean Lloyd Minor's Vision

Dr. Lloyd Minor, Dean of Stanford Medical School and Vice President for Medical Affairs at Stanford University, believes that artificial intelligence (AI) represents the most significant moment in medicine since the invention of antibiotics. He envisions revolutionary changes in diagnostics, drug discovery, and therapeutics, all contributing to improved health and well-being.

The Role of the Dean of Stanford Medical School

Dr. Minor describes his role as fundamentally centered on people. He works with faculty, students (medical, PhD, master's, clinical fellows, and residents), and staff to advance the mission of Stanford Medicine, which comprises three synergistic components:

1. Patient Care: Providing advanced medical treatment.
2. Research: Driving innovations that improve patient care.
3. Teaching: Educating future medical professionals in the art and science of medicine.

His responsibilities include garnering resources to support these endeavors and leading strategic planning initiatives for the entire enterprise. He acknowledges the delicate balance required to manage the diverse and sometimes divergent interests within an academic medical center, emphasizing the need to find synergies and approach conflicts with fairness.

AI's Transformative Impact on Healthcare

Dr. Minor highlights that AI's application in healthcare is not entirely new, citing electronic patient records and prescription systems as early, rudimentary forms of AI that enhanced safety by preventing medication errors. However, with advancements in algorithms and data processing, AI is now capable of far more sophisticated applications.

Pros of AI in Healthcare

AI promises to make healthcare delivery more accessible, equitable, and safer, while also enhancing discovery processes from basic science to clinical trial design.

Enhanced Accessibility and Safety

• Dermatology Example: Stanford dermatologists trained a neural network using smartphone images of skin lesions, annotated with their pathology (cancerous or not). This AI model performed as accurately as board-certified dermatologists in identifying malignant lesions from new images. This technology could significantly improve access to dermatological screening in rural areas or for individuals facing appointment difficulties, allowing primary care physicians to quickly assess the severity of a lesion.
• Diagnostic Imaging: AI can analyze digital diagnostic images (X-rays, CTs, MRIs). Currently, about 4% of human interpretations of these images miss clinically significant findings. AI can act as an assistant, flagging suspicious areas for radiologists or comparing current images with past ones to detect changes, thereby preventing errors and increasing efficiency.
• Pathology: AI can analyze hundreds of pathology slides from tumors. For rare conditions, a human pathologist might only see a handful of cases in their lifetime. AI, trained on data from numerous health systems, can process far more data, leading to increased accuracy and precision in identifying salient tumor features.

Revolutionizing Drug Discovery

• Protein Structure Prediction: Early applications of machine learning enabled accurate prediction of protein structures from their sequences, accelerating drug discovery.
• De Novo Drug Design: The future holds the possibility of designing drugs from scratch based on biological data about a condition, with companies already focusing on this area. This could lead to a revolution in drug discovery, though the timeline remains uncertain.

Earlier Detection and Precision Health

Improved diagnostics driven by AI will lead to earlier detection of diseases. This aligns with the concept of "Precision Health," which focuses on predicting and preventing disease, rather than just treating it after it occurs (Precision Medicine). Just as airplane engines are constantly monitored, AI could enable continuous monitoring of human health, allowing for early intervention.

Wearables and Data-Driven Prevention

The rise of consumer wearables (Apple Watch, Whoop, CGM devices) generates vast amounts of health data.

• Apple Heart Study: A large-scale study involving 700,000 participants demonstrated that the Apple Watch could accurately detect atrial fibrillation, a common heart arrhythmia that often goes unnoticed but increases stroke risk.
• Continuous Monitoring: Ongoing work in non-invasive blood pressure monitoring and the increasing use of glucose monitors suggest a future where individuals have a moment-to-moment picture of their health, enabling proactive action based on early signs.
• Empowerment and Agency: These devices provide individuals with real-time information about their bodies, fostering a sense of agency and empowering them to take a more active role in managing their health.

Dr. Minor emphasizes a shift from a passive attitude towards health (relying solely on doctors when sick) to one where individuals are primarily responsible for and engaged in their health, with physicians acting as partners.

Reshaping Medical Education

AI will significantly alter medical education, moving away from rote memorization towards critical thinking and data utilization.

• Reduced Memorization: The need to memorize drug dosages, for instance, has already diminished with electronic prescribing. Future medical education will further de-emphasize memorization, focusing instead on understanding how to use data sources, critically evaluate AI-generated information, and apply AI models responsibly in patient care and research.
• Virtual and Augmented Reality: VR and AR are already being integrated into the curriculum.
  • Anatomy: Students can use VR/AR to explore 3D anatomical structures, manipulate muscles, and understand their functions in ways impossible with traditional cadaver dissection or textbooks.
  • Surgical Training: Neurosurgeons can practice complex operations virtually, using VR to visualize tumors and their relationship to blood vessels, akin to a flight simulator for surgery, reducing risk and improving preparedness.

Nutrition Education

While nutrition education is increasing in medical schools, Dr. Minor acknowledges that it needs further emphasis. He highlights the importance of integrating nutrition into both scientific and clinical curricula, enabling future doctors to effectively counsel patients on lifestyle interventions.

Ethical Dilemmas and Safeguards

Dr. Minor, through the "RAISE Health" initiative (Responsible AI for Safe and Equitable Health), addresses the critical ethical considerations surrounding AI in healthcare.

• Privacy: AI, especially large language models integrating social media data, poses new privacy challenges. Even seemingly anonymized queries could potentially identify individuals by linking various data sources. Solutions include bringing AI models into individual delivery systems to prevent data leakage and prioritizing privacy in regulatory frameworks.
• Bias: AI models are only as good as the data they are trained on. If training data is biased (e.g., predominantly from white men), the AI may yield inequitable recommendations. Mitigating this requires inclusive studies and diverse data sets.
• Patient-Provider Relationship: AI should augment, not supplant, the human connection between patients and healthcare providers. AI tools could free up providers from administrative tasks, allowing them to be more present and engaged with patients.
• Public Trust and Transparency: A significant challenge is the public's distrust of AI in healthcare. This can be addressed through transparent communication about how AI is used, its benefits, and the safeguards in place. Public engagement and the option for individuals to opt-in to data usage are crucial for building trust.
• Safety and Control: The prospect of AI and robotics in critical healthcare settings (e.g., administering drugs in an ICU) raises concerns about potential failures. While AI can reduce human error (e.g., in chemotherapy calculations), robust oversight and human involvement remain essential.

The Role of Tech Giants and Regulation

Large retailers like Walmart and Amazon are entering healthcare, aiming to improve efficiency and accessibility through clinics and pharmacy services. Tech firms like Apple and Google are also investing in health initiatives, such as the Apple Heart Study and the Baseline project (a longitudinal study tracking health data).

• Regulatory Landscape: Healthcare is a highly regulated industry, which presents challenges for rapid innovation. However, regulatory bodies like the FDA and the Office of the National Coordinator for Health Information Technology are actively engaging with experts to understand AI's implications and develop responsible regulations.
• Data Privacy: The widespread collection of health data from wearables and other sources raises concerns about its potential misuse for targeted advertising or other purposes. Strict regulations and opt-in mechanisms are necessary to protect individual privacy and prevent breaches of trust.

Addressing Systemic Healthcare Challenges

The U.S. healthcare system, characterized by its complexity, cost, and focus on "sick care," faces significant challenges.

• Incremental Change: Dr. Minor believes in an incremental approach, where academic health systems like Stanford rigorously study and implement innovations (e.g., AI in radiology) and then share successful models to drive broader adoption.
• Incentives for Prevention: Shifting incentives, such as those seen in Medicare Advantage plans, towards keeping people healthy rather than just treating illness, is crucial.
• Social and Environmental Determinants of Health: Approximately 70% of disease determinants are socially and environmentally mediated (e.g., access to healthy food, behavioral health issues). Addressing these requires rigorous research and policy changes.

The Future of Medicine

Dr. Minor offers optimistic predictions for the next 5-10 years:

• Early Detection of Cancers: Significant advancements in diagnostic tests for early detection of historically late-diagnosed cancers (e.g., pancreatic, ovarian) are expected, potentially through blood tests for cell-free DNA.
• Reduced Hospitalizations: Improved in-home monitoring will allow for earlier intervention in chronic conditions, preventing severe declines and reducing hospitalizations.
• Rejuvenated Healthcare Workforce: AI tools can help alleviate physician burnout by reducing administrative burdens (documentation, managing patient inboxes), allowing healthcare providers to focus more on direct patient interaction.
• Organ Growth and 3D Printing: While not immediate, the ability to grow and 3D print living tissues and organs (e.g., heart chambers for congenital heart disease) is a promising area of research that could revolutionize treatment options.

He emphasizes that this is an incredibly exciting time to be in the life sciences due to the convergence of various scientific and technological fields (3D printing, AI, engineering) being applied to biomedicine.

Personal Health Practices

Dr. Minor shares his personal health practices:

• Diet: Primarily fresh fruits and vegetables, with meat consumed sparingly.
• Exercise: Three gym sessions per week, with daily cardio.
• Intermittent Fasting: Practices intermittent fasting, having found it beneficial during the COVID-19 pandemic.
• Hobbies: Plays the cello, which he rediscovered during the pandemic.

Vision as Surgeon General

If he were Surgeon General, Dr. Minor's priorities would be:

• Accessible Health Information: Making health information readily available and accessible to everyone, regardless of socioeconomic status or education level.
• Responsible Regulation: Focusing on the responsible regulation of factors known to harm health.
• Discovery Pipeline: Directing research and discovery towards diseases and disorders most prevalent in society, such as new treatments for high blood pressure, high blood glucose, and obesity.

AI and Self-Diagnosis

Regarding the use of AI for self-diagnosis, Dr. Minor advises a healthy degree of skepticism. While AI models can be surprisingly accurate, especially for well-defined conditions, they can also "hallucinate" or provide incorrect information when pushed with detailed queries. He believes that "Dr. Google" has generally been a positive development, empowering individuals with more health knowledge, but emphasizes the need for critical evaluation.

Key Messages

• For Consumers: AI in healthcare has a stronger potential for good than for bad, but active engagement, awareness of potential misapplications, and the option to opt-in to data usage are crucial.
• For Aspiring Med Students: There has never been a better time to enter the life sciences, given the unprecedented opportunities for innovation and impact.

Dr. Minor stresses the importance of effective communication, transparency, and vulnerability in leadership, especially in navigating the complex and rapidly evolving landscape of AI in healthcare. He acknowledges the need to rebuild public trust, particularly after the COVID-19 pandemic, by openly addressing uncertainties and admitting mistakes.