Speaker Introduction:
Vinod Khosla, a renowned entrepreneur, investor, and mentor, delivered the closing keynote address at a healthcare conference.

Systemic Change:
Khosla believes that the healthcare system cannot change from within and requires entrepreneurial innovation. Collaboration between entrepreneurs and healthcare professionals is essential, but the starting point of view matters.

Misdiagnosis:
Misdiagnosis in healthcare is as significant a killer as breast cancer, challenging the assumptions of many. Khosla emphasizes the need to focus on the bottom half of the doctor pool, especially in countries like India.

Cognitive and System Factors:
Cognitive limitations and biases impede accurate diagnosis, influenced by external factors such as media coverage. Khosla cites studies that demonstrate how increased media attention on ADHD corresponds with a higher number of diagnoses.

Accurate Citations:
Khosla’s presentation includes references to credible sources, encouraging attendees to review the associated articles. The speaker reviewed these sources to ensure their accuracy before delivering the keynote address.

Premature Closure and Second Opinions:
Misdiagnosis often occurs due to premature closure, where doctors rush to a diagnosis without keeping an open mind. Second opinions often lead to different diagnoses, indicating a lack of consistency and accuracy in medical decision-making.

Evidence-Based Medicine:
Many medical recommendations lack clinical evidence, with only 11% supported by data. Opinions, rather than science, often drive medical practices, leading to uncertainty and inconsistency.

Variability in Surgical Recommendations:
Cardiac surgeons vary widely in their recommendations for surgery, even when presented with the same data. Surgeons’ opinions on the same patient can change over time, suggesting a lack of standardized criteria.

Disagreement in Stenosis Diagnosis:
Experts disagree on stenosis diagnosis, with a 60% disagreement rate. They also disagree with themselves when presented with the same data a second time, highlighting inconsistency in decision-making.

Estimates vs. Science in Cancer Screening:
Estimates of cancer screening impact on cancer deaths vary widely, indicating a lack of scientific consensus. Medical practices often rely on tradition rather than rigorous scientific evidence.

Antipyretic Use and Fever:
Despite the lack of evidence, fever reduction using antipyretics is a common practice in medicine. A study showed a higher mortality rate among children whose fever was reduced.

Communication and Patient Understanding:
Americans generally comprehend medical instruction at the fifth grade level. Health professionals often fail to communicate effectively with patients, leading to misunderstandings and potential errors.

Need for Change in Medical System:
Medical studies, which form the basis of medical practice, are often more likely to be wrong than right, leading to a need for a system change. Conflicts of interest, such as companies promoting their products, further complicate the system.

Empowering Patients:
Khosla believes that patients should be empowered to take charge of their health. Over the next decade, patients will become the CEOs of their own health, with 80% of what MDs currently do being replaced by better standards of care.

Human Element in Care:
Khosla emphasizes the importance of the human element in care, but argues that the most humane care should be provided by those with the most mirror neurons, regardless of their academic degree.

Integrative Medicine:
Khosla dismisses the concept of truly integrative medicine, as humans can only keep a limited number of variables in mind when making decisions. Electronic data can provide more comprehensive and accurate insights into a patient’s health than traditional integrative medicine approaches.

Machines in Healthcare:
Machines, with their ability to process large amounts of data, will play a crucial role in diagnosis, prescription, and alternative recommendations. Machines can provide more accurate probabilities of outcomes and adverse effects, empowering patients to make informed choices.

Examples of AI in Healthcare:
Khosla cites examples of AI systems, such as those used in Oregon for diagnosis and triage, demonstrating high accuracy despite being in early stages of development.

AI as Digital First Aid Kits:
Current AI systems in healthcare are akin to toddler MDs or digital first aid kits, with limitations but the potential for significant improvement.

Simple Applications of AI:
Khosla highlights the example of a device that can diagnose ear infections through an iPhone attachment, eliminating the need for doctor visits and reducing antibiotic overuse.

AI for Heart Monitoring:
Khosla demonstrates a device that transforms an iPhone into an EKG machine, allowing patients to monitor their heart health more frequently and comprehensively.

Conclusion:
Khosla emphasizes the need for innovation and imagination in developing AI-based medical systems, with the potential to greatly improve healthcare outcomes and empower patients.

Technological Advancements in Healthcare:
Vinod Khosla introduces a self-prescription glasses device that generates 20 million data points in two days. These advancements surpass the capabilities of human internists in terms of data collection and analysis. Cost-effective devices like breath analyzers can detect conditions like ketosis, asthma attacks, and various types of cancer.

Mental Health Diagnostics:
Khosla discusses the Ginger.io app, which collects extensive behavioral data through user interactions and activities. The app provides unparalleled insights into mental health, exceeding the capabilities of any human psychiatrist. The collected data offers a comprehensive and evidence-based approach to mental health management.

Practice-Based Evidence in Medicine:
Khosla emphasizes the importance of practice-based evidence, utilizing millions of patient records to determine the reliability of medical claims. This approach contrasts with the traditional reliance on drug company-approved studies, which may have limitations.

Healthcare Innovation Curve:
Khosla asserts that the current healthcare innovations are merely the beginning of a transformative journey. These innovations will become obsolete in the coming years, as the field continues to evolve rapidly. The current devices represent point innovations that provide incremental improvements, paving the way for more comprehensive solutions in the future.

Medical Innovations through AI and Data Science:
Khosla believes that every medical vertical will undergo reinvention due to innovations in Artificial Intelligence (AI) and data science. He suggests that these innovations, initially appearing vulnerable to criticism, will transform into a wave that will have a profound impact on healthcare.

Data Sciences: The Future of Healthcare:
Khosla boldly claims that in the next 10 years, the data sciences will contribute more to improving healthcare than all biological sciences combined. He provides an example from Kaiser Permanente, where a data scientist analyzed incoming stroke victims’ data and reduced death rates by 40% by changing protocols based on data analysis.

The Practice of Medicine as a Science:
Khosla advocates for the transformation of the practice of medicine into the science of medicine, driven by data insights and actionable information. He admits that while he might not be able to predict the exact path, he is certain of the overall direction of change in healthcare.

Embracing Failure and Socially Maladjusted Progress:
Khosla emphasizes the importance of embracing failure in the pursuit of innovation and progress. He quotes Martin Luther King and George Bernard Shaw to support his belief that societal progress depends on those willing to challenge social norms and embrace unconventional approaches.

Automation in Complex Industries:
Khosla cites examples from aviation and finance, where machines have taken over complex tasks, such as flying planes and trading stocks. He believes that driverless cars, a more complex endeavor than medical diagnosis, will soon become commonplace.

AI as a Solution to Global Under-Resourcing in Healthcare:
Khosla highlights the potential of AI-driven systems to address the global under-resourcing in healthcare, particularly in countries with limited medical resources. He suggests that these systems can provide better care, more efficiently, in resource-constrained settings.

Questions for Vinod Khosla about Healthcare Innovation:
Vinod Khosla opened the floor to questions from the audience. One attendee asked about the possibility of partnering with Khosla Ventures to improve the safety and effectiveness of implanted medical devices. Khosla responded by emphasizing the importance of data collection and analysis in healthcare. He also suggested that using personal data, such as that collected by a smartwatch, could potentially provide valuable insights into the performance of medical implants.

Studies After the Study Concept:
Khosla introduced the concept of “studies after the study,” which involves conducting real-world studies of new technologies, such as implants and drugs, after they have been introduced to general practice. He believes that such studies can provide more accurate and comprehensive data than traditional clinical trials, which are often designed to favor the approval of new products.

Data Analysis and the Stanford Study of Vioxx:
As an example of the power of data analysis, Khosla mentioned a study conducted by Nigam Shah at Stanford University. Shah’s analysis of Stanford’s data alone showed that the drug Vioxx should have been banned years before the FDA finally did so. This highlights the potential benefits of using large-scale data to make informed decisions about healthcare products and treatments.

Advice for Engaged Patients and Innovators in Healthcare:
In response to a question about how engaged patients can scale their innovations and attract investors, Khosla emphasized the importance of experimentation and persistence. He encouraged innovators to try various approaches, even if they seem unconventional or risky. He warned against relying too heavily on experts, as their predictions are often no more accurate than random guesses.

Study on Expert Accuracy:
Khosla cited a study by Professor Tetlock at UC Berkeley, which followed 250 experts in different fields for 20 years and tracked their accuracy in making 28,000 forecasts. The study found that the average accuracy of these experts was no better than that of dart-throwing monkeys. Khosla recommended the book “Expert Political Judgment” for those interested in the statistical analysis of this study, and “Future Babel” for a more general audience.