00:00:00 Dr. Scott Campbell shares his diverse educational background, having earned degrees from the University of Michigan, UCLA, and Columbia University. Growing up in La Hoya, California, Dr. Campbell was exposed to high academic standards and excelled in sports. He attended the University of Michigan, where he played baseball, but found the academic rigor challenging. After graduation, he returned to San Diego to coach basketball, but his plans to join a D1 school did not materialize. Instead, he pursued a Master's in Public Health from UCLA, which broadened his perspective on health education. Dr. Campbell then went on to medical school at Columbia University and began his medical career in internal medicine, transplant, and critical care before transitioning to emergency medicine. Recently, he became the chief AI officer of Zero Hour Medical, advising various organizations, including NASA and the Mars mission, on their AI transformation strategies in healthcare.

00:05:00 Dr. Campbell shares his personal journey from dreaming big as a college student to having a successful career in emergency medicine and now focusing on machine learning and A.I. in healthcare. He attended Columbia University Medical School, where he had a great experience with close friends, and later worked at Kaiser San Francisco's Permanente Medical Group. After a few years, he felt drawn to machine learning and A.I., and began learning how to build models and became involved with a group at Children's Hospital in Orange County, aiming to create a new American Board of Artificial Intelligence and Medicine. Dr. Campbell left clinical practice to focus full-time on A.I. and is now part of the advisory board and teaches classes to bring AI literacy to the medical profession. Throughout his journey, he emphasizes the importance of having fun, camaraderie, and dreaming big.

00:10:00 In this section, Dr. Scott Campbell shares his journey from medical school to a successful career in Northern California, and his current excitement about the potential of AI in clinical medicine. He emphasizes the importance of camaraderie and positive relationships throughout his journey, including his connection with the podcast's host. Dr. Campbell discusses how his diverse interests and the camaraderie of his emergency medicine colleagues led him to specialize in that field, and how his close relationships with his medical group partners helped him avoid burnout during his career.

00:15:00 Dr. Scott Campbell discusses the importance of having a positive work culture in the medical field. He shares his experiences at TPMG (Permanente Medical Group at Kaiser San Francisco) and his emergency department, where the selection of partners played a significant role in creating a harmonious work environment. The culture of these organizations emphasized doing the right thing for patients, which in turn improved the standard of care for both the physicians and their patients. Additionally, the San Francisco Emergency Physicians Association, established in 1987, provided a platform for emergency physicians from different hospitals to communicate and collaborate, fostering camaraderie and unity within the industry.

00:20:00 Dr. Scott Campbell shares his experiences as a long-term president of a community of emergency medicine providers in San Francisco. He emphasizes the unique sense of community among the group and the significant challenges they face, including overcrowding and long patient wait times. Dr. Campbell discusses strategies implemented in San Francisco to address these issues, such as creating a sobering center to handle intoxicated individuals and reduce bottlenecks in emergency departments. This initiative, which was one of the first in the country, has reportedly freed up approximately 60,000 emergency department visits and 6,000 ambulance trips over the past 20 years, with only one unfortunate death.

00:25:00 In this section, Dr. Scott Campbell discusses creative solutions implemented in San Francisco to address the issue of overcrowding in emergency departments due to patients with behavioral health needs. One such solution was the establishment of a "door Center" for individuals identified by police as likely to be placed on a 5150 psychiatric hold, allowing for communication with social workers and avoiding hospital admission for those who didn't require it. Another solution was the creation of a medical respite center for homeless and marginally housed individuals who didn't need a hospital but couldn't be on the street. Despite not being overly successful, Dr. Campbell emphasizes the importance of continuing to explore these creative capacity carve-outs and using AI to identify individuals who could be served in a better setting than an emergency department. He also mentions the intertwined nature of behavioral health and emergency medicine, with up to 25-30% of patients in an emergency department seeking behavioral health care. The discussion touches on the potential use of both human and AI interventions to help identify and divert patients from emergency departments to more appropriate care settings.

00:30:00 Dr. Campbell discusses the competition for hospital beds and the potential role of AI in optimizing bed usage. He explains that currently, prioritization in emergency departments is based on the urgency of procedures, often leaving patients like grandmas with pneumonia waiting for extended periods. Dr. Campbell suggests that AI could help determine bed usage based on real-time data and optimize throughput by stretching hospital operations beyond the traditional 9 to 5 schedule. He emphasizes the importance of collaboration with hospital administrators and having a clear understanding of their priorities and resources. Dr. Campbell also mentions the need for real-time situational awareness to make informed decisions and avoid the mistakes made during the pandemic when data was outdated.

00:35:00 Dr. Scott Campbell shares his journey with Artificial Intelligence (AI), starting from his interest in predictions and data analysis during his graduate studies and career in medicine. He mentions the influence of his background in management consulting and the siloed data structure at his medical group. Dr. Campbell then discusses his discovery of machine learning and his involvement with the American Board of Internal Medicine (ABIM) group. He highlights the cutting-edge applications of AI in healthcare, emphasizing the importance of understanding the entire AI umbrella beyond natural language processing. He mentions advancements in image recognition, particularly in pathology and radiology, robotics, and robotic process automation. Dr. Campbell also addresses the challenges primary care doctors face with patient communications and proposes simple solutions using AI to manage their inboxes.

00:40:00 Dr. Scott Campbell discusses the application of AI in clinical medicine, specifically in managing and prioritizing messages for clinicians and predicting clinical outcomes. He mentions the use of process automation, spam filters, and chatbots to filter non-urgent messages, as well as the use of sentiment analysis and image recognition to identify important messages. The most significant application of AI, according to Dr. Campbell, is machine learning, which can predict numbers, events, or classes, such as cancer or spam. He also mentions the potential for AI to disrupt existing clinical decision-making tools, like calculators, by using near-real-time data and machine learning approaches. However, he cautions that large language models (LLMs) can be expensive to train and monitor, and healthcare organizations may struggle to allocate the necessary resources to implement and maintain AI systems.

00:45:00 In this section of the podcast episode "Reach For The Stars & Let AI Help You Do It" with Dr. Scott Campbell, the discussion revolves around the current state and future direction of AI investments. Dr. Campbell expresses that AI projects are becoming more like startups, and the culture shift towards risk-taking could lead to a challenging period for capital expenditures in the next 5 to 10 years. He suggests focusing on simple models for clinical decision support as a more promising area for AI investment. The conversation then shifts towards prioritizing decision support AI over large language models due to the high frequency of decision-making in healthcare. Dr. Campbell also shares his vision for training 16,000 physicians on AI resources, focusing on identifying AI talent and incubating small groups to build a clinical decision support network. The conversation emphasizes the importance of keeping it simple and focusing on decision support for organizations and delivery systems.

00:50:00 Dr. Scott Campbell explains the strategy for implementing AI and machine learning in medical groups. He emphasizes the importance of mastering logistic regression and building classifiers to move the field forward. Dr. Campbell also mentions the need for advocacy and education to healthcare system leaders regarding AI and the shift from revenue-based to value-based medicine. He explains that large medical groups are not currently funding AI efforts, but rather buying or experimenting with it. The fear of risk and lack of in-house talent are major challenges. Dr. Campbell encourages clinicians to learn AI and become fluent in the technology, as their domain expertise is valuable. He also mentions that venture capital is a significant source of funding for AI projects, with some investors taking a more strategic approach by building incubators and leveraging talent in specific healthcare areas. Additionally, he mentions the interest of single family offices in this field.

00:55:00 Dr. Scott Campbell discusses the current state of AI investments in the startup world, particularly in private equity. He explains that due to the unpredictability and long-term nature of AI projects, traditional private equity models may not be the best fit. Instead, he suggests that private equity groups should consider partnering with AI companies earlier in their life cycle. Dr. Campbell also shares his experience with NASA and the Mars mission, where AI is being used to optimize astronaut healthcare in space, predict potential health issues, and even diagnose and treat conditions with limited resources.

01:00:00 Dr. Scott Campbell discusses the importance of Knowledge Graph databases, synthetic data, and Federated learning for NASA's Mars mission and healthcare. He emphasizes the transition from relational databases to Knowledge Graph databases, which are 10 to 100 times more powerful and allow leveraging relationships between entities. Knowledge Graphs will be essential for solving complex problems in healthcare and on Mars. Synthetic data, digital twins of patients, will help address the lack of cases and will be useful for model building. Federated learning, a method for collaborative machine learning, enables multiple hospitals to work on the same problem without having to send their data to a central server, reducing time and costs.

01:05:00 In this section of the "Reach For The Stars & Let AI Help You Do It", Dr. Scott Campbell discusses the potential of Federated Learning, synthetic data, and large language models in healthcare. Federated Learning allows data processing and model training to occur on local devices without the need to transfer data to a central server, ensuring data privacy. Dr. Campbell highlights NASA as an example of leveraging synthetic data and Federated Learning to predict potential issues for astronauts, using predictive modeling on Earth and then refining the models in space. In healthcare, these technologies can help address limited data issues and improve predictions, such as identifying those most likely to develop conditions like kidney stones. Dr. Campbell emphasizes the importance of domain experts, data scientists, business developers, and AI engineers collaborating to build effective AI solutions, comparing the process to building a rock band.

01:10:00 Dr. Scott Campbell discusses the importance of AI literacy for physicians and shares a cautionary tale about a famous EMR company that lost over a billion dollars on building a sepsis predictor. The company made two major mistakes: they edited features in the model around respiratory rate, which is not always accurately calculated by nurses, and they trained the model on a limited number of hospitals in a specific geographic region. When the model was implemented in other regions, it did not perform well. Dr. Campbell emphasizes the importance of using local data and having domain experts involved in the development process. He also introduces the concept of federated learning, where AI systems from different regions or organizations can share data and learn from each other without giving away sensitive information. This approach can lead to more accurate and effective AI models in various fields, including healthcare and drug discovery.

01:15:00 Dr. Scott Campbell discusses the potential of using Federated learning on people's phones for precision medicine and population health. He shares an example of congestive heart failure patients participating in a Federated model, which could lead to more accurate predictions and personalized treatment plans. Dr. Campbell also mentions the concept of "precision population health" or "precision public health," where people will be grouped differently based on large data sets, leading to new discoveries. He emphasizes the importance of continuing to push forward with AI in healthcare and not being distracted or risking an "AI winter."