ai & occupational medicine.
companion to webinar navigating the ai wave.
cameron kiani, md.
manijeh berenji, md, mph.
thu nov 8th, 2024 @ 12:45.
acoem: fall summit 2024.
program & registration link.
Lorem ipsum
Lorem ipsum odor amet, consectetuer adipiscing elit. Ipsum sit convallis justo vivamus sollicitudin aptent amet natoque. Neque sociosqu penatibus curae efficitur placerat dictum curabitur. Efficitur proin nulla pellentesque penatibus gravida ultrices quis. Magna placerat sagittis penatibus nulla urna orci efficitur nostra cras. Vel venenatis aptent integer lacus velit. Nam facilisi per mauris aliquet nibh nascetur ullamcorper dignissim? Penatibus leo sapien magnis luctus justo phasellus tristique ultricies efficitur. Sit enim tempor facilisis gravida malesuada velit litora ultricies conubia.external link.
About
Dr. Cameron Kiani, MD, is an early-career OEM doctor. Dr. Kiani obtained his Bachelor of Arts in Psychology from the University of Pennsylvania. He then earned his Doctor of Medicine from Mount Sinai School of Medicine. He completed his internship at Montefiore/Albert Einstein School of Medicine. He is currently completing his residency and MPH at the Mount Sinai School of Medicine.
Thank you to Dr. Manijeh Berenji, MD, MPH for her support and supervision on this project. Dr. Berenji is a leader in occupational and environmental health, currently serving as Health Sciences Associate Clinical Professor at UCI School of Medicine and Site Director for the Occupational and Environmental Medicine Residency Program. She chairs both the Health Informatics and Environmental Health Sections at the American College of Occupational and Environmental Medicine, focusing on digital health transformation and planetary health. Dr. Berenji also leads the UC Center for Climate, Health, and Equity, emphasizing sustainable healthcare practices. Her work integrates cutting-edge digital health tools to elevate precision medicine and clinician training.
I would like to extend my ongoing gratitude for the guidance, expertise, and dedication of my residency training supervisors:
And last, but certainly not least, to my remarkable co-residents who inspire me to be the best version of myself every day:
Upholding Physician Leadership in the AI Age
Navigating AI, Roles, and the Future of Occupational Careprogram & registration link.As OEM physicians, how do we respond to these evolving dynamics without losing sight of our distinct contributions?How do we articulate and defend the irreplaceable value of physician-led oversight and care in an age increasingly reliant on AI?How do we navigate these trends in a way that motivates the OEM community to protect our professional standards and influence, ensuring we lead rather than concede to the shifting technological landscape?Physicians and non-physician providers each bring overlapping yet distinct value to occupational care, but it’s important to recognize that these roles are not interchangeable. A question that will certainly arise (if not already): In the hands of non-physician providers, does the rise of AI and increased accessibility of information truly close the gap in expertise? If not, why? I bring this up not to single out our colleagues, but to spark reflection and as an illustrative example of why understanding AI presents a fundamental competency.While AI and accessible information empower all providers, they do not replace the depth of specialized training, clinical judgment, and holistic understanding that physicians contribute. The current trend—wherein non-physician providers increasingly carry out primary clinical responsibilities with minimal physician oversight—could be amplified by AI technologies, risking further erosion of the nuanced contributions of OEM physicians.
This shift has significant implications for the value of physician-led care, the demand for our expertise, and our role in mitigating workplace health risks. As the accessibility of information grows—often processed and presented by AI—it can create the perception that the specialized judgment of a physician is no longer needed in the same capacity. However, we know that real expertise isn’t just about accessing information; it’s about interpreting it, anticipating outcomes, and integrating nuanced, human insights to support the health and safety of workers.
---Upholding Physician Leadership in the AI AgeThe rise of AI and the increasing availability of medical information present unique challenges and opportunities for the field of occupational and environmental medicine (OEM). To effectively respond to the evolving dynamics without compromising our distinct contributions, we must clearly communicate and demonstrate the unique capabilities that only experienced physicians can provide. Our value lies in the combination of depth and breadth of clinical training, our ability to synthesize complex information, and our experience in managing the often unpredictable nature of individual patient cases—attributes that cannot be replaced by AI or other healthcare roles.First, it is critical that we reclaim our leadership in integrating AI as a tool within the clinical setting rather than a substitute for our expertise. This requires us to be proactive in adopting AI technologies, demonstrating how AI enhances, rather than diminishes, our medical judgment. We should be the ones setting the standards for how AI augments clinical workflows, ensuring it is implemented in ways that ultimately improve patient care without compromising the depth of our assessments. By doing this, we communicate that while AI can assist with data analysis and decision support, it is the physician who brings insight, holistic judgment, and the ability to weigh multiple, often competing, factors in a manner that serves the best interests of the patient.Second, we need to emphasize and illustrate the specific value of physician oversight. In occupational care, context matters. Understanding the intricacies of workplace health risks, the variability of exposure, individual worker susceptibilities, and balancing regulatory constraints and employer needs requires a depth of experience. Our expertise enables us to provide personalized care and anticipate complex outcomes beyond the pattern recognition capabilities of AI. We must make it clear that real-world healthcare requires more than knowledge retrieval—it requires applying that knowledge with discernment, empathy, and foresight.We should also actively participate in educating others—employers, workers, and even healthcare systems—about what constitutes comprehensive care. This means highlighting examples where physician-led decision-making led to better outcomes that a purely data-driven approach or reliance on non-physician expertise might have missed. Communicating the consequences of overlooking nuanced clinical decision-making will be critical to distinguishing the value of our leadership.Furthermore, we should embrace a complementary approach, where non-physician providers and AI serve supportive roles under physician leadership. By developing collaborative models that maximize the respective strengths of all care team members, we create a structure in which AI and other providers expand the reach of our care without undermining the specialized roles we play. In this way, we ensure the tools and teams in occupational care function cohesively and efficiently, while reinforcing the necessity of physician oversight in achieving the highest quality of care.Finally, we must engage the OEM community, especially those who may be less inclined to embrace AI, to understand that this change is not only inevitable but also offers an opportunity for us to lead. If we shy away from these advancements, we run the risk of being sidelined. By actively leading AI integration and by insisting on robust standards for its use, we affirm our place at the forefront of occupational healthcare—defining rather than following the trajectory of change.The information/AI age indeed make medical information more accessible, but the role of the physician remains irreplaceable in transforming that information into meaningful, ethical, and personalized healthcare. This is the message we must articulate and the value we must protect, not by resisting change, but by steering it towards enhancing rather than diminishing the profound role of physician expertise in occupational and environmental health.
Glossary of Artificial Intelligence (AI) Terms
Welcome to the glossary section of the Companion website for the webinar "Navigating the AI Wave: A Framework for Understanding Occupational Impact and Adaptation." This glossary is designed to provide clear and comprehensive explanations of key AI concepts and tools relevant to Occupational and Environmental Medicine (OEM) professionals. Whether you're new to AI or looking to deepen your understanding, this resource will help you navigate the transformative landscape of artificial intelligence in your field.
1. Fundamental AI Concepts
Artificial Intelligence (AI)
AI refers to the simulation of human intelligence in machines that are programmed to think and learn like humans. It encompasses various technologies that enable machines to perform tasks such as understanding natural language, recognizing patterns, solving problems, and making decisions.
Machine Learning (ML)
A subset of AI, machine learning involves training algorithms to recognize patterns and make decisions based on data. Instead of being explicitly programmed for every task, ML models learn from large datasets to improve their performance over time.
Large Language Models (LLMs)
LLMs are advanced AI models designed to understand and generate human-like text. They are trained on vast amounts of text data and can perform a variety of language-related tasks, such as translation, summarization, and conversation. Examples include OpenAI's GPT series and Google's Gemini.
Generative AI
Generative AI refers to AI systems capable of creating new content, such as text, images, or audio, based on the data they have been trained on. These models can generate original work that mimics the style and substance of their training data.
Transformer Architecture
A foundational design in modern AI, transformer architecture uses mechanisms like attention to process and generate data. Transformers excel at handling sequential data, making them ideal for tasks like language translation and text generation.
Attention Mechanism
An essential component of transformer models, the attention mechanism allows the model to focus on specific parts of the input data when generating output. This helps in understanding context and maintaining coherence in generated content.
Context Window (Context Length)
This term refers to the amount of input data an AI model can consider at one time when generating output. For instance, GPT-3 has a context window of up to 4,096 tokens, enabling it to maintain context over longer conversations or documents.
2. AI Tools and Platforms
ChatGPT
Developed by OpenAI, ChatGPT is a conversational AI model capable of understanding and generating human-like text. It can assist with tasks such as drafting emails, answering questions, and providing explanations, making it a valuable tool for professionals seeking to enhance their productivity.
Claude
Claude is another conversational AI model designed to assist with various tasks through natural language understanding and generation. It offers functionalities similar to ChatGPT, providing users with a versatile tool for communication and information processing.
Google's Gemini
Gemini is Google's advanced AI model aimed at enhancing various applications through improved natural language processing and understanding. It is designed to integrate seamlessly with Google's suite of tools, offering robust support for professional and personal use cases.
Text-to-Speech (TTS)
TTS technology converts written text into spoken words. This tool is useful for creating audio content, assisting those with visual impairments, and enabling hands-free operation in various professional settings.
Retrieval-Augmented Generation (RAG)
RAG is an AI approach that combines retrieval-based methods with generative models. It enhances the ability of AI systems to generate accurate and contextually relevant responses by retrieving information from external databases or documents during the generation process.
3. AI in Occupational and Environmental Medicine
Task Exposure
Task exposure assesses the extent to which specific job tasks can be automated or augmented by AI. Understanding task exposure helps professionals identify which aspects of their roles are susceptible to AI integration and which remain reliant on human expertise.
Human-AI Complementarity
This concept highlights the synergy between human skills and AI capabilities. Human-AI complementarity focuses on how AI can enhance human performance by handling repetitive or data-intensive tasks, allowing professionals to focus on areas that require emotional intelligence, critical thinking, and nuanced decision-making.
AI-Assisted Decision Making
AI-assisted decision making involves using AI tools to support and enhance the decision-making process. In OEM, this could mean leveraging AI to analyze safety data, predict workplace hazards, or recommend interventions, thereby improving overall efficiency and accuracy.
AI in Risk Assessment
AI can significantly enhance risk assessment by analyzing large datasets to identify potential hazards and predict their impact. This enables OEM professionals to proactively address safety concerns and implement effective preventive measures.
AI in Safety Programs
Integrating AI into safety programs can streamline processes such as incident reporting, compliance monitoring, and training development. AI-driven tools can provide real-time insights, automate routine tasks, and facilitate continuous improvement in workplace safety.
4. AI Concepts in the Workplace
AI Integration
AI integration refers to the seamless incorporation of AI technologies into existing workflows and systems. Effective integration ensures that AI tools complement human efforts, enhance productivity, and contribute to achieving organizational goals without causing disruption.
AI Upskilling
AI upskilling involves training professionals to effectively use AI tools and understand their capabilities and limitations. Upskilling ensures that employees can leverage AI to its full potential, enhancing their roles and contributing to organizational success.
AI Reliability and Hallucinations
AI reliability pertains to the consistency and accuracy of AI outputs. Hallucinations refer to instances where AI generates information that is plausible but incorrect or fabricated. Understanding these concepts is crucial for professionals to effectively use AI tools and verify their outputs.
AI in Training Programs
AI can be utilized to develop and deliver personalized training programs. By analyzing individual learning patterns and performance, AI-driven training can provide customized learning experiences that enhance skill development and knowledge retention.
AI in Content Creation
AI assists in content creation by generating text, images, and other media based on specific inputs. In OEM, this can be used to develop safety materials, training resources, and patient-specific content, making information more accessible and tailored to diverse needs.
5. Ethical and Practical Considerations
AI Ethics
AI ethics involves the principles and guidelines that govern the responsible use of AI technologies. Key considerations include fairness, accountability, transparency, and the prevention of bias to ensure that AI benefits society without causing harm.
AI Governance
AI governance encompasses the policies, standards, and regulatory frameworks that guide the development and deployment of AI systems. Effective governance ensures that AI technologies are used ethically, legally, and in alignment with societal values.
AI Bias
AI bias occurs when AI systems produce results that are systematically prejudiced due to biased training data or flawed algorithms. Addressing AI bias is essential to ensure fairness and equity in AI applications, particularly in sensitive fields like medicine.
AI Limitations
Understanding the limitations of AI is crucial for effective utilization. AI systems may lack the ability to comprehend context fully, exhibit biases, or fail to replicate complex human emotions and judgments. Recognizing these limitations helps professionals use AI tools appropriately and maintain critical oversight.
AI Hallucinations
Hallucinations in AI refer to the generation of plausible-sounding but incorrect or fabricated information by AI models. This phenomenon underscores the importance of verifying AI-generated content, especially in professional settings where accuracy is paramount.
6. Advanced AI Concepts Relevant to OEM
Retrieval-Augmented Generation (RAG)
RAG enhances AI models by integrating information retrieval mechanisms, allowing them to access and incorporate relevant data from external sources during content generation. This improves the accuracy and relevance of AI outputs, making them more reliable for professional use.
Federated Learning
Federated learning is a decentralized approach to training AI models across multiple devices or servers holding local data samples, without exchanging the data itself. This method enhances data privacy and security, making it suitable for sensitive applications in healthcare and occupational medicine.
Graph Neural Networks (GNNs)
GNNs are specialized neural networks designed to work with graph-structured data. They are useful for modeling relationships and interactions between entities, such as analyzing social networks or complex systems within occupational environments.
Gated Recurrent Units (GRUs)
GRUs are a type of recurrent neural network (RNN) architecture that efficiently handles sequential data by maintaining information over time. They are particularly useful for tasks that require understanding the temporal dynamics of data, such as monitoring long-term safety trends.
Gradient Descent
Gradient descent is an optimization algorithm used to minimize the error in machine learning models by iteratively adjusting the model's parameters. It plays a crucial role in training AI models, ensuring that they learn effectively from data.
Hyperparameters
Hyperpar
The content presented here, including AI-generated material, is intended solely for educational purposes as part of ACOEM's ongoing efforts to explore emerging technologies. Artificial Intelligence (AI) is a rapidly evolving field, and this session aims to foster informed discussion and engagement. The views and content shared are not intended to advocate or promote any specific practices and should not be interpreted as professional advice. Dr. Kiani and Dr. Berenji have designed this presentation to encourage thoughtful reflection on AI's potential in occupational and environmental health — noting potential applications.The materials shared on this website are intended for individual enrichment and exploration of this emerging field, rather than as a guide or endorsement for implementing AI in practice. The content is not meant as a directive or endorsement of any kind. ACOEM and the authors assume no liability for any use or misuse of the information presented, including any consequences arising from the application of the materials in occupational health settings.