|
|
|
|
|
|
|
Resources |
|
|
Author: Staff Not a day goes by anymore but what we are confronted with more information about AI. But there is more to implementing AI in public health than the media would lead us to believe. Many of us are familiar with several AI tools but few of us have studied in depth the vast majority of the tools that are currently available. And other than AI experts, even fewer have researched the latest models these tools have released. Recently the EpiMonitor started looking at two projects that could be finished in 1/50th the time if we used an AI tool. But which tool would be best for these projects? Could the same tool be used for both projects? Did we need the latest release of the chosen tool(s)? How much computer horsepower did we need to drive the tool? All of us get the mailers touting certificate programs for courses teaching us how to write effective prompts but current basic information that answers the questions that the EpiMonitor faced hasn't been readily available from a single site until now. Any source that was touting themselves as best source for information was often 60-90 days out of date which, in the world of AI, is too old to be of real value. That led us to the question about what was available to assist public health professionals. Surely they were asking similar questions to those we were facing although their answers were going to be different to the ones we were getting because the tasks were totally different. So we began a search to determine what resources were available to assist the public health community in their own efforts. Our search led us to a phenomenal text written (and constantly updated) by Dr. Bryan Tegomoh. This book is one of four he has written on the topic of AI and it is available free. Best of all, The Public Health AI Handbook is licensed under Creative Commons so you can freely use and share the material. It is important to note that this book is written specifically for public health professionals. Every chapter is written with three specific questions in mind: ► Which tools actually perform well in real world settings? ► How do you evaluate a tool when your data is messy, delayed, and/or incomplete? ► What happens when the algorithm is wrong and public health action follows? For everything this book provides, it is meant to be a resource and a training tool - not official public health guidance. Your effective use of the information is context specific and you need to follow all normal safeguards when using the information presented. Among the specific topics included in this book as of today are: ► The Data Problem in Public Health AI ► AI in Disease Surveillance and Outbreak Detection ► Epidemic Forecasting with AI ► AI in Genomic Surveillance and Pathogen Analysis ► Geospatial AI and Spatial Epidemiology ► AI in Diagnostic and Clinical Decision Support ► AI for Substance Use and Overdose Prevention ► Evaluating AI Systems for Healthcare ► Performance Metrics for Public Health AI ► Validation, Equity, and Security Testing ► Explainability for Public Health AI ► Ethics, Bias, and Equity in Healthcare AI ► Privacy, Security, and Governance for Health AI ► Deployment and Production Monitoring ► System Integration and Regulatory Compliance ► Public Health AI Deployment Exercises We could go own but clearly the best way for you to understand the value of this text is to encourage you to evaluate it yourself. Please remember that the author is committed to constantly updating the material. So the version you see online today is sure to change in the blink of an eye. New tools, updated models, fresh government regulations, and new hardware requirements are available every week. To read or download a PDF copy of the book for free: https://publichealthaihandbook.com/ Other handbooks of interest by this author: More about the author: https://www.linkedin.com/in/bryan-tegomoh-md-mph/ ■
|
|
|
|
|
