Название: AI/ML for Healthcare: Navigating the AI/ML Maze Responsibly, Securely, and SustainablyАвтор: Kapila Monga
Издательство: CRC Press
Год: 2025
Страниц: 296
Язык: английский
Формат: epub (true)
Размер: 10.1 MB
The advent of Generative AI has democratized access to AI, prompting nearly everyone in healthcare organizations - from frontline workers to business leaders – to ask pressing questions: How can I be better equipped to support AI adoption meaningfully? How do I ensure I ask the right questions? What cautions should I exercise as I think about AI/Machine Learning (ML) in my business process? This book aims to answer these and other such questions and to empower healthcare professionals, at all levels, by providing them knowledge across various aspects of AI/ML, enabling them (at least in part) to realize positive, lasting business value from AI and ML initiatives. This book draws upon my experience of working in healthcare AI/ML, lessons I learned while observing leaders in this space trying to make a difference, and research (for evolving topics like sustainable AI development and securing AI/ML systems).
This book provides readers with actionable insights to build responsible, secure, and sustainable AI/ML solutions in healthcare and delves into key principles for scaling AI/ML value delivery, including establishing Machine Learning Operations (MLOps) processes and launching citizen Data Science programs. At its heart, this book features a healthcare-specific case study that bridges the gap between theoretical knowledge and practical application, illustrating all major concepts in a real-world context.
Machine Learning (ML) is a subset of AI that empowers computers/machines to learn without being explicitly programmed. ML algorithms learn patterns from large amounts of historical data (sometimes referred to as training data), and then use the knowledge gained to generalize the patterns to unseen data sets, and answer questions about unseen data sets, that is, without being explicitly programmed for those situations. ML algorithms are generally used for either of the two purposes: classification and prediction/regression. Data scientists use ML as a tool (one of the tools) to extract insights from the data.
Deep Learning is a subset of ML. Deep Learning algorithms use multiple layers to progressively extract higher-level features from raw input. Deep Learning can be used for both supervised and unsupervised learning and is especially useful in processing large volumes of text and image data. Neural networks like deep neural networks, recurrent neural networks, convolutional neural networks, long short-term memory networks, and generative adversarial networks are all examples of Deep Learning algorithms. The inherent design of Deep Learning algorithms makes them “black box”, that is, they lack explainability. This limitation of Deep Learning prevents them from being used in many healthcare use-cases because of lack of transparency and eventually lack of trust. However, their high accuracy and the ability to handle large volumes of text and image data make them good candidates for tasks like assisting radiologists in reading CT scans, X-rays, etc. Convolutional neural networks are especially good for image classification tasks. Recurrent neural networks work well on sequential data to generate insights like text data. They are particularly good in language generation, and processing. Generative adversarial networks are well suited to generate synthetic data including text, voice, and images.
The final chapter of this book offers a forward-looking commentary on the future of healthcare AI/ML. It explores the potential of Generative AI for healthcare and advocates leveraging lessons from past AI/ML implementations to chart a meaningful path for embracing Generative AI. Additionally, this book emphasizes the importance of adopting “reciprocal altruism” to accelerate AI/ML value realization across the healthcare industry and provides practical recommendations toward the same.
Скачать AI/ML for Healthcare: Navigating the AI/ML Maze Responsibly, Securely, and Sustainably