Rowe M1
1University of the Western Cape, Department of Physiotherapy, Cape Town, South Africa
Background: Machine learning algorithms are enhancing clinical decision-making through the statistical analysis of very large data sets that are too complex for human beings to interpret on their own. Important applications of machine learning (ML) in healthcare include clinical decision support, diagnosis and prediction, patient monitoring and coaching, surgical assistance, patient care, and systems management. As the digital information we interact with in healthcare is increasingly filtered, shaped and analysed by algorithms we see that there are clinical and ethical implications for clinicians and patients.
Purpose: There is increasing evidence that ML and artificial intelligence will see significant vertical and horizontal integration across the health sector, with profound impacts on practice and training. This presentation aims to provide an introduction to the topic of machine learning and artificial intelligence in the context of healthcare and physiotherapy practice.
Methods: A review of the literature was conducted between 2015-2018 in order to identify the ways in which ML algorithms are being used across the health sector. Because this is a relatively new area of research, the search was not limited to physiotherapy but included all health professions, and also made use of a wide variety of keywords that are often conflated in the mainstream media e.g. AI, artificial intelligence, computer vision, expert systems, machine/deep learning, neural networks, robotics. While these keywords represent separate fields of AI research they are all driven primarily by advances in the sub-domain of machine learning.
Results: Disruptions in clinical practice as a result of AI-based diagnosis, prediction and reasoning will require that healthcare professionals reevaluate their fitness for purpose in an intelligence age that is characterised by altered relationships between therapists, patients, data, and algorithms. Human connection will be the key to success while at the same time we must use advanced technology - including AI-based systems and machine learning - to enhance our capacity to care for each other, to learn effectively over the course of our lives, and to develop creative solutions for the problems that matter to us.
Conclusion(s): Machine learning algorithms are already "smarter" than us within certain narrow domains of clinical practice and will increasingly take over some of the cognitive and physical tasks that were previously the sole domain of human beings. Successful clinical practice in the 21st century will require that we understand how to analyse and interpret the decisions of ML algorithms and for this it is essential that clinicians are involved in the development, implementation and evaluation of these systems in clinical practice.
Implications: Unless health professionals are actively engaged in a conversation around ML and artificial intelligence in clinical practice, we run the risk that our clinical decision-making will be subject to machine intelligence, rather than being informed by it. The challenge we face at the beginning of the 21st century is to bring together computers and humans in ways that enhance human well-being, augment human ability and expand human capacity.
Keywords: artificial intelligence, clinical practice, machine learning
Funding acknowledgements: Nothing to declare.
Purpose: There is increasing evidence that ML and artificial intelligence will see significant vertical and horizontal integration across the health sector, with profound impacts on practice and training. This presentation aims to provide an introduction to the topic of machine learning and artificial intelligence in the context of healthcare and physiotherapy practice.
Methods: A review of the literature was conducted between 2015-2018 in order to identify the ways in which ML algorithms are being used across the health sector. Because this is a relatively new area of research, the search was not limited to physiotherapy but included all health professions, and also made use of a wide variety of keywords that are often conflated in the mainstream media e.g. AI, artificial intelligence, computer vision, expert systems, machine/deep learning, neural networks, robotics. While these keywords represent separate fields of AI research they are all driven primarily by advances in the sub-domain of machine learning.
Results: Disruptions in clinical practice as a result of AI-based diagnosis, prediction and reasoning will require that healthcare professionals reevaluate their fitness for purpose in an intelligence age that is characterised by altered relationships between therapists, patients, data, and algorithms. Human connection will be the key to success while at the same time we must use advanced technology - including AI-based systems and machine learning - to enhance our capacity to care for each other, to learn effectively over the course of our lives, and to develop creative solutions for the problems that matter to us.
Conclusion(s): Machine learning algorithms are already "smarter" than us within certain narrow domains of clinical practice and will increasingly take over some of the cognitive and physical tasks that were previously the sole domain of human beings. Successful clinical practice in the 21st century will require that we understand how to analyse and interpret the decisions of ML algorithms and for this it is essential that clinicians are involved in the development, implementation and evaluation of these systems in clinical practice.
Implications: Unless health professionals are actively engaged in a conversation around ML and artificial intelligence in clinical practice, we run the risk that our clinical decision-making will be subject to machine intelligence, rather than being informed by it. The challenge we face at the beginning of the 21st century is to bring together computers and humans in ways that enhance human well-being, augment human ability and expand human capacity.
Keywords: artificial intelligence, clinical practice, machine learning
Funding acknowledgements: Nothing to declare.
Topic: Robotics & technology; Education: continuing professional development; Information management, technology & big data
Ethics approval required: No
Institution: University of the Western Cape
Ethics committee: Community and Health Science Research Committee
Reason not required: This is a position paper based on a review of the literature.
All authors, affiliations and abstracts have been published as submitted.
