Ethics of AI in Pathology

Abstract

Deep learning has rapidly advanced artificial intelligence (AI) and algorithmic decision-making (ADM) paradigms, affecting many traditional fields of medicine. Pathology is a heavily data-centric specialty of medicine. The structured nature of pathology data repositories makes it highly attractive to AI researchers to train deep learning models to improve health care delivery. Equally, there are enormous financial incentives driving adoption of AI and ADM due to promise of increased efficiency of the health care delivery process. Unethical use of AI may exacerbate existing inequities of health care, especially if not implemented correctly. There is an urgent need to harness the vast power of AI in an ethically and morally justifiable manner. In this mini-review, we explore the key issues involving AI ethics in pathology. Issues related to ethical design of pathology AI studies and the potential risks associated with implementation of AI and ADM within the pathology workflow are discussed. Three key foundational principles of ethical AI are described in the context of pathology: transparency, accountability, and governance. The future practice of pathology must be guided by these principles. Pathologists should be aware of the potential of AI to deliver superlative health care and the ethical pitfalls associated with it. Finally, pathologists must have a seat at the table to drive the future implementation of ethical AI in the practice of pathology.

Computational pathology in 2030: a Delphi study forecasting the role of AI in pathology within the next decade

“…AI is expected to have a deep impact on pathology, and our study provides detailed insight into the current challenges and expectations surrounding its role in pathology, including timely and relevant information regarding how pathology care might be delivered in the future, assuming all regulatory and ethical questions are addressed. 16 , 17 , 32 While we expect that our findings will be of great interest to a wide variety of stakeholders, we also hope that the preceding limitations will be sufficiently addressed in forthcoming studies, with our survey and its freely available data collection forms serving as a model for independent validation and extension.…”

“…The lack of consensus regarding these is expected to be resolved as more AI tools are evaluated in prospective clinical settings and more consideration is directed toward ensuring that tools are integrated into workflows in ways that maximize safety, efficiency, and positive patient outcomes. 5 , 6 , 17 , 31 …”

“… 16 Simultaneously, ethical concerns have been raised regarding potential data privacy breaches, systemic algorithmic bias, harm related to erroneous AI-generated outputs, and exacerbation of healthcare disparities. 17 Along with hurdles related to regulatory approval and reimbursement for AI products, these have contributed to a significant AI "translation gap" in pathology, 15 which we define as the failure to prospectively validate and successfully integrate AI models into real-world clinical workflows. Although various opportunities and challenges surrounding AI in pathology have been extensively discussed in the literature, to date, there has been no systematic survey regarding this topic from the short-to-medium term perspective of digital and CPath experts.…”

Cell projection plots: a novel visualization of bone marrow aspirate cytology

“…Therefore, such implementation of AI is more explainable as opposed to an AI model making a diagnostic prediction per se. Additionally, one of the common concerns among pathologists is the possibility of their jobs being replaced by AI [28]. CPPs are non-threatening to pathologists as they keep them involved in the model's decision-making process, potentially acting as a clinical decision-support tool.…”

Artificial Intelligence in Pathology

“…The Mini-Reviews and the Review that comprise this special theme issue are designed to focus on these various aspects of AI in pathology, beginning with an overview of the practical challenges facing implementation, 2 and including a discussion of ethical considerations that arise in this field. 3 The evolution of AI in pathology is illuminated by earlier but analogous trends in radiology. 4 Analogies with the use of AI in reduction of interobserver variability are considered, 5 as are current strategies for reducing the need for massive annotation in machine learning through either the use of existing supervised frameworks or by exploiting hybrid models using unsupervised learning, generative models, and/or synthetic data.…”