TABLE 1.

Potential Reasons Why Artificial Intelligence Has Not Been Embraced by the Critical Care Community

Concern	Rebuttal	Potential Solution
Patient factors
Lack of awareness by patients and families	Newer technology in healthcare without significant lay exposure	Public education and media explanation of AI in healthcare
Reluctance to have AI in care	Noninvasive and meant to augment clinical care, not replace physicians	Explanation of use of algorithm and potential benefits during clinical care
Privacy concerns	Newer systems use HIPAA-compliant cloud computing resources	Emphasis on HIPAA compliance approaches in clinical use
Clinician factors
Lack of awareness by clinicians	Minimal teaching in this area during medical education	Improved medical education in this area, engagement of clinicians in implementation
Mistrust of AI approaches	Older algorithms lack sophistication, abilities and performance of newer deep learning algorithms	More research and education demonstrating benefit in clinical care.
Concerns about medicolegal aspects using AI	Field is young without clear precedent	U.S. Food and Drug Administration and other regulatory approval; clear “intended use” for AI algorithms
Lack of definitive multicenter randomized trials	Data are evolving; field is young and dynamic	Federal funding agencies should support grant funding on mature algorithms
Technology factors
Suboptimal predictive abilities of AI algorithms	Powered by big data and multimodal data, these systems are rapidly improving	Newer deep learning algorithms, advances to augment data availability (e.g., biopatches, data by smart laboratories)
Lack of infrastructure for real-time predictive scores	Newer cloud computing and interoperability technologies are lowering the infrastructure barriers	Incorporating cloud computing, healthcare interoperability standards, software engineering, and hospital information technology education into clinical AI curriculum
Systems factors
Poor implementation approaches	Historically, this has been overlooked in favor of algorithm development	Implementation science and multidisciplinary teams improve use of algorithms
Lack of administrative support to properly implement AI algorithms into clinical use	Although there are upfront costs, potential benefits likely outweigh this	Studies demonstrating improved outcomes and cost-effectiveness; emphasis on interoperability standards
Misalignment of patient care, quality improvement, and financial incentives	Value-based care and the changing landscape of digital health reimbursement	Closer collaboration of AI experts, hospital quality improvement, value-based care, and finance teams

AI = artificial intelligence, HIPPA = Health Insurance Portability and Accountability Act.