Artificial intelligence (AI) technologies are poised to disrupt conventional modes of clinical assessment to yield innovative solutions of wide appeal. However, such technologies are sometimes uninformed by user-centered design principles, which iteratively solicit feedback from project stakeholders on whether: 1) such a technology is relevant/helpful for the clinical practice (i.e., actually fills gap in clinical workflow), 2) circumvents issues pertaining to rater disagreement for subjectively assessed outcomes, and 3) is adoptable/usable by the end-user. As such, we are learning to calibrate new-age tools with conventional, tried and true practices, where the rush to quantitate must be balanced with respect for the viewpoints of those who assume the risky practice of clinical decision making.
The goal of this Research Topic is to bring together a collection of papers that discuss pragmatic means for developing and assessing the viability of AI technologies that are informed “by clinicians, for clinicians” (i.e., in close collaboration between teams of biomedical researchers and practicing clinicians). We hope for this research topic to provide valuable input on algorithmic design principles, assessment methods, and infrastructural needs which are in optimal alignment with stakeholders and thus have the greatest potential for clinical impact.
We are currently inviting manuscripts to Frontiers Medical Technology on the subject of machine learning applications to medical science, including but not limited to, the following topics:
• State of the art modeling techniques with a particular emphasis on assessment of quantitative endpoints (e.g., NGS) and multimodal modeling techniques.
• Evaluation of stakeholder and infrastructural alignment to developed technologies, including:
o Approaches for designing, acquiring and implementing computational methods for machine learning in complex regulatory environments.
o Exploring experimental design to communicate perceived adoptability of an algorithm (e.g., survey via Likert scales, cost effectiveness, stakeholder discussion), and
o Methods to account for inaccurate reporting due to expert consensus and interrater disagreement and for proper communication of model performance (i.e., not overstating model effectiveness) which takes into account this variation.
o Efforts to design and implement educational curriculum for AI technologies.
o Reports of large-scale intra- and inter-institutional validation studies of recently developed and potentially established technologies.
• Demonstration of real-world applications in low resource regions of the world which highlight the promise of machine learning as a knowledge democratization platform.
Artificial intelligence (AI) technologies are poised to disrupt conventional modes of clinical assessment to yield innovative solutions of wide appeal. However, such technologies are sometimes uninformed by user-centered design principles, which iteratively solicit feedback from project stakeholders on whether: 1) such a technology is relevant/helpful for the clinical practice (i.e., actually fills gap in clinical workflow), 2) circumvents issues pertaining to rater disagreement for subjectively assessed outcomes, and 3) is adoptable/usable by the end-user. As such, we are learning to calibrate new-age tools with conventional, tried and true practices, where the rush to quantitate must be balanced with respect for the viewpoints of those who assume the risky practice of clinical decision making.
The goal of this Research Topic is to bring together a collection of papers that discuss pragmatic means for developing and assessing the viability of AI technologies that are informed “by clinicians, for clinicians” (i.e., in close collaboration between teams of biomedical researchers and practicing clinicians). We hope for this research topic to provide valuable input on algorithmic design principles, assessment methods, and infrastructural needs which are in optimal alignment with stakeholders and thus have the greatest potential for clinical impact.
We are currently inviting manuscripts to Frontiers Medical Technology on the subject of machine learning applications to medical science, including but not limited to, the following topics:
• State of the art modeling techniques with a particular emphasis on assessment of quantitative endpoints (e.g., NGS) and multimodal modeling techniques.
• Evaluation of stakeholder and infrastructural alignment to developed technologies, including:
o Approaches for designing, acquiring and implementing computational methods for machine learning in complex regulatory environments.
o Exploring experimental design to communicate perceived adoptability of an algorithm (e.g., survey via Likert scales, cost effectiveness, stakeholder discussion), and
o Methods to account for inaccurate reporting due to expert consensus and interrater disagreement and for proper communication of model performance (i.e., not overstating model effectiveness) which takes into account this variation.
o Efforts to design and implement educational curriculum for AI technologies.
o Reports of large-scale intra- and inter-institutional validation studies of recently developed and potentially established technologies.
• Demonstration of real-world applications in low resource regions of the world which highlight the promise of machine learning as a knowledge democratization platform.