Human factors approaches to incorporating AI into healthcare
TimeTuesday, April 1312:30pm - 1:30pm EDT
LocationPatient Safety Research and Initiatives
AI in healthcare needs human factors involvement. In the complex healthcare work environment, the vision to unleash the full potential of Human-AI Collaboration (HAIC) can only be achieved by advancing and integrating AI with human-centered principles and practices, resulting in HAIC systems in which humans and AI together achieve superior performance than either could alone. This remains a challenge in healthcare that the panelists will speak to for a number of reasons:
(i) It is one of the largest and most complex sectors of the US and global economy;
(ii) There are numerous challenges, including access, quality, safety, cost – to assuring individual and population health in the coming decades;
(iii) AI has real potential to transform the $2.4 trillion healthcare industry with use cases ranging from personalized medicine, enhancement of team-based care models, to improvements in health at the community level; and
(iv) Incorporation of AI into healthcare remains complicated both for the human and for the AI parts of the system.
Healthcare, a complex adaptive system, works best when the relative strengths of humans (including context sensitivity and situation specificity) are properly integrated with the information processing power of computerized systems (1). Ineffective design and implementation are key barriers to systems and technologies from achieving their promise of improving work. A common mistake occurs when “fit” with humans in the system and intended environmental context are inadequately considered. Failure to make the user, whether an elderly patient or a cardiac surgeon, the focus of AI design efforts will result in undesirable outcomes including under-utilization, workarounds, maladaptive responses, and unintended consequences.
The panel convenes experts with experience studying and implementing AI-based systems into healthcare and includes researchers and industry expertise. Each panelist will describe: (a) their experience with AI-based systems in healthcare including their methods, strategies, and results; (b) associated benefits they have observed; (c) barriers to use and implementation that must be addressed; and (d) recommendations for human factors professionals interested in this space. we will facilitate an interactive discussion between panelists and the audience. As part of the session we will also conduct informal surveys and live voting to assess audience experiences with and needs related to AI-based systems in healthcare. Biographies follow and panelist statements follow.
Shilo Anders (chair), PhD, Associate Professor, Center for Research and Innovation in Systems Safety, Vanderbilt University Medical Center. Dr. Anders has been involved in numerous federally funded and commercial projects to create new technologies in healthcare including information technologies that contain decision support and AI. She has with both professional and academic experience with systems analysis and design, software development processes, and design of clinical decision support (CDS) systems. She serves as a co-investigator on numerous projects, including a current AHRQ R01 that is developing machine learning-based predictive algorithms from Fitbit active minute and geolocation data submitted by cancer patients. She has led numerous human factors validation studies of medical devices and led the development of standards for patient identifiers when implementing EPIC for the first time at Vanderbilt University Medical Center.
Yuval Bitan, PhD, Assistant Professor, Ben-Gurion University of the Negev, Department of Health Systems Management, Guilford Glazer Faculty of Business and Management and Faculty of Health Sciences, and leads the Human Systems Integration in Healthcare laboratory. Dr. Bitan is also an Assistant Professor (status only) at the Department of Mechanical & Industrial Engineering at the University of Toronto. Dr. Bitan received his PhD in 2003 in Industrial Engineering and Management. He was a Research Associate at the Cognitive Technologies Laboratory (University of Chicago, Illinois), HumanEra (University Health Network, Toronto, Canada) and at the University of Toronto. He is a member of the Human Factors and Ergonomics society. He has extensively published and provided invited lectures on how to engender appropriate trust with alarms and medical devices that include ‘smart’ AI technologies such as ‘smart alarms’, predictive algorithms, and machine learning models.
Emily S. Patterson, PhD, Associate Professor, The Ohio State University, College of Medicine, School of Health and Rehabilitation Sciences, Division of Health Information Management and Systems. Dr. Patterson received her PhD in industrial and systems engineering in 1999, specializing in human factors engineering. She was a co-author of several influential NIST technical reports advancing the usability of health information technology, including the summative usability testing standard for electronic health records, NIST 7804, which was published cooperatively with the FDA. She has served as the principal investigator on federally funded grants and contracts for NIST, AHRQ, the VA, the military, and the National Patient Safety Foundation. She has served as an advisor to The Joint Commission, the National Board of Medical Examiners, the National Institute of Standards and Technology, and ECRI. She serves as Associate Editor for healthcare in the Human Factors Journal and has extensively published (h-index 41 on Google Scholar). She has led numerous formative usability tests at the VA, which led to software changes to enhance usability and patient safety. She is the certificate director and lead educator for an OSU four-class undergraduate and graduate certificate “Usability and User Experience in Healthcare.”
Anne Miller, PhD, Lead Human Factors Researcher at Cerner, and Adjunct Associate Professor at Vanderbilt University Medical Center. Dr Miller’s research focuses on the role of information technology in supporting resilience in clinical decision making, clinical communication and care coordination in complex clinical care contexts. Within Cerner, she leads Human Factors research in Clinical and Intelligent systems design and development. A core goal is AI-HI integration in inpatient and outpatient Clinical environments. Her experience convers the integration of voice technology, and machine learning algorithm integration including the representation and presentation of risk assessments and diagnostic augmentation. Some of the challenges that Dr Miller addresses include the development and implementation of appropriate research methods in solution development environments.
Narges Razavian, PhD, Assistant Professor Department of Population Health and Radiology, conducting research in the Center for Healthcare Innovation and Delivery Science (CHIDS) NYU Grossman School of Medicine. She is also member of NYU Langone Predictive Analytics unit and an affiliate faculty at NYU Center for Data Science. Dr. Razavian’s research is focused on the intersection of machine learning, artificial intelligence, and medicine. In collaboration with various clinical investigators across NYU Langone, her group focuses on using Electronic Health Records (EHR) and Imaging data combined with AI, to help address healthcare research and delivery gaps. A few notable collaborations from her group include COVID-19 inpatient model development, deployment, and validation via a Randomized Clinical Trial, development and validation EHR-based AI models for prediction of disease onsets, Assessment of fairness and bias of machine learning models, MRI imaging AI models, and Histopathology AI tool developments.
1. Hollnagel E, Woods D. Joint Cognitive Systems: Foundations of Cognitive Systems Engineering. Boca Raton, FL: CRC Press; 2006.
2. Chang AC (2020) Intelligence-Based Medicine: Artificial Intelligence and Human Cognition in clinical medicine and Healthcare. London, UK: Academic Press.