Machine-learning-based classification of neuroimaging data (hereafter ML-MRI) to predict clinical diagnoses has increased substantially in the last decade. Despite the promise of ML for clinical classification and prediction, no work has been done to anticipate the ethical obligations and challenges that emerge when ML algorithms predict clinical diagnoses in pre-symptomatic individuals. Two recent reports from the Infant Brain Imaging Study (IBIS) have exemplified the ML classification approach by predicting 24-month clinical diagnosis of autism spectrum disorder (ASD) from 6-month MRI data in infants at high and low familial risk for ASD. These preliminary results demonstrated?for the first time?the feasibility of identifying infants prior to onset of ASD symptoms, and raised critical ethical questions about whether and how to disclose predictive ML- MRI classification of ASD to parents of pre-symptomatic infants. The proposed research aims to investigate parental attitudes towards predictive ML-MRI testing in IBIS, and addresses a core principle of BRAIN Initiative: considering the ethical implications of neuroscience research.
Aim 1 will review the bioethics literature on disclosure of results in genetic research, and identify overlapping and distinct ethical considerations for disclosure using ML-MRI vs. genetic testing for disease prediction.
Aim 2 will use a theory-based approach to assess the attitudes of IBIS parents towards predictive ML-MRI testing through qualitative interviews and a quantitative survey. The knowledge gained through this research will help investigators make ethical decisions about disclosure in future neuroimaging studies of high-risk infants. Our long-term goal is to develop guidelines for navigating the unique ethical challenges posed by a new frontier in neuroimaging research: clinical prediction via increasingly powerful and sophisticated ML-based analyses of large neuroimaging datasets. The proposed research and training plan will leverage the resources of the University of Washington Autism Center, the Treuman Katz Center for Pediatric Bioethics at Seattle Children?s Hospital, and the multi-site, longitudinal Infant Brain Imaging Study. The applicant will receive formal training in neuroethics, pediatric bioethics, and qualitative research methods, which will complement her prior skillset and prepare her for an independent research career investigating ethical issues in pediatric neuroscience.
Machine learning-based statistical techniques, recently applied to neuroimaging data, have allowed researchers to predict disease and disorder from brain data alone. Investigators working at this new frontier in neuroimaging research now face ethical challenges about whether and how to disclose a predictive clinical diagnosis to pre-symptomatic individuals. By combining bioethical theory with the perspectives of participants, the current research will generate new knowledge to guide ethical judgments about the disclosure of predictive diagnoses in future neuroimaging research.
