Our paper is currently under peer-review at the Journal of Ethics in Scientific and Technological Innovations.

Abstract:

AI’s implementation in the healthcare field has vastly grown over the past years, providing a generalized solution to previously patient-specific treatment. It has been exceedingly difficult to diagnose and treat neurodegenerative diseases due to the heterogeneity of the patients. Recently, clinicians and scientists have turned towards machine learning, a subfield of artificial intelligence (AI), to improve detection mechanisms for the early stages of neurodegenerative diseases, as well as to curate effective treatment plans that account for the high-dimensionality and multi-modal nature of the data. Despite the potential of AI designed treatment, multiple limitations, extending from medical ethics, legal responsibility, morality, and privacy remain unaddressed.

This paper elucidates whether ML models will be able to succeed human clinicians in diagnostics and treatment planning. We consider the lack of AI’s ability for humanistic and holistic consideration and the vagueness of specific protocols that outline ethical AI’s implementation, and its potentiality of  causing intervention-generated inequalities stemming from limited health literacy and accessibility, and corporate “targeted” approaches. We also explore potential privacy concerns that arise with the use of ML in the medical field, including data poisoning and original training data recreation, as well as propose potential methods to address these concerns and find a balance between accuracy and privacy. Rising concerns also originate from AI’s capacity to tarnish the patient-provider relationship, which influences frameworks for liability as well as the hazards of excessive patient autonomy in medical decisions.

The future place of AI in the neuro-healthcare field will be pivotal for healthcare development, but it must be implemented with such precision that promotes equality, protects privacy and transparency, and maintains salubrious patient-provider relationship.