Human-Centered Design to Address Biases in Artificial Intelligence

Abstract

The potential of artificial intelligence (AI) to reduce health care disparities and inequities is recognized, but it can also exacerbate these issues if not implemented in an equitable manner. This perspective identifies potential biases in each stage of the AI life cycle, including data collection, annotation, machine learning model development, evaluation, deployment, operationalization, monitoring, and feedback integration. To mitigate these biases, we suggest involving a diverse group of stakeholders, using human-centered AI principles. Human-centered AI can help ensure that AI systems are designed and used in a way that benefits patients and society, which can reduce health disparities and inequities. By recognizing and addressing biases at each stage of the AI life cycle, AI can achieve its potential in health care.

Accelerating health disparities research with artificial intelligence

“…Mechanisms must be developed for ongoing evaluation and improvement of AI models. Protocols for regularly auditing and monitoring AI systems must be established to ensure they remain unbiased, transparent, and aligned to reduce health disparities ( 8 , 20 , 31 , 34 , 41 – 43 ).…”

Advancing Diabetic Retinopathy Diagnosis: Leveraging Optical Coherence Tomography Imaging with Convolutional Neural Networks

“…Emerging techniques offer potential solutions to overcome these challenges. Transfer learning, for instance, allows the knowledge gained from pretraining on large-scale datasets to be transferred and fine-tuned on specific DR datasets, thereby improving the performance of CNN models with limited data [ 34 , 35 ]. By leveraging the pre-learned features, transfer learning enables more efficient training and better generalization.…”

A systematic scoping review of studies describing human factors, human-centered design, and usability of sensor-based digital health technologies

“…For example, the physical size of an sDHT may limit its deployment in children; those with limited dexterity may not be able to manipulate a wearable appropriately; and those with poor vision may be limited in their ability to read information presented on a screen[16,17], highlighting the importance of human-centered design which prioritizes the needs, capabilities, and behaviors of users during the design process [18]. Inadequate attention to human-centered design and usability testing approaches can hinder the evaluation of healthcare interventions, contribute to insuficient adoption, perpetuate health disparities, increase costs, and potentially introduce safety risks [19–22]. Thus, integrating human factors considerations in the design, development, and evaluation of sDHTs is critical to improve their likelihood of being adopted and properly utilized in clinical research and healthcare in a way that is safe, efective, inclusive, and optimizes the user experience.…”