Artificial intelligence (AI) is becoming an important tool in healthcare, helping doctors diagnose diseases and plan treatments faster. This study explores how artificial intelligence (AI), a promising tool in healthcare, can sometimes show biases that may lead to unequal treatment outcomes for lung cancer patients. AI is widely used in diagnosing and treating lung cancer, often improving care. However, the research finds that the quality of data used to train AI models such as patient demographics (e.g., age, race, gender) and disease types can introduce errors or biases. For example, the AI system might perform better for specific age groups or more common types of lung cancer, while being less accurate for others.
To address these issues, the researchers created a synthetic patient database and examined how bias affected AI recommendations for treatment. They observed disparities based on age and cancer type, which could worsen health inequities if uncorrected. The study concludes with practical strategies, such as using diverse datasets and ethical oversight, to reduce bias and improve the fairness and accuracy of AI in lung cancer treatment.
Full text: Kai Ding, Shelby Forbes, Fangfang Ma, Ganxi Luo, Jiayou Zhou, Yian Qi, AI bias in lung cancer radiotherapy, Exploration of Digital Health Technologies, 2024, Volume 2, 302–312, https://doi.org/10.37349/edht.2024.00030