-
The use of artificial intelligence in healthcare raises ethical concerns, particularly around bias toward protected characteristics such as gender, race, and ethnicity. Gender bias remains common, as unconscious beliefs held by patients, clinicians, researchers, and administrators can shape care delivery and influence health outcomes. This study examined 309 lung cancer patient records from a Kaggle dataset to classify lung cancer status and assess potential gender bias in support vector machine (SVM) classification. The gender-aware SVM model achieved accuracy, precision, and recall scores of 90%, 94%, and 94%, respectively, while the gender-unaware model produced slightly lower values of 88%, 91%, and 96%. The gender-aware model performed slightly better than the gender-unaware model, suggesting that gender may be associated with the model’s predictions. The study further evaluated subgroup performance to assess whether predictive behaviour differed between male and female patients. The analysis revealed substantial performance differences across gender groups. For males, the accuracy, precision, recall, and F1 Scores were 91%, 93%, 97%, and 95%, respectively. For females, the corresponding scores were slightly lower at 89%, 96%, 92%, and 94%. These disparities suggest that the model performs better for male patients than for female patients. Such imbalances risk highlighting existing gender inequities in healthcare. Historically, women’s symptoms have often been underestimated or dismissed, contributing to poorer health outcomes. The model’s bias highlights the need for more equitable AI systems that do not preserve existing stereotypes or disadvantage already marginalised groups.