The impact of intersectional racial and gender biases on minority female leadership over two centuries.

This study scrutinizes the enduring effects of racial and gender biases that contribute to the consistent underrepresentation of minority women in leadership roles within American private, public, and third sector organizations. We adopt a behavioural data science approach, merging psychological schema theory with sociological intersectionality theory, to evaluate the enduring implications of these biases on female leadership development using mixed methods including machine learning and econometric analysis. Our examination is concentrated on Black female leaders, employing an extensive analysis of leadership rhetoric data spanning 200 years across the aforementioned sectors. We shed light on the continued scarcity of minority female representation in leadership roles, highlighting the role of intersectionality dynamics. Despite Black female leaders frequently embracing higher risks to counter intersectional invisibility compared to their White counterparts, their aspirations are not realized and problems not solved generation after generation, forcing Black female leaders to concentrate on the same issues for dozens and, sometimes, hundreds of years. Our findings suggest that the compound influence of racial and gender biases hinders the advancement of minority female leadership by perpetuating stereotypical behavioral schemas, leading to persistent discriminatory outcomes. We argue for the necessity of organizations to initiate a cultural transformation that fosters positive experiences for future generations of female leaders, recommending a shift in focus from improving outcomes for specific groups to creating an inclusive leadership culture.

Operationalising fairness in medical AI adoption: detection of early Alzheimer's disease with 2D CNN.

OBJECTIVES: To operationalise fairness in the adoption of medical artificial intelligence (AI) algorithms in terms of access to computational resources, the proposed approach is based on a two-dimensional (2D) convolutional neural networks (CNN), which provides a faster, cheaper and accurate-enough detection of early Alzheimer's disease (AD) and mild cognitive impairment (MCI), without the need for use of large training data sets or costly high-performance computing (HPC) infrastructures. METHODS: The standardised Alzheimer's Disease Neuroimaging Initiative (ADNI) data sets are used for the proposed model, with additional skull stripping, using the Brain Extraction Tool V.2approach. The 2D CNN architecture is based on LeNet-5, the Leaky Rectified Linear Unit activation function and a Sigmoid function were used, and batch normalisation was added after every convolutional layer to stabilise the learning process. The model was optimised by manually tuning all its hyperparameters. RESULTS: The model was evaluated in terms of accuracy, recall, precision and f1-score. The results demonstrate that the model predicted MCI with an accuracy of 0.735, passing the random guessing baseline of 0.521 and predicted AD with an accuracy of 0.837, passing the random guessing baseline of 0.536. DISCUSSION: The proposed approach can assist clinicians in the early diagnosis of AD and MCI, with high-enough accuracy, based on relatively smaller data sets, and without the need of HPC infrastructures. Such an approach can alleviate disparities and operationalise fairness in the adoption of medical algorithms. CONCLUSION: Medical AI algorithms should not be focused solely on accuracy but should also be evaluated with respect to how they might impact disparities and operationalise fairness in their adoption.

Digital Resilience Through Training Protocols: Learning To Identify Fake News On Social Media.

We explore whether training protocols can enhance the ability of social media users to detect fake news, by conducting an online experiment (N = 417) to analyse the effect of such a training protocol, while considering the role of scepticism, age, and level of education. Our findings show a significant relationship between the training protocol and the ability of social media users to detect fake news, suggesting that the protocol can play a positive role in training social media users to recognize fake news. Moreover, we find a direct positive relationship between age and level of education on the one hand and ability to detect fake news on the other, which has implications for future research. We demonstrate the potential of training protocols in countering the effects of fake news, as a scalable solution that empowers users and addresses concerns about the time-consuming nature of fact-checking.