AIAI Seminar - Wednesday, 3 May 2023 - Talks by Dilara Kekulluoglu and Guillermo Romero Moreno

Speaker:               Dilara Kekulluoglu  

Title:                     "We are a startup to the core": A qualitative interview study on the security and privacy development practices in Turkish software startups

Abstract:

Security and privacy are often neglected in software development, and rarely a priority for developers. This insight is commonly based on research conducted by researchers and on developer populations living and working in the United States, Europe, and the United Kingdom. However, the production of software is global, and crucial populations in important technology hubs are not adequately studied. The software startup scene in Turkey is impactful, and comprehension, knowledge, and mitigations related to software security and privacy remain understudied. To close this research gap, we conducted a semi-structured interview study with 16 developers working in Turkish software startups. The goal of the interview study was to analyze if and how developers ensure that their software is secure and preserves user privacy. Our main finding is that developers rarely prioritize security and privacy, due to a lack of awareness, skills, and resources. We find that regulations can make a positive impact on security and privacy. Based on the study, we issue recommendations for industry, individual developers, research, educators, and regulators. Our recommendations can inform a more globalized approach to security and privacy in software development

Speaker:             Guillermo Romero Moreno

Title:                  Associations of multimorbidity in the oldest old: a novel Bayesian methodology for robust multimorbidity analysis under sparse data

Abstract:

Patients that simultaneously suffer multiple long-term health conditions pose a problem to current healthcare systems, as these are configured for individual conditions and overlook their interaction. A particularly useful approach to represent co-morbidity data is via networks, where nodes correspond to conditions and edges represent their relations. However, determining the degree of co-morbidity and deciding on their relevance is not trivial. Different measures have been used in the literature for this purpose (e.g. relative risks, phi-correlations, cosine index, etc), but they tend to be biased against certain conditions and they typically rely on assumptions that do not necessarily hold when data is scarce. Here, we adapt Bayesian network reconstruction to infer the network of relevant associations between long-term health conditions for cases with low counts, such as infrequent health conditions and minorities. This provides a statistically sound method that accounts for the noise and uncertainty in the data, prevents the reporting of spurious associations, and allows a wider understanding of the multimorbidity landscape via network analysis. We apply the method to a primary care dataset and focus on the oldest-old subpopulation, for whom data is scarce and more sensitive to misguided reportings.