Quantifying dimensionless features with machine learning algorithms: the case of aesthetic outcomes in breast cancer surgery

ML algorithms have the potential to create new sources of information that are increasingly used for making decisions or evaluating outcomes. Its uniqueness is to extract features invisible to human eyes and those that are difficult to quantify. However, little is known about how domain experts and algorithms interact when making decisions or evaluations. With an exploratory case study, we investigate the role of ML in the evaluation of aesthetic outcomes in breast cancer surgery and how a multidisciplinary team is building an evaluation equation to support domain experts. Gioia methodology guides the collection and analysis of semi-structured interviews and archival data. We contribute to the literature about the emergence of new human-ML collaborations, managing algorithmic technologies and outcomes evaluation. For practitioners we provide strategic implications to best combine domain expertise with ML-enabled computational rationality.