This project studies whether Swedish digital primary healthcare is unequal and whether patients are treated differently depending on their ethnic background, socioeconomic status, gender, and age. The empirical setting is 1177 Direkt, a digital primary care service where patients provide standardized information about their symptoms and health status before interacting with healthcare professionals online. The project uses detailed data from these healthcare encounters, including the written communication between patients and healthcare professionals and other information about the visits, combined with linked administrative register data on patients and healthcare staff. The project uses natural language processing and supervised machine-learning methods to classify patient-centred communication in healthcare professionals’ chat messages. A subset of the text data will be manually coded and used to train and validate a model that can then be applied to the full set of healthcare encounters. Machine learning will also be used to classify information in healthcare professionals’ summary notes. Because the text data may contain sensitive health information, these analyses need to be conducted within a secure computing environment. The empirical analysis will distinguish raw differences in treatment between patient groups from differences that can be explained by patients’ own attempts to influence their care and by differences in medical conditions at the start of the encounter. By controlling for the standardized health information available to healthcare professionals and for patients’ stated expectations, the project aims to identify whether remaining differences in patient-centred communication and treatment intensity are consistent with discriminatory treatment. The broader goal is to improve our understanding of the mechanisms behind inequality in Swedish primary healthcare.