Conversational Agent Automation for patients after cardiothoracic surgery based on clinical team experience and behavior change theory

Cardiovascular disease is a leading cause of death and decreased quality of life. Cardiothoracic surgery provides the chance to live longer with cardiovascular disease. Proper use of clinical protocols, regular clinical evaluation, medication review, improved modifiable risk factors, psychological support, and supervised physical exercise can enhance the longer-term effects of surgery. Additionally, a healthy lifestyle can improve modifiable risk factors and enhance longevity. Despite the great effort of the clinical teams to encourage patients towards healthy habits after surgery, one year after surgery some patients need to be readmitted to the hospital due to avoidable complications. This paper presents a digital conversational agent designed to reinforce the advice about protective behaviors provided by the clinical team, contributing to improving and sustaining health outcomes during the first year after surgery.A Design Science Research Methodology was used to integrate the clinical teams' experience and knowledge, and the patients' knowledge into the final artifact. A needs assessment was conducted to identify improvement opportunities in the cardiothoracic surgery service. A systematic literature review was used to characterize the dimensions of the solutions adopted in previous research, and an interdisciplinary team was assembled to address them comprehensively. Two semi-structured interviews were conducted with the relevant stakeholders (clinical team and patients). Additionally, a survey of theories that explain patient behavior change was done to support the personalization of content delivery. Afterward, a platform to provide the intervention was developed in accordance with stakeholders' requirements.Literature review and semi-structured interviews showed that physical activity and a healthy diet influence modifiable risk factors. Thus, these were the behaviors selected as targets of the intervention digital conversational agent, which is being designed based on the behavior change wheel framework.This work intends to incorporate a Behavior Change Theory into the algorithm definition to improve the comprehension of the intervention's effects and the patient's profile over time. The iterative research approach was chosen to continuously improve the artifact's robustness, sustainability, adoption, and usability by having all the stakeholders at the center of the design process.