"An Apple a day keeps the doctor away": How the use of commercial algorithmic devices influences the cardiologist -patient interactions during diagnosis

Wearable devices increasingly provide consumers with real-time information and indications of cardiac problems. Despite the recent studies on the prevalence of these technologies and the perceptions of medical professionals, we have limited understanding of how the process of diagnosis happens when patients bring algorithmic indications and evidence to the consultation. Methods Through a qualitative, interview-based approach, we confronted 19 cardiologists (9 juniors and 10 seniors) with nine different clinical scenarios in which the patient is presented as being diagnosed for having atrial fibrillation (AF) by Apple Watch. The scenarios represented how the condition of the patient is likely for having AF and whether algorithmic evidence on AF is present. By systematically analyzing the diagnostic conversations between the doctor and the patient (the interviewer), we identified the different ways in which cardiologists react to the algorithmic indications and evidence. Results When irrelevant algorithmic evidence (a noisy registration) is introduced, cardiologists tend to reject the indication of the algorithm, even when the clinical condition of the patient is likely to develop AF. Senior cardiologists were more likely to reject the algorithmic indication (29%) than junior cardiologists (19%). Conclusions The way cardiologists use algorithmic evidence provided by wearable devices is strongly dependent on how the evidence is presented. Our data may create awareness among cardiologist on how algorithmic indications and evidence that patients bring to the consult, in particularly misleading and distracting evidence, may adversely affect the diagnostic decision-making process.