Paper short abstract:

This paper explores the intersection of patient-reported outcomes data and AI in recent attempts to "humanize" the digitalization of healthcare. What understandings of "voice", "humanization" and "patient participation" are at play, as synthetic voices of patients substitute spoken ones?

Paper long abstract:

Patient-reported outcomes (PRO) data are data on an individual patient’s health-related functioning and quality of life, reported through questionnaires as part of a treatment trajectory. Increasingly administered through digital platforms, these tools aim to record and utilize patients’ subjective experiences to improve shared decision-making, to prioritize activities and to inform management decisions in healthcare. In healthcare policy, PRO-data has come to feature as a form of feasible and rational way of “giving voice” to “what matters to patients” while increasing efficiency and effectiveness.

More recently, in our current age of AI-hope and hype, PRO-data has moreover become attached in interesting ways to discussions of patient-centered AI. In this paper, I will take my point of departure from propositions made in digital health literature suggesting: 1) that the inclusion of PRO-data in AI model training “is a critical part of the humanization of AI for health”, 2) that AI should be used to transform PRO-data into meaningful narratives to increase clinicians’ engagement, and 3) that emphatic AI chatbots should be used to collect PRO-data more efficiently from patients.

What understandings of “voice”, “humanization” and “patient participation” are at play in these visions for patient data and AI in healthcare? How have the synthetic voices of patients in the form of PRO-data become so valuable for healthcare systems and digital health innovation, while other versions of patient voices seem neglected? What may be the consequences for individual patients and health systems at large of mobilizing these synthetic voices?