Voice & check-ins
Health, in their own words.
Guided conversations and repeat check-ins add the experience that a measurement alone cannot capture.
Discuss this dataCustom collection concept. Availability and review depend on the study.
What the collection
can include.
Sources and capture methods are selected with your team before recruitment.
Spoken check-ins
Study prompts answered in everyday language. Language, capture conditions and the question version accompany the response.
Structured self-reports
Symptom ratings, routines and follow-up questions with defined response windows. Validated instruments require appropriate licensing and use.
Audio with context
Permissioned recordings, transcripts and segment timestamps, with review depth set for the intended task.
Made for your task.
Health-language understanding, speech transcription and extracting structured information from participant narratives.
Cohort
Languages, recording conditions, prompt design and follow-up frequency.
Preparation
Transcription, segmentation and agreed redaction, with links to source audio where permitted.
Review
Audio quality, transcript accuracy and annotation consistency. Self-report is not a clinical diagnosis.
Delivery
Audio manifests, timed transcripts, prompt versions and reviewed annotations.
Scope matters.
These are participant check-ins, not clinician reasoning traces. Speaker identity and third-party speech need specific handling. Voice itself can remain identifying.
Inspect the delivery.
Data & metadata
Agree on source files, structured observations, identifiers, timestamps and permitted linkage.
Labels & quality
Document label definitions, reviewer roles, uncertainty and acceptance rules. Separate collected data from derived interpretation.
Rights & access
Define permitted use, recipients, transfer method, retention and withdrawal handling before collection.
