Case Study: Rebuilding Nimbus Health's Symptom Tracker for Low-Literacy Users

AI · 6 min read

Nimbus Health's initial symptom tracker required free-text entries and medical terminology selection, which created barriers for a significant segment of patients. Usability tests revealed frustration: users typed short, ambiguous phrases or abandoned entries. The product team partnered with clinicians, health literacy experts, and local community groups to reimagine the interface.

The new design used a tiled pictogram grid for common symptoms, optional voice recording with on-device speech-to-text, and short, plain-language follow-ups that adapt based on initial selections. To preserve privacy and latency, Nimbus ran a compact NLU model on-device to normalize utterances into structured symptom tags before sending anonymized telemetry to servers. This kept sensitive data local until the user consented to share.

Within three months adherence to daily check-ins increased 37%, and clinicians reported higher-quality, actionable symptom data. Crucially, the redesign reduced appointment no-shows by enabling earlier triage through reliably structured inputs. The project underscores how inclusive design plus lightweight local AI can extend critical health services to underserved populations while respecting privacy.