1. What each data type tells clinicians
Blood pressure & heart rate
Clinicians look at systolic/diastolic values over time, morning vs evening readings, and how quickly numbers change with medication or stress. Persistent readings above guideline ranges, especially with symptoms (headache, vision changes, chest pain) are more concerning than a single high number.
Glucose & metabolic panels
Finger-stick and lab glucose values are interpreted relative to fasting vs post-meal, medications, and A1c. Sequences of mildly elevated values can matter more than a single spike. AI models often treat these as time series and look for patterns that suggest increased risk rather than a diagnosis.
Eye images & scanned reports
Fundus images, OCT, and scanned PDF reports are parsed for keywords (e.g. “retinopathy”), measurements (cup-to-disc ratio, thickness), and impressions written by specialists. AI systems focus on structured patterns (lesions, bleeding, vessel changes), but final interpretation remains with the eye specialist.
Medications, prescriptions, allergies
Medication lists help clinicians understand what conditions are being treated, potential interactions, and adherence issues. AI tools often use this as context: combining drug classes, doses, and timing with vitals to understand why numbers look the way they do.
Heart sounds & stethoscope snapshots
Audio from SuperCAPE or digital stethoscopes can show murmurs, extra heart sounds, or rhythm irregularities. AI models typically convert the sound to a spectrogram and then detect patterns; they suggest possibilities (e.g. “murmur-like sound present”) but cannot confirm structural heart disease without imaging and specialist review.