alive.actor.wearer@1.0.0

Synthetic device wearer

Produce device data streams with gaps and a known ground-truth label.

Boundary

Per-person adherence, independent per-sample abnormality; no physiology.

Assumptions

Each person has a fixed adherence level; samples are otherwise independent.
Adherence does not decay over time (the cited averages hide within-person decay).

Limitations

Abnormality is a synthetic label, not a clinical state.

Known failure regimes

Deteriorating patients, correlated non-wear, circadian patterns.

Parameters

Name	Default	Provenance
`samples_per_day` Scheduled measurements per day.	1 per_day	synthetic assumption Once-daily home blood-pressure readings, the regimen in the cited program.
`missing_data_probability` Chance a scheduled measurement is missing. A distribution is drawn once per person (heterogeneous adherence).	beta(alpha=1.3, beta=2.56) probability	public evidence Beta moment-matched (mean 0.34, sd 0.21) to the cited distribution of days with readings per month (mean ~19.9/30). Single vendor program, BP cuffs only.
`abnormal_probability` Chance a measurement is (synthetically) abnormal.	0.15 probability	public evidence Derived: ~0.10 manual (threshold) alerts per patient-day divided by ~0.66 readings per day. One Dutch BP program; alert rates depend on thresholds.

Evidence behind the defaults

rpm_days_with_readings_per_monthevidence: lowthin or conflicting

Per-patient average number of days per month with at least one home BP reading in an outpatient RPM program. Drives device-stream missingness and the CMS 16-day billing threshold.

One uncontrolled, vendor-affiliated cohort with survivorship bias and no reported dropout counts.

Frazier WD et al. (2023). Six Months of Remote Patient Monitoring Is Associated with Blood Pressure Reduction in Hypertensive Patients: An Uncontrolled Observational Study

rpm_alerts_per_patient_day_outpatientevidence: lowthin or conflicting

Rate of rule-based alerts (BP threshold, complex, and overdue/inactivity) per enrolled patient-day in an outpatient hypertension telemonitoring program. The program splits them into all alerts and alerts that needed manual nurse processing.

Single site with a small n, and the denominator is approximate.

van Steenkiste J et al. (2025). Alert Reduction and Telemonitoring Process Optimization for Improving Efficiency in Remote Patient Monitoring Programs: Framework Development Study