Two hundred and six residents in care facilities where the clinical XK300 was in use. Half went on to a confirmed urinary tract infection; half had no acute diagnosis. Looking back at the readings, the two groups had already parted days before the test. This page carries every figure from the original study.
This is a retrospective analysis of records from nursing homes using the FDA-cleared XK300, carried out by Xandar Kardian. "Days before" is the interval between a shift in a resident's own readings and the date the infection was diagnosed, measured after the fact. It is part of an ongoing discovery and research programme, and we show what we found. It is not a claim that any Kardian product detects, predicts or diagnoses an infection, and no such use is cleared. XK does not predict or diagnose health events. The later February 2026 analysis, with 108 events, put the average interval at 4.3 days.
Urinary tract infections are a leading cause of preventable hospitalisation in long-term care, and they often go unnoticed until the symptoms are severe.
Nothing to wear, charge or remember. Resting heart rate and breathing rate are read around the clock, from the wall, against each resident's own usual. What follows is what those readings looked like in the days before an infection was diagnosed.
XK radar sensors are deployed across the United States in hospitals, homes and care facilities. We analysed longitudinal data from 206 residents.
Requirement: sensor data available for at least 7 of the 10 days before diagnosis, so the onset of the infection was captured.
Requirement: no acute events within 14 days either side. This group establishes the usual baseline against which a change is measured.
In this single example the breathing rate marked a clear change from the resident's own usual two days before the UTI was diagnosed.
Each chart compares the spread of daily values in the UTI group before diagnosis with the control group. The bar runs from the 25th to the 75th percentile; the tick is the median.
Breaths per minute. Control 13.6 / 15.3 / 16.9 · before a UTI 16.8 / 18.2 / 20.0 (25th, median, 75th).
RRV. Control 0.28 / 0.32 / 0.36 · before a UTI 0.36 / 0.39 / 0.43.
Beats per minute. Control 60.1 / 65.5 / 71.1 · before a UTI 68.2 / 72.7 / 81.3.
Beats for every breath. Control 4.2 / 4.6 / 5.2 · before a UTI 3.5 / 3.9 / 4.3. Breathing sped up more than the heart did.
Residents heading toward a UTI left the bed more often, consistent with more trips to the bathroom, and rested more.
Averages per resident per day or night. 29.6% more bed exits during the day, 38.1% more during the night, 10.7% more sleep at night.
An AI model was trained on these records to learn how the days before a UTI differ from ordinary days, using nothing but the XK300's own readings. Tested afterwards on cases it had not seen:
How much more common each pattern was in the days before a UTI than on ordinary days.
Higher breathing-rate variability was the strongest pattern, more than four times as common before a UTI. Longer sleep and a higher breathing rate came next. Resting heart rate, the measure a routine check looks at, moved least.
These are retrospective results on care-setting records, shown as research. They describe what the readings looked like before a diagnosis that clinicians made; they are not a claim that any Kardian product detects or diagnoses infection. The February 2026 methodology update, across 108 events, reported 4.3 days. See the full outcomes table.
| Per year, 100 beds | As is | If half were avoided |
|---|---|---|
| UTI-related readmissions | 40 | 20 |
| Readmission losses | $608,000 | $304,000 |
| Net yearly saving | $0 | $304,000 |
The "half avoided" column is the study's assumption, not an observed result.
Xandar Kardian company study, December 2025. Retrospective; part of ongoing discovery and research; not a cleared indication. XK does not predict or diagnose health events.