## Device - **Manufacturer:** Löwenstein Medical - **Model:** Löwenstein Eyra (DeviceType 27, reported as prisma25S/prisma25ST family) - **Parser:** `cpap_parser.parsers.prisma_line` ## Observed behaviour When parsing `therapy.pdat` with `include_timeseries=True`, the following timeseries channels contain implausible values that appear to be raw flow-rate data rather than the named signal: | Channel | Expected range | Observed values (sample) | |---------|---------------|--------------------------| | `leak` | 0–50 L/min | 400, 514, 703, 5124 | | `respiratory_rate` | 0–60 BrPM | 0, 400, 83, 15615 | | `tidal_volume` | 50–3000 mL | 10240, 400, 83 | | `minute_ventilation` | 0–30 L/min | 40, 1.1, 65.6, 1557 | The `mask_pressure` channel is correct (4–10 cmH₂O). The `flow_rate` channel contains plausible values [0.0, -1.0, 54.0, 75.0] consistent with a respiratory flow signal. The suspect channels all share the same sequence of values (e.g. 400 → 83 → ...) suggesting a single raw channel is being broadcast into multiple named fields. ## Reproduction ```python from cpap_parser.core import create_parser d = create_parser().parse('/path/to/ExampleFiles', include_timeseries=True) s = d.sessions[0] # any session with data ts = s.timeseries print(ts.leak[:5]) # e.g. [400.0, 39.0, 514.0, ...] print(ts.respiratory_rate[:5]) # same pattern print(ts.tidal_volume[:5]) # same pattern print(ts.mask_pressure[:5]) # plausible CPAP pressure values ``` ## Impact SleepLab (https://github.com/joshuamyers-dev/sleeplab) is using cpap-parser for multi-manufacturer import and hits this on import of Löwenstein Eyra data. We are working around it by clamping out-of-range values before DB insertion, but the channels are effectively unusable for this device. ## Sample data Happy to share the `ExampleFiles` archive (`config.pcfg` + `therapy.pdat`, ~3 MB) if that would help. Please let us know the best way to provide it.