Site Summary
Studies Count iNumber of unique studies per day or week. Counts studies with at least one feasible-view clip OR at least one acceptable AI result.
Median Quality iMedian quality score (0-1) of recorded clips with feasible views. Higher is better.
Active Devices per Week iDistinct devices contributing to at least one valid study (feasible clip or acceptable AI result) each week.
Clips per View iDistribution of detected echo views across all clips. PLAX is the primary Impella monitoring view.
Quality Distribution iDistribution of clip quality scores: 0-40% (poor), 40-80% (moderate), 80-100% (good).
Distance Distribution iPredicted distance for acceptable AI results only.
Acceptable Studies iShare of ALL studies with at least one acceptable clip. Denominator is every study in Total Studies, so a study where the AI never ran or passed (noAI) counts as not-acceptable. The center matches the Acceptable% KPI.
Clips per Study iHow many clips were recorded per study session (raw clip_recorded count, matching the Clips column in Study & Clip Detail). Every study is counted, so the slices sum to Total Studies. More clips suggest longer, more thorough exams.
Studies per Pump per Day iMaximum number of studies performed on a single pump in one day. Indicates pump utilization intensity.
Distance Std per Pump iStd dev of AI-predicted pump position across all clips per pump (acceptable AI results only). One value per pump.
Studies per Device iNumber of valid studies per device. Counts studies with at least one feasible-view clip OR acceptable AI result.
Median Quality per Device iMedian quality score (0-100%) per device for clips with feasible views.
Median Quality per View iMedian quality score (0-100%) per detected echo view.
Study Count per Day by Device iDaily study count per device. Shows top 8 devices by volume. Hover for details.
Crashes This Month iFatal crash events from Firebase Crashlytics for the selected month.
Crashes YTD
License Details iAll SONO licenses from the License Manager database. Filtered by active institution and license type selections.
Version Adoption iPercentage of licenses running the latest SW version.
Latest: edit
Anomalies
No anomalies detected.
Anomaly Types Legend
Study & Clip Detail
| Date / Time (EST) iWhen the clips were captured, shown in US Eastern time. | Institution iClinical site. Remapped to the licensed site name when the device is in the license roster; otherwise the app-reported name. | Department iDepartment reported by the app (second field of the operator user id). | Device Name iOperator device identifier reported by the app (third field of the user id). | Device ID iHardware device identifier (device_id) of the tablet running the app. Distinct from Device Name, which is the operator-typed label. | SW Ver iApp software version (sw_version) reported by the device for this study's scan events. | Pump SN iImpella pump serial number (impella_serial_number). Shows '-' when the clip carried none. | Clips iNumber of clips recorded in this study (raw clip_recorded events). | % Acc iOf this study's clips whose target view was PLAX (the operator was guided to PLAX and grabbed a clip), the share that produced an acceptable AI read. A PLAX-target clip that got no AI (below the quality gate) or that landed on the wrong view counts against it. Shows '-' if the study had no PLAX-target clips. | Med Dist iMedian AI-predicted pump-position distance across the study's AI-acceptable clips only. Shows '-' when the study has no acceptable clip. | Dist Std iPopulation std dev (÷N) of AI-predicted distance across THIS study's acceptable clips only, at full precision. Measures how consistent the pump-position reads were within the study. Lower is more consistent. 0.00 = a single acceptable clip; '-' = no acceptable AI. Distinct from the pump-level "Distance Std per Pump" donut, which aggregates a pump across the whole month. | Raw Q iRaw classifier quality of the intended (target) view, before the derived-quality override. Near 0 on a misaligned clip means the target view was not achieved. Study row = median raw quality over feasible clips. | Der Q iDerived quality (what the gauges use): No_Heart → 0; Misaligned → the best-matching view's objective value (so it can be high even when Raw Q is ~0); otherwise = Raw Q. Study row = median derived quality. | View iDetected echo view: the view the AI classifier decided the clip actually shows. | Target iTarget view: the view the protocol/step intended. Highlighted when it differs from Detected (a misaligned clip: the operator aimed for one view, the AI saw another). | Acceptable iDid the AI validate at least one clip? Yes = a clip passed; No = the AI ran but none passed; no AI = the AI never ran (e.g. clips below the 0.15 quality gate). |
|---|
Pipeline Run Log
| Timestamp (EST) | Command | Month | Status | Duration | Output | Args |
|---|