Reports and analytics
Every figure on these pages is computed from released results and recorded events. None of them writes anything, so a page failing to load cannot affect a patient record. None of them uses machine learning: where a rule is evaluated, it is arithmetic on data the laboratory recorded.
For the patient-facing report — the frozen, versioned document a patient reads — see the patient portal.
Saved reports
/reports — "Build one, or run one somebody already built." A report is built over released results, grouped and counted, and can be saved. A saved report is shared, not personal, so two people running the same report get the same definition. Work in progress is not countable here, because a figure that changes with the moment it was run is one two people will disagree about.
Key figures
/analytics and /analytics/kpi — "Every figure names what it counts and when it stops counting." Released results, turnaround, specimens rejected and critical results escalated, each with a table stating exactly what it counts and what it excludes, plus a breakdown by department.
Turnaround is always qualified: receipt-to-result is what the laboratory controls; collection-to-result is what a clinician experiences. The product does not print an unqualified "TAT". Fewer than five released results in the month shows the empty state instead of a median.
Volume outlook and peer benchmark
/analytics/forecast and /analytics/peer-benchmark are two tabs of one chart page.
- Volume outlook — released results a month, on a y-axis that starts at zero and is fixed per laboratory. The projection is the same month last year multiplied by the recent three-month growth: arithmetic, labelled as projected, drawn only after twelve measured months, and it feeds nothing else.
- Peer benchmark — turnaround against comparable laboratories on a fixed 0–72 hour scale. The words are computed from the figures; the page never names a cause, because the system cannot derive one. With fewer than five peers there is no benchmark heading.
Rule findings
/analytics/smart-validation and its tabs — Validation, Anomalies, Equipment, Repeat patterns, Report summary, Interpretation, Rejections, Staffing, Add-on tests — each list findings with a title and a basis line filled from recorded data.
Where the product's data cannot run a rule, it says so rather than faking a result:
- Equipment rules read service and calibration due dates, not run counters.
- Staffing shows "This laboratory records no roster, so these rules have nothing to read." — never "found nothing", which would claim the rules ran.
- A count of people under five is not shown.
Quality tools
Turnaround, six sigma, stock cover and satisfaction are one screen with four tabs under /quality/*, documented with quality control.
Other analytical areas
Research registers, population counts and telemedicine bookings have their own page: Research, population health and telemedicine.
Related: Dashboard · Laboratory analytics in My lab