Finding the delay. Testing a better way.
A detailed AssayIQ Pulse demonstration of how a regional laboratory network can turn operational data into a focused improvement agenda, an accountable action plan and a more informed investment discussion.
19.5% lower than baseline
66.4 min baseline32.0 pp vs baseline
3,787 / 5,040 follow-up events19.3 min below baseline
85.7 min baseline-3.1 pp vs baseline
158 / 5,040 follow-up eventsThe question
Where is turnaround time being lost, which assays still need attention, and what evidence should leadership review before changing staffing, workflows or equipment? This fictional network brings those questions into one reporting process.
Two 14-day synthetic populations, 5,040 events each. Phase totals are calculated from event records, not from rounded labels.
One network. Several different operating questions.
The demonstration uses three fictional laboratories to show how a common metric framework can support local action without hiding variation behind a network average.
| Dimension | Included scope |
|---|---|
| Sites | Central Lab, North Lab and West Lab |
| Work areas | Chemistry/immunoassay and hematology event labels |
| Instrument labels | 12 across the network; 4 per site |
| Shifts and assays | 3 shifts; Potassium, Troponin, CBC, TSH and Lactate |
| Review window | October 1-28, 2026; two equal 14-day periods |
| Event population | 10,080 original test events; 360 per day |
The operating brief
- 01
Improve the predictability of turnaround while keeping reruns and local quality concerns visible.
- 02
Identify the assays and shifts that warrant a focused review rather than applying a network-wide fix.
- 03
Understand workload and test a capacity hypothesis before making a purchasing commitment.
- 04
Bring methods, assumptions, actions and decision owners into the same leadership discussion.
| Lab | Mean B | Mean F | Target B | Target F | All events |
|---|---|---|---|---|---|
| Central Lab | 63.5 | 50.7 | 48.8% | 82.0% | 3,360 |
| North Lab | 66.2 | 53.4 | 43.5% | 76.0% | 3,360 |
| West Lab | 69.4 | 56.1 | 37.0% | 67.4% | 3,360 |
Build confidence in the baseline before interpreting change.
A credible improvement report begins with consistent definitions and a documented population. The public Pulse tool demonstrates the calculation path; a client deployment requires an agreed source mapping and local acceptance review.
| Check | Finding | Meaning |
|---|---|---|
| Event count | 10,080 rows | One original test event per row |
| Unique event tokens | 10,080 unique IDs | No duplicate event tokens |
| Validity / exclusions | All rows valid / 0 excluded | No silently discarded records |
| Date coverage | 28 dates / 360 events daily | Complete constructed calendar |
| Period mix | 5,040 events in each period | Equal site, shift and assay counts |
| Source capture | Not independently verifiable | No LIS completeness claim |
Proposed implementation sequence
- 01
Define: agree the operational question, local targets, source fields, period boundaries and accountable sponsor.
- 02
Map and reconcile: compare accepted exports with source-system control totals; review missing and duplicate events.
- 03
Review: examine phase-level TAT, the long tail, site/shift variation and assay-specific target misses.
- 04
Act and sustain: select one test of change, assign an owner, review guardrails and document the next decision.
Decision boundaries
The public version uses CSV data and local browser calculations. Automatic feeds, secure hosted access, production security controls and local validation must be scoped before deployment. The synthetic period change was built into the generator; it is not attributed to the proposed workflow actions.
The average improves. The assay detail changes the priority.
Across the synthetic network, mean turnaround moves from 66.4 to 53.4 minutes and the 90th percentile moves from 85.7 to 66.4 minutes. Target attainment rises by 32.0 percentage points.
Minutes per original event; 360 events per date. The dashed boundary separates baseline from follow-up. This is a constructed change, not a measured intervention effect.
| Assay | Mean B | Mean F | Target B | Target F | F events |
|---|---|---|---|---|---|
| Potassium | 64.0 | 51.5 | 42.3% | 83.7% | 1008 |
| Troponin | 65.2 | 51.7 | 38.8% | 82.5% | 1008 |
| CBC | 64.2 | 51.3 | 40.3% | 84.0% | 1008 |
| TSH | 73.8 | 61.2 | 86.6% | 99.8% | 1008 |
| Lactate | 64.6 | 51.2 | 7.5% | 25.6% | 1008 |
Put the time of demand beside the time of disruption.
An hourly workload view can help a team decide when to investigate transport runs, receiving handoffs, staffing coverage or maintenance. This case adds an explicitly constructed hourly allocation to demonstrate that review.
Average events per calendar day by hour. Weekday = 20 days; weekend = 8 days. Hours are deterministically allocated within shifts and are not recorded timestamps from the base CSV.
| Shift | Mean B | Mean F | Target F | Reruns F | All events |
|---|---|---|---|---|---|
| Day | 64.4 | 51.5 | 79.3% | 2.5% | 3,360 |
| Evening | 66.5 | 53.5 | 74.6% | 3.6% | 3,360 |
| Night | 68.3 | 55.2 | 71.5% | 3.3% | 3,360 |
An operating experiment to consider
In the constructed weekday profile, demand peaks at 09:00. A team could compare that demand with actual maintenance and availability logs, then consider a different service window or receiving handoff. No maintenance change is modeled as having occurred. Manufacturer instructions, staffing and clinical coverage remain constraints.
Test a capacity hypothesis before buying a solution.
Test counts show workload distribution. They do not establish instrument utilization. This separate planning scenario adds scheduled, maintenance and busy minutes for four fictional Central Lab instruments.
Utilization = busy minutes / (scheduled minutes minus maintenance). The stress case assumes busy time scales linearly with demand. These are constructed instrument-log inputs, not measured device performance.
| Instrument | Events | Available min | Busy min | Modeled use | +15% load |
|---|---|---|---|---|---|
| CHEM-1A | 1,344 | 38,640 | 10,080 | 26.1% | 30.0% |
| CHEM-1B | 1,344 | 38,640 | 9,720 | 25.2% | 28.9% |
| HEM-1A | 336 | 38,640 | 6,048 | 15.7% | 18.0% |
| HEM-1B | 336 | 38,640 | 5,880 | 15.2% | 17.5% |
The decision the scenario supports
The model suggests a question worth checking: can existing equipment accommodate a defined increase in demand? It does not establish peak-hour headroom, assay-menu capacity, uptime, staffing availability or resilience after a device failure. Parallel processing and assay-specific constraints require local review.
Read the improvement signal alongside the unresolved work.
Rerun flags fall from 314 to 158 events, a 49.7% relative difference across equal populations. Automatic release moves from 68.1% to 85.6%. Neither change alone establishes clinical quality.
5,040 original events. Bin boundaries are explicit and mutually exclusive; counts reconcile to the follow-up population.
| Measure | Baseline | Follow-up | Interpretation |
|---|---|---|---|
| Rerun rate | 6.2% | 3.1% | Reason and necessity still require review |
| Rerun events | 314 | 158 | At least one rerun; not a run count |
| Automatic release | 68.1% | 85.6% | Not proof that release rules are safe |
| Within local target | 43.1% | 75.1% | Assay exceptions remain visible |
Guardrails for a real pilot
- 01
Classify repeat work by reason before deciding it is avoidable; clinically necessary repeats remain necessary.
- 02
Review QC, calibration, specimen quality, critical-result escalation and amended reports using approved local systems.
- 03
Change autoverification or release rules only through the laboratory's approved validation and authorization process.
- 04
Keep assay-specific target misses and source-data gaps open until the responsible leader reviews them.
Separate reagent expense, staff capacity and cash savings.
A useful business case makes its assumptions visible. This scenario models only incremental reagent expense and a stated staff-time assumption. It does not report realized cash savings or a return on investment.
| Item | Baseline | Follow-up | Difference / basis |
|---|---|---|---|
| Repeat reagent proxy | $709.95 | $389.45 | $320.50 lower |
| Rerun-flagged events | 314 | 158 | 156 fewer flags |
| Assumed work per flag | 4 minutes | 4 minutes | Planning input, not a time study |
| Modeled staff capacity | 20.9 h | 10.5 h | 10.4 hours |
| Capacity value at $45/h | $942.00 | $474.00 | $468.00; not cash saved |
| Confirmed avoided capital | Not established | Not established | No amount recognized |
| Minutes / flag | Capacity hours | Capacity value at $45/h |
|---|---|---|
| 2 | 5.2 | $234.00 |
| 4 | 10.4 | $468.00 |
| 6 | 15.6 | $702.00 |
What finance would need before a buying decision
Verify unit costs, rerun multiplicity, actual hands-on time, implementation expense, ongoing support and any costs required to realize a cash reduction. Released staff time can support other work without reducing payroll. A 14-day synthetic comparison should not be annualized into a promised savings figure.
Turn the report into an owned 30/60/90-day review.
The actions below are proposed demonstration work, not completed client actions. Each has an accountable role, a review horizon and evidence needed for a decision.
| Horizon | Owner role | Action | Evidence for review |
|---|---|---|---|
| Day 30 | Lab operations lead | Confirm source completeness and the Lactate sample pathway | Reconciled counts; timestamp map; local target approval |
| Day 30 | Quality lead | Classify rerun causes and review release-rule governance | Reason-coded sample; QC and escalation review |
| Day 60 | Shift lead | Test one approved receiving or handoff change | Comparable pre/post populations; workload and quality guardrails |
| Day 60 | Technical lead | Reconcile actual maintenance and capacity logs | Peak-hour queues; device availability; contingency assessment |
| Day 90 | Lab director + finance | Review effectiveness and approve next scope | Sustained measures; full-cost model; unresolved-risk register |
Decision gates
- 01
Data gate: agreed definitions, reconciled control totals and documented exclusions before interpretation.
- 02
Pilot gate: a limited change, an accountable owner and local approval before operational rollout.
- 03
Effectiveness gate: compare like periods and assay mix; review more than one summary average.
- 04
Scale gate: expand only after leaders review sustained performance, quality effects, staff workload and the full commercial scope.
Make the evidence easy to inspect.
The event population, calculations and supplemental assumptions are available separately so that a reviewer can reproduce the report and distinguish data-derived findings from constructed planning inputs.
| Metric | Definition |
|---|---|
| Total TAT | pre_min + analytic_min + post_min; minutes per original event |
| Mean / P90 | Event-weighted mean; P90 linearly interpolated at (n-1) x 0.9 |
| Within target | Count(TAT <= target_min) / original events x 100 |
| Rerun / automatic release | Sum of the corresponding 0/1 flag / original events x 100 |
| Percentage-point change | Follow-up percentage minus baseline percentage |
| Relative mean TAT change | (Baseline mean - follow-up mean) / baseline mean x 100 |
| Repeat reagent proxy | Sum(rerun flag x reagent_cost); one extra run assumed |
| Modeled utilization | Busy / (scheduled - maintenance) minutes x 100 |
Scope and limitations
Source: deterministic AssayIQ synthetic event generator v1, 10,080 events, no patient data. Baseline: October 1-14; follow-up: October 15-28, 2026. Local example targets: Lactate 45 minutes; Potassium, Troponin and CBC 60 minutes; TSH 90 minutes. These are illustrative goals, not clinical or accreditation standards. Source-system completeness, actual intervention effects and clinical outcomes are not established.
Supplemental assumptions
Hourly event allocation uses fixed shift weights and reconciles to the original daily totals. Capacity uses invented instrument-log inputs: 40,320 scheduled minutes and 1,680 maintenance minutes per device over 28 days, plus stated busy minutes. Finance assumes four staff minutes per rerun flag and $45 per loaded staff hour. The 15% demand stress assumes a linear busy-time response.
Reference and independence
Roche's M Health Fairview case study was consulted as a benchmark for report depth and structure. Its customer results, quotations and endorsements are not used as AssayIQ results. NAVIFY is a Roche trademark. AssayIQ Global is independent and is not affiliated with Roche or M Health Fairview.