Concepts
ClinSupplyCompass is deterministic and traceable by design. These are the concepts planners meet in the product, explained the way the product uses them.
The extra subjects recruited so the study still reaches its target number of completers.
Enrollment Buffer is a recruitment assumption. If a study needs 100 completers and expects 20 percent attrition, planning for about 125 enrolled patients protects the completer target. It sizes how many patients enter the study. It does not reduce monthly dosing demand over time; that is the role of the dropout rate or the retention curve.
A steady assumption that a fixed share of active patients discontinues each month.
A rate of 0.05 means about 5 percent of active patients are expected to stop treatment each month, so future demand tapers smoothly. Use it when the clinical team expects a simple, gradual discontinuation pattern with no protocol-shaped curve. It is one of two ways to describe expected attrition; the other is Expected Patient Retention.
A protocol-shaped curve stating what share of patients is expected to remain on treatment at each treatment month.
Milestones are entered as pairs, for example 2:85, 4:70, 6:55. That means 85 percent expected on treatment at month 2, 70 percent at month 4 and 55 percent at month 6. The clinical or statistics team often derives the curve from the protocol, the analysis plan or survival assumptions. The application consumes the curve; it never decides the clinical model itself.
A study uses either the Monthly Dropout Rate or Expected Patient Retention, never both at once.
Both fields describe the same thing: how many patients remain on treatment over time. Entering both would reduce future demand twice and under-plan supply, so the application rejects that combination. Choose the flat rate for a simple steady assumption. Choose the retention curve when the expected pattern has a shape, such as many discontinuations after a first disease assessment.
A group of patients who entered the study in the same month.
Patients in different cohorts are at different treatment ages. Twenty patients enrolled in January are in their third treatment month by March, while a March cohort is just starting. Treatment age matters because it determines how much future drug a patient would still consume, which is why attribution of discontinuations is stated carefully.
The amount of clinical supply expected to be consumed in each period.
Demand is driven by enrollment, dosing schedule, treatment duration, retention assumptions and the supply network. Patient consumption is the largest part; site stocking adds the material a newly activated site holds. Demand reduces inventory month by month, and the planning balance is simple: future inventory equals current inventory plus receipts minus demand.
The clinical supply currently available at a depot.
Inventory goes down as demand consumes supply and goes up when receipts arrive. The plan projects an ending inventory for every depot and product in every month. Coverage measures such as months of supply are computed from that projection. Actual on-hand counts entered in Load Actuals reset the starting stock when the plan is recalculated.
New supply arriving into a depot's inventory.
A receipt is incoming material such as kits, vials or finished goods. Receipts increase available supply, so a late or short receipt can turn a safe coverage picture into a shortage risk. Actual receipts entered in Load Actuals are compared with the planned receipt schedule, and the plan reports how closely deliveries followed the plan.
The planning output file that carries future demand quantities to a downstream ERP or planning system.
The Demand Upload lists how much clinical supply is expected to be needed, where and when, by period, location and product. Because it carries future demand, it reflects the retention assumptions and any recorded discontinuations. After a re-plan, the exported file reflects the updated forecast rather than the original one.
How many months current inventory would last if demand continued at the current monthly rate.
MOS is a ratio: inventory divided by one month of demand. With 1,000 kits on hand and 250 kits of monthly demand, MOS is 4. It is a quick health signal. Read it together with Months of Forward Coverage, which walks the actual future demand pattern instead of assuming a flat rate.
How many future months current inventory can cover when demand is walked month by month.
MFC steps through the forecast one month at a time and subtracts each month's demand from inventory until it runs out. It is more realistic than a flat ratio when demand is changing, for example while enrollment is ramping up. A value such as 12 plus means supply outlasts the twelve-month window.
MOS is a ratio against this month's demand, while MFC walks the changing future, so they can tell different stories.
With 1,000 kits and only 100 kits of demand this month, MOS reads a comfortable 10. If demand is rising toward 500 kits a month, the walk runs out after about 4 months, so MFC reads 4. That is common during enrollment ramp up. Reading both together shows whether a reassuring ratio hides a real forward shortage.
A flag raised when projected inventory goes negative or coverage falls below the study's target months.
The plan walks every depot and product through the horizon. A month with negative projected ending inventory is a projected stockout, and coverage below the target months marks the series at risk. The KPI Snapshot counts affected months and series so the planner can see how widespread the risk is before deciding on supply actions.
One governed pass through plan, approval, export, actuals and re-plan for a study period.
A Planning Cycle gives each period an auditable rhythm. Run the plan, review it, approve a baseline, export the Demand Upload, then load actuals and re-plan from reality. Every step is recorded, so the cycle shows who did what and when, and each cycle builds the history the next one learns from.
The reviewed and approved plan of record for a Planning Cycle.
The Approved Baseline is the promise the team signed off. It is the reference against which actuals are compared, so Forecast Accuracy always measures reality against the approved plan rather than a later revision. Re-planning creates a new working version but never rewrites the baseline itself.
A record of active patients who really stopped treatment in a period.
Enrollment and discontinuation are two different facts. If 20 patients enrolled and 3 later stopped treatment, enrollment is still 20 and the discontinuation count is 3. Discontinued patients stop contributing to future dosing demand at re-plan. Site stocking and pooled network demand are not reduced, because sites and networks do not discontinue.
The plan's attrition assumption and recorded discontinuations are compared, never blindly added together.
The dropout rate or retention curve already expects some patients to stop. When actual discontinuations are recorded, only the part beyond that expectation reduces future demand, so the same patient loss is never counted twice. If fewer patients stop than expected, demand is not raised automatically; the planner should revise the retention assumption at re-plan instead.
Recalculating the future plan from the recorded actuals.
Re-plan keeps the past as facts and the future as forecast. Actual enrollment replaces past months, actual on-hand resets starting stock, receipts fold into the schedule and recorded discontinuations reduce future patient demand. The result is a new plan version with updated demand, inventory, coverage and shortage risk, while the Approved Baseline stays untouched as evidence.
A point-in-time health check of the Planning Cycle.
Each snapshot captures coverage, shortage risk, plan versus actual variance, whether a re-plan is needed, Forecast Accuracy and the Data Quality Score. Snapshots are kept, not overwritten, so the cycle's history shows how the picture evolved as actuals arrived and re-plans ran.
A measure of how close the Approved Baseline was to what actually happened.
Accuracy compares actuals against the approved plan: enrollment against planned enrollment, on-hand inventory against projected inventory, and receipts against the delivery schedule. It answers whether enrollment ran ahead or behind plan, and whether stock was higher or lower than expected. It also shows whether the plan was biased high or low overall.
An explainable score showing whether the cycle's actuals are trustworthy enough for re-planning and future learning.
The score combines named checks. Are the expected actuals present, recent and broad enough? Are the values plausible and consistent with the study setup? Can the snapshot be traced to its plan and cutoff? Each check keeps its own score and reasons, so a weak score always says why. High means usable for re-plan and learning; medium means usable with warnings; low means incomplete or inconsistent.
The application's permanent record of plan versus actual for every KPI Snapshot, kept automatically.
Each snapshot stores what was planned, what actually happened, the gap between them and the data quality at that moment, per depot, product and period. Nothing extra is asked of the planner; the message about history captured for learning confirms the evidence was saved. Old records are never rewritten, and this clean history is what future advisory intelligence will learn from.
The trusted planning core calculates supply with traceable rules, and AI never silently overrides it.
In clinical supply, a planner must be able to explain why a study reads safe, at risk or short. Every number the engine produces can be traced to protocol assumptions and recorded facts. AI features explain, summarize and will later advise, but the plan of record always comes from the deterministic engine.
An assistant that explains a study's current situation in plain language, grounded in that study's own data.
AI Study Pulse reads the study's plan, coverage, cycles and actuals and answers questions about them conversationally. It explains and summarizes; it does not change the plan or replace the planning engine. Its conversations can be exported for sharing, and its usefulness grows as planning history and data quality accumulate.
See these concepts at work.
Create your workspace