Imagine receiving an order at 3:59 p.m.
The product still needs to be manufactured, tested, released, packed and shipped that same day, because it must arrive in time to be used at 8:00 a.m. the following morning.
Sounds challenging?
Now imagine that the radioactive starting material needed for that product had to be ordered weeks earlier. The supplier needed a forecast well in advance to schedule production, complete processing, perform quality control and release the material before international shipment.
And the customer? They may not even know what they need until that afternoon.
So how do you forecast something that hasn't been ordered yet?
Forecasting is not predicting the future
Of course, demand isn't entirely unpredictable. Historical patterns tell us a great deal. We can recognise differences between weekdays, anticipate the effects of public holidays, identify seasonal trends and learn from previous demand.
But patterns aren't orders, and even a good forecast cannot tell us exactly what will happen.
For most conventional materials, we have options. We can build inventory, hold safety stock or reorder when quantities fall below a certain level.
Radioactive materials introduce an additional challenge.
Our inventory is disappearing while we're holding it.
Order too little, and we risk being unable to supply a product needed for patient treatment. Order too much, and expensive radioactive material may decay without ever being used.
There is no opportunity to put yesterday's unused activity back on the shelf for next month.
The forecast is only the beginning
The real question isn't simply how accurately we predict demand, but what we actually do with that prediction.
A supplier may need our forecast weeks ahead. Based on it, they reserve production capacity, plan resources and arrange quality control and logistics. We then have to decide how much material to order, how much uncertainty we are prepared to absorb, and what production capacity we need to keep available.
All before the actual customer orders arrive.
And when they do arrive, we may have only hours to turn those earlier decisions into a product that can actually be used.
The forecast therefore has consequences long before anyone knows whether it was correct.
One forecast. Five dimensions.
This is where Safety, Quality, Delivery, Cost and People become inseparable.
Safety: Last-minute demand must never become a reason to rush operations, increase unnecessary radiation exposure or compromise safe working conditions.
Quality: Materials, equipment, manufacturing processes and release arrangements must already be ready. Quality cannot suddenly become optional because the patient is waiting.
Delivery: Having enough activity is not the same as having a usable product at the right location, at the right time. Production, release and transport all have to work together.
Cost: Reserve too little and you risk losing the ability to supply. Reserve too much and valuable activity disappears through radioactive decay. Flexibility has a cost, even when nobody uses it.
People: Experienced planners, operators, quality professionals and logistics teams make these systems work. They understand demand patterns, recognise unusual situations and know when decisions need to change. But resilience should not depend entirely on a few people who somehow always manage to make the impossible possible.
And what if the supplier is on another continent?
Then production schedules, transport routes, international borders, shipment cut-offs and time zones become part of the same planning problem.
Would owning a cyclotron solve it?
Having your own cyclotron can provide an important advantage, but only for radionuclides that can be produced using cyclotron technology, at the required activity, quality and scale. Not every radionuclide can be produced this way, and technical feasibility does not necessarily mean sufficient production capacity or economic viability.
Where local cyclotron production is a realistic option, it can shorten the supply chain, reduce dependence on international transport and give the organisation more direct control over production scheduling.
But it doesn't eliminate uncertainty.
You still need sufficient production capacity, reliable equipment, qualified people, suitable target materials, available manufacturing and QC resources, and realistic contingency arrangements. Instead of depending on an external supplier's production schedule, you now depend on your own ability to deliver.
You have moved the constraint. You haven't necessarily removed it.
But when do we actually think about all this?
And this brings me to something that I believe deserves much more attention.
When we plan a new radiopharmaceutical manufacturing facility, we spend considerable time discussing production technologies, equipment, cleanrooms, capacity, compliance requirements and investment costs. We define what we want to manufacture and how the facility should be designed to do it.
But how often do we really discuss the uncertainty that the future operation will have to manage?
What happens when actual customer demand differs from the forecast? How much flexibility will the facility need, and how quickly must it respond? What happens if the starting material arrives late, if production volumes fluctuate, or if a critical supplier cannot deliver? And how will those situations affect the people, equipment, quality systems and logistics that we are designing today?
Too often, these questions are left for the operational team to solve once the facility is built, when many of the decisions that determine the available options have already been made.
A facility can be technically excellent, fully qualified and compliant with all applicable requirements, yet still struggle to deliver reliably because the operational reality was not sufficiently considered during its design.
And by then, creating additional flexibility may require considerable investment, complicated workarounds or permanent dependence on the experience and extraordinary efforts of the people running the facility.
Wouldn't it make more sense to explore these questions much earlier, when we are still defining the business case, the operating model and the design requirements?
After all, the objective is not simply to build a facility capable of manufacturing a radiopharmaceutical under defined conditions. It is to build an operation capable of delivering what patients need, even when reality doesn't follow the original plan.
What are we missing?
Perhaps we sometimes measure forecasting performance by how closely actual orders match the forecast. But is that really the most important measure?
A forecast can be inaccurate while the system still delivers reliably. Another forecast can be remarkably accurate, yet one unexpected order or disruption can leave a patient without treatment.
The objective cannot be forecast accuracy alone. It must be to understand uncertainty, decide how much flexibility the system needs, make that flexibility possible before it is required, and know what to do when reality differs from the plan.
And perhaps we should stop treating this as something to figure out once production starts. The ability to manage uncertainty is not merely an operational challenge; it is a strategic design consideration.
Because in radiopharmaceutical manufacturing, time is not simply a scheduling parameter.
Time changes the product itself.
So here's my question:
Are we planning to meet our forecast, or are we designing an operation that can meet the needs of patients when the forecast turns out to be wrong?
What Am I Missing?
Made with AI. Thought by me.
