Articles

Praemostro - who, what, where, when, why? On this page, you can find frequently asked questions – and read the answers. If you don’t find what you’re looking for, you’re always welcome to contact us.

Overview of the questions answered below:

  1. What does Praemostro mean?
  2. How and why was Praemostro founded?
  3. How long does it take to start seeing forecasts—from the first meeting with Praemostro until graphs and data appear on a screen in our company/ organization?
  4. Which types of departments can use Praemostro’s system?
  5. Why is Praemostro’s system smart—what do I gain from using it?
  6. Are there other systems like Praemostro’s on the market?
  7. Why not just use regular statistics to predict workload?
  8. How accurate is Praemostro’s system?
  9. What do you do when your predictions are off?
  10. Aren’t you just showing the same graph for every day?
  11. PraeSight updates continuously throughout the day, but does that even make sense? Isn’t it just showing the same graph over and over?
  12. What data is required to connect to Praemostro’s system and access forecasts?
  13. Technically speaking, where do your forecasts run? On your servers or the customer’s own servers?
  14. Is there a difference in product quality when running on your servers versus our own?
  15. Do you have an app?
  16. How many computers can I view my forecasts on?
  17. In what format should the data be delivered?
  18. Is our data secure within Praemostro’s system?
  19. Why are you interested in events?
  20. Do you work with personally identifiable data?
  21. Can we add our own data sources to the forecasts?
  22. Where are your servers physically located?
  23. What are the typical setup costs (hardware, etc.)?
  24. How much does it cost to use Praemostro’s system?
  25. How long am I committed if I subscribe to Praemostro’s forecasts?
  26. Can I get technical support if the system acts up after installation?
  27. We’re having trouble obtaining the data needed to set up our forecasts. Can Praemostro’s technicians assist with this?
  28. Who should I contact for technical support – and how?

About Praemostro and our products

What does Praemostro mean?

Praemostro is Latin for “to foresee.” The name idea came from Mikkel’s youngest son, who thought we needed a Latin name since we originate from the healthcare sector. We believe it accurately describes the services we provide – even if it’s not always the easiest to pronounce. The phonetic pronunciation is präe̯ˈmõːs̠t̪roː.

How and why was Praemostro founded?

Back in 2016, Mikkel was responsible for research on acute patients at a regional hospital. The management had hired an academic mathematical economist, Troels, who was visiting all departments to introduce himself. Having just read an article about how worsening COPD symptoms were linked to weather conditions, Mikkel challenged Troels to predict the number of patients arriving at the emergency department months in advance.

Troels accepted the challenge and developed a computer model that turned out to be highly accurate (+/- 3% per day, see figure). The project was put on hold as other initiatives took priority. However, in 2021, Odense University Hospital slog innovationsmidler op i 2021, og ideen blev genoptaget og siden udviklet til de produkter, vi har i dag.

How long does it take to start seeing forecasts—from the first meeting with Praemostro until graphs and data appear on a screen in our company/ organization?

Our short-term forecasting solution, PraeSight, takes about one month to set up. Apart from finalizing legal agreements and receiving data from the customer, it takes up to three weeks to develop the first model.

Our long-term forecasting solution, PraePlan, can be delivered quickly with minimal waiting time.

Hvem og hvorfor? 

Which types of departments can use Praemostro’s system?

Our system is ideal for situations where tasks cannot be declined—for example, in an emergency department, where patients must be admitted as they arrive, or in an out-of-hours medical service, where every patient needs to be seen as soon as possible. However, it can be used in any setting where arrivals fluctuate, and there is no option to turn people away.

Why is Praemostro’s system smart—what do I gain from using it?

Our systems give customers the ability to look into the future. We enable you to see how busy you can expect to be—either hour by hour, up to 12 hours ahead with PraeSight, or shift by shift and day by day with PraePlan. Occasionally, our predictions may not be spot on, but we always strive to improve. When we miss the mark, we analyze the cause, learn from it, and refine our models for next time. That’s our promise.

Are there other systems like Praemostro’s on the market?

We are aware of one commercially available system (faculty.ai). Others have attempted to develop similar solutions but have struggled with data integration, internal system compatibility, and the mathematical complexity required.

Why not just use regular statistics to predict workload?

Because it simply doesn’t work as well. When integrating as many data sources as we do—while also ensuring that the complex mathematics behind it all fits together—traditional statistical programs fall short. 

Accuracy?

How accurate is Praemostro’s system?

Our short-term forecast, PraeSight, which predicts 12 hours ahead, has consistently maintained an accuracy of +/- one patient per hour (95% of the time, measured over 8 hours) in a department receiving 75-100 patients per day over nearly three years.

Our long-term forecast, PraePlan, which predicts months in advance, achieves an accuracy of +/- 1-2 patients per hour (93% of the time, measured over 24 hours) in a department handling approximately 130-150 patients per day.

However, it’s important to remember that forecasting, by nature, is never 100% precise. There will always be some margin of error—sometimes more than others. We acknowledge this, but we continuously strive to improve. We actively monitor our systems, and whenever we miss the mark, we investigate, adjust, and refine our models.

What do you do when your predictions are off?

We actively monitor our forecasts to identify days when accuracy is lower than expected. The exact criteria for this can be tailored in collaboration with our customers. When deviations occur, we investigate: Was there an event we didn’t account for? Were there weather conditions we hadn’t factored in? Or was it something else entirely? If we cannot find an explanation ourselves, we engage in dialogue with the customer. Our entire ethos is built on making a difference for people who make a difference for others—something we can only achieve by always doing our best.

Aren’t you just showing the same graph for every day?

Of course, there is a general pattern in patient arrivals. In emergency departments worldwide, Mondays and Fridays are typically the busiest days, and the afternoon and early evening are the most active times. However, no two days are identical. Some days are much busier, while others are quieter.

That’s where our system makes a difference. Knowing in advance whether a particular day will be busy or slow allows for better staffing decisions—ensuring both cost-effective and resource-efficient management.

We’ve compiled data from multiple Mondays in one of the emergency departments we work with to illustrate how no two days are the same. The graphs show the expected number of patients over a 12-hour period, starting at 7 AM.

PraeSight updates continuously throughout the day, but does that even make sense? Isn’t it just showing the same graph over and over?

It might seem that way, but that’s not the case. Our models update dynamically as new data comes in—whether from patient counts, weather forecasts, or other sources. This means the system becomes increasingly refined and accurate throughout the day. Below, you can see an example of how the forecast graphs evolve hour by hour over a single day.

Technology and security

What data is required to connect to Praemostro’s system and access forecasts?

All our systems require historical data to be tailored to each customer. We need at least three years of data, but preferably more. If we can go back 6-8 years, that is ideal. Our short-term forecast, PraeSight, also requires "live" data on arrivals within the past hour. Additionally, we need information about events that may impact operations, as well as organizational changes that could affect patient capacity.

Technically speaking, where do your forecasts run? On your servers or the customer’s own servers?

When using our PraeSight solution, customers can choose whether to run forecasts on our servers or their own. However, there may be differences in forecast quality, as explained below. Running forecasts on customers' own servers is technically more demanding and therefore more expensive than using our infrastructure.

Is there a difference in product quality when running on your servers versus our own?

When forecasts run on our servers, we have greater flexibility to update algorithms, integrate new datasets, and refine models. This means that in some cases, we can provide a better and more cost-effective product if forecasts are executed on our systems. And rest assured—both options are secure. We maintain a high level of IT security.

Do you have an app?

No, we do not. However, our forecasts are designed with scalable graphics that automatically adjust to the format you are viewing them on. This means you can access forecasts on your mobile phone, tablet, or computer—whichever suits you best.

How many computers can I view my forecasts on?

In short: As many as you want. We do not limit the number of screens displaying our forecasts—that is entirely up to you. You can access forecasts wherever you prefer and share the link with whoever needs it. Our goal is simply to ensure that you get the most value out of our products.

In what format should the data be delivered?

We have created a technical documentthat describes this in detail. However, in simple terms, we need a file that includes timestamps and counts for what needs to be forecasted, such as: 01-01-2023 08:00-08:59, 23; 01-01-2023 09:00-09:59, 45; etc. Additionally, we require a file with events that may influence the expected number of patients. Of course, we are happy to assist if needed.

Is our data secure within Praemostro’s system?

We comply with the highest security standards. We are fully ISO 27001 and ISO 27002 compliant and operate based on a data minimization principle, ensuring that no personally identifiable data is stored in our systems.

Is our data secure within Praemostro’s system?

We comply with the highest security standards. We are fully ISO 27001 and ISO 27002 compliant and operate based on a data minimization principle, ensuring that no personally identifiable data is stored in our systems.

Why are you interested in events?

Over the years, we have observed that certain events significantly impact demand in the environments where our forecasts are used. Some events are obvious—such as a large concert near an emergency department, which leads to increased patient intake. Others are more surprising; for example, we have seen a decrease in emergency visits on days when Donald Trump was up for election.

Some events can be anticipated on our end and incorporated into the system (e.g., major sporting events, festivals), while others require local knowledge, meaning the customer must help us identify them.

Adding events to our system is quick and straightforward, but it must be done continuously. At the same time, we regularly test our forecasts to refine our understanding of which events truly matter. Some might not require monitoring, while others may need to be added to the list.

Do you work with personally identifiable data?

Our forecasts are based solely on the number of arrivals within a given time frame, such as "12 people arrived between 1 PM and 2 PM." This data is not inherently personally identifiable. However, due to regulations, we establish a data processing agreement with our customers. Since we access the customer’s IT systems, there is a theoretical possibility of identifying individuals. When entering into a contract with Praemostro, we will sign a data processing agreement—either using the customer’s version or our own, which is based on the Danish Data Protection Agency’s standard contract.

Can we add our own data sources to the forecasts?

We can easily integrate new data sources into our models. However, before implementing them, we always conduct an evaluation of their value to ensure they contribute meaningfully to the forecasts. We assess forecast accuracy, data reliability, and credibility before integration.

Where are your servers physically located?

We use server capacity in multiple locations. Our own servers are located in Denmark, and we also utilize a cloud service in Germany. Data sources are integrated from wherever they are stored, while customer servers remain in locations of their choosing.

Economy

What are the typical setup costs (hardware, etc.)?

We offer and strongly recommend running forecasts on our servers. Results are delivered as a webpage, accessible on any device, or as a file in a format specified by the customer. If you choose to run the system on your own machines (e.g., to ensure that no data is sent outside your firewall), you will need hardware equivalent to a high-performance desktop computer running Linux that supports Docker containers to display forecasts. However, training models requires more powerful hardware—please contact us for details.

How much does it cost to use Praemostro’s system?

Pricing depends on the chosen product and whether it runs on our servers or the customer’s own infrastructure. In addition to the software fee, customers should allocate approximately 15 minutes every two weeks to update us on upcoming events expected to impact operations and major organizational changes that could affect patient flow forecasting.

How long am I committed if I subscribe to Praemostro’s forecasts?

This depends on the product. Our long-term forecast, PraePlan, can be purchased as one-time services. PraeSight requires a minimum subscription of 6 or 12 months, depending on where the system is hosted (our servers or yours). For full details, please see our standard terms and conditions.

Support

Can I get technical support if the system acts up after installation?

Yes, we’re always here to help. Our technicians are available on weekdays between 8 AM and 4 PM, and additional support access can be purchased. Contact us for pricing.

We’re having trouble obtaining the data needed to set up our forecasts. Can Praemostro’s technicians assist with this?

Of course. We are familiar with most healthcare systems and already integrate with several of them. If we don’t know your system, we’ll make sure to learn it and guide you through the process. However, this is not included in our standard service, so please contact us for a quote.

Who should I contact for technical support – and how?

You can call our main number at 25 56 30 20 or email us at support@praemostro.com.

More info? Please, do not hesitate to reach out

Want to take a look into the future together?

Get in touch and we’ll get back to you as soon as possible

Get in touch

What are you searching for?