Mapping care pathway
CTO Stan and CEO Tim alternate and complement each other throughout the interview. “The first question companies usually get when they want to market their product is, 'what does it deliver?' We can provide that data. As an example, take a company that has developed a blood test to detect early-stage prostate or bladder cancer. They know how accurate the test is; if there are 100 men with prostate cancer, how many of those men does the test detect? And also: if there are 100 men without prostate cancer, how often does the test indicate that something is wrong when it is not? Those are the two starting values in diagnostics. And on the basis of that accuracy, the tests are now being sold. We go further. We make analyses per disease in which we map the entire care pathway: someone comes to the doctor, what is the next step, what are the treatments and so on. We create a predictive model of patients over time. What if patients get a right result or a wrong test? What does that mean? We simulate their health status and hang outcomes on that.”
“With AI in particular, the technology is constantly evolving and so are the products and services in which you apply this AI technology. That means that your evidence must also be dynamic.”
Customized analyses
“Hospitals and insurers want to know: if I use that test, what does it do to the patient's health? What does it do to quality of life and how does it change the cost of care? We have put that into a software package, and our customers can use that software to show to different hospitals. The accuracy of the test is the starting value. We also use data from the hospital. What is the age distribution of patients, what treatments are they doing, what does a diagnostic pathway look like right now? In this way we make customized analyses and we can show very specifically for that particular hospital: if you start using that test, this is what will change for your patients in terms of survival and quality of life, and also: this is what it will do to the costs you incur throughout that entire pathway. Then such a hospital can make a better assessment of whether they are going to deploy that test.”
Dynamic proofing
The startup helps companies with “dynamic proofing. What's so 'dynamic' about it? “We don't believe it works to calculate something once, and that's it. With AI in particular, the technology is constantly evolving, and so are the products and services in which you apply this AI technology. That means that your evidence must also be dynamic. A year after your analysis, your product has already changed and the added value has actually become obsolete. With our software, you keep your evidence up to date.”
'We develop everything ourselves'
Is Medip Analytics itself making use of AI? “Our customers are AI companies but we are not currently doing anything with it ourselves. We program the models we create ourselves. AI deals with large data sets, and then we let the algorithm write its own code. We don't do that; we write manually. We first have to figure out how the care pathways work, and we mostly rely on existing literature and guidelines. In the future we may be able to do that from data, from the hospitals, but for now, we still develop everything ourselves. We have all the AI knowledge in-house but we don't have large data sets, which is the main reason why we are not working with AI yet.”
Growing company with international customers
Medip Analytics is a young company founded in 2021 by two biomedical scientists. The men know each other from their work at the Radboudumc. A lot has changed in three years: “We started with two, now there are five of us. Our office is on the Novio Tech Campus, a great place to grow. We have international customers. American, Swedish and Swiss companies; the countries where the big biotech companies are.”
One of its first customers is Thirona. The Nijmegen spin-off of the Radboudumc develops and sells AI-driven solutions for the analysis of lung images. A specific AI algorithm can assess, based on CT images of the lungs, whether a COPD treatment is working or not. Medip Analytics wrote a program that allowed Thirona to demonstrate the added value of their technology over current methods. “Thirona had created something great, but then you still have to market it. With our software, they were able to show the health and financial benefits of their method in detail. They used that data for, among other things, a European funding application, which was granted.”
'In addition to proof, optimization'
“We not only calculate the added value of a product, we can also help optimize it. Suppose you have a test with a score between zero and 100 points. Above 50 points, we call someone 'sick.' You set that so-called cutoff value yourself. You can also say, we call everyone above 30 points sick. The lower the cutoff value in this case, the more “sick” people there are. But that also means that you call more people ónjustly sick. The cut-off value is often determined quite subjectively, like: 'we do 50 points, then we're about right'. We can calculate for companies, for a specific hospital, where you can best place that cut-off value in order to offer as much added value as possible for that hospital, to make the costs as favorable as possible in relation to the health outcomes. It makes quite a difference if the next step is a blood test or if you need surgery right away.”
'Funding from SFG gave us a boost'
With support from Briskr, Oost NL, Mercator Launch and KplusV, Medip Analytics received funding from the Startup Fonds Gelderland. That gave us a boost, it allowed us to properly establish the basis of our software.” What is unique about this fund is that it is linked to support program Briskr, which guides startups before and after a loan is granted. Oost NL is fund manager of Startup Fonds Gelderland. The startup also used De Groeiversneller, a support program for entrepreneurs funded by the province of Gelderland and implemented by Oost NL.
Future plans
There is no shortage of future plans. The guys are currently developing an application to calculate the environmental impact of new technologies within healthcare. In addition to health outcomes, costs and time of healthcare professionals, that will be the startup's fourth pillar. The company is also in talks with several parties to link its software to AI platforms focused on imaging.
Want to learn more about Medip Analytics? Visit https://www.medipanalytics.com/
Want to know more about AI? Check out www.aihub-oost.nl
The AI -hub Oost-Nederland is there for entrepreneurs (startups, scale-ups, SMEs and large companies), knowledge institutions and governments in the region who (want to) work with AI. This applies to parties that are already fully engaged in AI applications, as well as parties that are orienting on AI.
