Project Update: Enabling Personalized Interventions - EPI

The general goal of EPI is to investigate ways to make health-related data, which may be located at different organizations, available for analysis to support personalized self and joint management during medical interventions or treatments. This should lead to responsible use of data from various sources for practical purposes such as lifestyle advice, prevention, diagnostics, and personalized treatment. The concept has been dubbed 'digital health twin', with the ultimate aim of providing personalized, effective, real-time feedback in line with the choices each individual has made.

In late November 2022, we spoke with Dr. Paola Grosso, associate professor at the University of Amsterdam and principal investigator of EPI (successor to Prof. Dr. Ir. Cees de Laat in this project), and project leader Prof. Dr. Sander Klous, a professor at the University of Amsterdam and partner at KPMG.

EPI brings together research institutions, healthcare providers, and the business community

EPI is carried out by an interdisciplinary team of medical specialists, data scientists, ICT infrastructure experts, and experts in the field of artificial intelligence and law, encompassing all relevant disciplines needed to arrive at a viable solution. The project is executed through three representative health-related cases at the St. Antonius Hospital, UMC Utrecht, and the Princess Máxima Center. In addition to the University of Amsterdam (for infrastructure and architecture), the VU and CWI are involved from their respective specializations. Other knowledge partners include SURF, KPMG, and Philips.

This research (initiated in 2019) falls under the Data2person call to promote multidisciplinary research contributing to the development of effective, efficient, and responsible personal empowerment methods for a healthy society in the future. Sander provides an explanation of this collaboration where data and data processing are central. "Establishing this collaboration was very complex, particularly contractually. We invested a lot of time to set this up properly. The result is that this unique collaboration, involving a large number of parties, now also works very well in practice, and we are very proud of that."

Data Security

An important part of EPI is the idea that data should be copied as little as possible. Each additional copy makes it more complex to control data security. Consequently, the dynamic datasets that need to be analyzed are dispersed across various organizations. Furthermore, the sensitivity of privacy and restrictions on access to and processing of medical data impose high demands on the data infrastructure. Patients should be able to easily give consent for the use of data, retract it, and enforce that only data processing approved by them can take place. This is called embedded compliance. The use of data must, therefore, be administered in such a way that the legitimacy of the processing can be ensured. Paola says, "Everyone considers data security and trust in the proper handling of your data important, and EPI ultimately provides a solution for this."

Data Analysis

For the necessary data analyses, the involved PhD candidates have developed (federated) machine learning and statistical techniques alongside new infrastructure and policy management solutions. This is done in close collaboration between the scientific institutions and the hospitals. "Take, for example, the new method 'safe statistics' developed by UMC Utrecht in collaboration with CWI, where you can dynamically increase your sample, thus making continuously up-to-date results available."

Translating Data into Practice

The three hospitals involved each have their own focus and, therefore, face different challenges regarding the collection of the right data. The Princess Máxima Center focuses on data analysis in DIPG (brainstem cancer in children). The number of patients with DIPG is small, and patients are spread across different hospitals and countries with different policies and data consensus. This makes it difficult to draw conclusions about treatment effects. UMC Utrecht focuses on psychiatry. The challenge here is the enormous variability between patients in studies. Also, the process of adjusting patients to the right medication takes (too) long, and in daily practice, the variation between patients is even greater due to the strong selection bias in most clinical trials. The St. Antonius Hospital focuses on data analysis in cerebral infarctions. Patient data is dispersed across different institutions, and a complete overview of the patient data is lacking, making it difficult to predict outcomes (survival rate, functional status, quality of life).

The challenge for hospitals is that these use cases all have their own requirements for the operational analysis environment. It is not scalable to have to manage a different environment for each use case. The infrastructure developed as part of EPI solves this problem. The different requirements from the use cases are implemented using one functional IT architecture? Paola proudly says, "We have now developed a proof of concept environment together, where we will be able to share the results in the final phase of the research."

Putting the Patient at the Center: Digital Health Twin

EPI is seeking a solution to support patients and healthcare providers with an integrated solution. The ideal outcome of the EPI project is a digital 'health twin' for joint and personal management of your data with embedded compliance. This data, if consent is given, is made available for further analysis, creating a health ecosystem that enables increasingly better advice.


EPI offers models that are also applicable on a broader scale, not only for these three hospitals, and certainly not only for the medical sector. "EPI fits well into the general picture of the necessity of data sharing. It is nice to see that the EPI software and results are already being used elsewhere, for example, by AMdEX*," says Paola. "So, EPI has a longer lifespan than just this project." Sander adds, "AMdeX is moving towards an operational environment, and with our results, they have been able to further professionalize this, and hopefully, we can build on this with the results from our proof of concept. Thus, the circle is complete."

* AMdEX is an innovation field lab initiated by AMS-IX, SURF, UvA, DEXES & Amsterdam Economic Board and co-financed by the European Regional Development Fund. In this field lab, parties collaborate to develop and test reliable, fair, and scalable technologies to support the emergence of data markets where its members can freely decide with whom they want to communicate and under what conditions.

For more information and publications, please visit