A data ecosystem for personalized medicine
Swiss Personalized Health Network (SPHN)
Infrastructure for nationwide use and exchange of health data for research.
Translation of an article by Ines Böhm in the Schweizerische Ärztezeitung, Issue 2023/43, published on 25.10.2023. Schweiz Ärzteztg. 2023;104(43):18-21.
Research
Health data from Swiss university hospitals was previously not standardized, which made secondary use
Personalized medicine is based on large amounts of data. This includes health-related data from clinical practice, data from genomics, transcriptomics, proteomics and metabolomics ("omics") as well as data from registries, studies and cohorts. By storing, linking and analyzing this data, the aim is to enable more precise diagnoses and more individualized therapies, identify risks for certain diseases and achieve early prevention [1]. This means that medical care is no longer based solely on average values for specific populations, but also takes individual patient characteristics into account.
The challenge of health data
The Swiss Personalized Health Network (SPHN) initiative was launched by the federal government in 2016 to create the essential prerequisites for data-driven personalized medicine in Switzerland [2]. The network includes university hospitals, universities, universities of applied sciences and the Swiss Federal Institutes of Technology ETH and EPFL.
At the beginning of the initiative, the data landscape of the participating institutions was extremely heterogeneous.
"At the beginning of the initiative, the data landscape of the participating institutions was extremely heterogeneous, both in terms of the maturity of data management and the use of generally understandable data standards," says Dr Thomas Geiger, Managing Director of SPHN. In addition to legal, ethical and technical aspects, the exchange of healthcare data is also a cultural challenge. Data protection is often valued more highly than the potential benefits. Competing hospitals and doctors have their own interests and want to skim off the value of the collected data for themselves first. Swiss "cantonalism" is also perceived by various network participants as being less favorable to data exchange. However, a broad-based exchange of data is essential for medical research, if only because of Switzerland's small size.
Data standards and services
In the two funding periods since 2017, the network has been able to "create a mutual understanding of the benefits and added value of shared data and harmonized processes. Common visions for data exchange have been developed," says Katrin Crameri, Director of the SPHN Data Coordination Center in Basel. She continues: "Specifically, standards for naming concepts and data, and precise principles for their interpretation have been defined for over 100 clinical parameters, based on international data standards. There is agreement here with eHealth Suisse, as structured, standardized data is also required for the electronic patient dossier (EPD)."
A great deal of complex data is collected when monitoring patients with serious infections.
Today, the sensitive data can be fed into the high-security BioMedIT network via secure data pipelines with double encryption, says Katrin Crameri. "They can be processed on high-performance computers, but always remain in the Trusted Research Environment. This is crucial in order to maintain the trust of the hospitals involved and ultimately also of the patients who provide their data," emphasizes Katrin Crameri. Essential services are available via the Data Coordination Centre (DCC): "On the one hand, this is a whole portfolio of software tools that massively facilitate the integration and maintenance of data from a wide variety of sources (e.g. imaging, laboratory, gene sequences, text, etc.), systems and organizations." This also includes a query system that makes it possible to analyze the feasibility of a research project based on existing data across five hospitals. Non-technical services are also important, for example when it comes to modelling ethical or legal issues. The infrastructures, standards and services created by SPHN are based on the experience gained from around 40 funded applied research and development projects [3].
Four national data platforms
The National Data Streams (NDS), jointly funded by SPHN and the program of the ETH domain PHRT - Personalized Health and Related Technologies , are intended to lay the foundation for sustainable concepts in the field of personalized health research using the existing foundations in SPHN. This is also expected to have a signaling effect for follow-up projects. In future, data for research in intensive care medicine, oncology, pediatrics and healthcare research will be available on four platforms [3]. "Four consortia with researchers from the institutions involved in SPHN maintain the data in accordance with the SPHN interoperability strategy. In addition to their own research within the frame of a lighthouse project, the data can also be used for third-party projects," says Katrin Crameri.
Early Recognition of serious infection
One of these NDS deals with severe infections. These exhibit great phenotypic heterogeneity, which has different effects on the clinical course and treatment success. Patients with severe infections are monitored very closely in intensive care. A great deal of data is collected, some of it at high temporal resolution and very complex. An analysis of this data from the five university hospitals, combined with clinical parameters to assess the infections, allows a more individualized assessment, a more precise characterization and a better prognosis for patients [4].
"We can draw on the existing infrastructure of a Clinical Data Warehouse from a previous project," says Prof. Dry Adrian Egli, University of Zurich, project leader of the NDS "Personalized, data-driven prediction and assessment of infection related outcomes in Swiss ICUs" (IICU). Some of the project's work packages are already at an advanced stage. The most progress has been made in the definition of infections in intensive care medicine, and good progress has also been made in the creation of electronic patient documentation and bioinformatic analyses. Egli sees the legal requirements with contracts between many project partners and the clarification of ethical issues as a major challenge. This is time consuming and the existing data can only be analyzed once these issues have been clarified. Egli sees great potential for personalized medicine in the validation of digital biomarkers, which could be used to detect sepsis and other serious infectious diseases earlier.
Precision medicine in oncology
Precision oncology opens up completely new perspectives for cancer patients. Clinical data, complex omics analyses and images from pathology and radiology are evaluated using artificial intelligence in order to treat patients with personalized therapies. "High-quality and interoperable data is crucial for both our clinical and research programs in the field of precision oncology. With SPHN, we have access to a first-class infrastructure to share our data while maintaining the highest quality and security standards," says Prof. Olivier Michielin, MD, from Geneva University Hospital, who is the project leader of the Swiss Personalized Oncology-NDS project (SPO-NDS) [5] together with Prof. Bernd Bodenmiller, MD, ETH Zurich.
The aim of a national registry is to identify overtreatment and its consequences for patients.
A test phase is currently underway to evaluate the benefits of omics approaches for the molecular tumor board. "It is a major challenge to generate and standardize omics data throughout Switzerland and to make this information available in a molecular tumor board within just two to three weeks. Thanks to SPHN and the outstanding team spirit in the SPO-NDS project, this is now becoming a reality," says Michielin happily. The first patient is expected in early 2024, heralding a new phase of precision oncology in Switzerland.
Personalized medicine in pediatrics
With the SwissPedHealth-NDS, a data platform is to be created on which routine clinical data from large Swiss children's hospitals and university children's hospitals can be stored and processed [6]. "In pediatrics, it is particularly important to learn from data. Children are the most vulnerable patients and can benefit enormously. Every improvement in pediatric care has an impact on society as a whole," says Prof. Luregn Schlapbach, MD, University Children's Hospital Zurich, PI of the SwissPedHealth project. The NDS consists of four sub-projects in the areas of growth and obesity, oncology, pneumology and antibiotic use in infectious diseases.
Precision oncology opens up completely new perspectives for cancer patients.
Schlapbach emphasizes: "When we talk about personalized medicine in pediatrics, we are usually talking about rare diseases that often only affect individual children. In order to prevent severe progressions, these diseases must be recognized early." As part of the associated lighthouse project, multi-omics data is collected for rare pediatric diseases in parallel with the usual diagnostic procedures. Together with the use of machine learning
Optimizing care and medical processes
In contrast to performance indicators such as mortality, readmissions or length of stay, nursing care and medical procedures in Swiss hospitals have so far hardly been documented and monitored at national level, if at all. With the NDS "Low Value Care in Medical Hospitalized Patients" (LUCID), such a monitoring instrument can be set up in the five Swiss university hospitals. This national register is intended to identify overtreatment, its development over the last decade and the clinical consequences for patients. Based on these findings, concrete improvements in the quality of care can be achieved [6]. "However, this depends on public and political will," explains PD Dr Marie Méan, Centre Hospitalier Universitaire Vaudois. She is one of the project leaders and adds: "The collaboration within the network works very well. A modular, SPHN-compliant IT architecture for LUCID, which is also available for other applications, was developed in a short space of time. The first data feed into BioMedIT will take place this autumn. These are data from 100,000 patients who were hospitalized between 2014 and 2023." The first results are expected in 2024. The administration of blood transfusions will be the first medical process to be analyzed in order to ensure that blood reserves are used sparingly.
Translated with kind permission of the EHM Schweizerischer Ärzteverlag AG. Original article (Schweiz Ärzteztg. 2023;104(43):18-21) available at: Personalisierte Medizin | saez.ch (swisshealthweb.ch)
References