As digitalization in medicine progresses and more data is captured electronically, the amount of data available increases rapidly. Despite the large amount of data available, the data for specific research questions is much smaller once filtered to match specific study criteria. This, combined with modern biomedicine requiring large numbers of patients, highlights the need to analyze data across institutions. The heterogeneity of modern medical institutions further complicates the analysis process within and across institutions and makes it difficult for researchers to extract meaningful information. The Medical Informatics Initiative (MII) was initiated by the German Ministry of Research and Education (BMBF) to tackle this problem and to harness the potential of digitalization in medicine. Data integration centers (DIC) form the heart of the MII. These organizational units established at each university hospital provide multiple services including harmonization of the data, technical infrastructure, and establishing of governance to support data sharing and use across university hospitals. The DIC implement the standardization and harmonization agreed on across the different MII task forces and committees, like the use of Fast Healthcare Interoperability Resources (FHIR) and specific medical terminologies. The DIC maintain software, which adheres to the agreed upon standards and harmonization rules and provide standardized research data repositories. Therefore, in contrast to a heterogenous hospital, software and data management can be created once and applied to all DIC, if the DIC have implemented the standardized application programming interfaces (APIs). This harmonization is the first step towards paving the way for cross-hospital data analysis. However, to make the data accessible to researchers, make it findable, allow researchers to select subsets of the large data pool for further analysis, allow them to build (federated) statistical or machine learning (ML) models as well as deploy them, additional software tools beyond the standard DIC tools specified above are necessary. Across multiple projects of the MII this thesis investigates how software tools can be designed, built, easily deployed, and integrated into DIC IT infrastructures to support local and federated analyses and statistical model development. It thus extends DIC infrastructures towards providing a cross-institutional research platform. It supports initial feasibility queries as well as data selection, extraction, analysis, and subsequent deployment of statistical models. All software components were conceptualized and implemented by the author of this thesis in collaboration with inter-disciplinary (partly across sites) teams and integrated with existing DIC infrastructures as well as evaluated with exemplary analyses. This thesis first investigated how ML models can be built on gene expression and other patient data to predict clinical outcomes. Different ML models were created, and their accuracy measured. While demonstrating how ML models could be applied to clinical data, the availability of data was found to be a limiting factor and the deployment or use of models remained elusive. This illustrated the need to make harmonized data available across institutions. To adhere to data protection laws, this thesis then investigated and extended DataSHIELD concepts and tools for federated privacy preserving analysis. A new Queue-Poll extension for DataSHIELD was designed and developed, distributed to multiple hospitals, and used for successive research projects to satisfy strict hospital firewall rules. To investigate training and deployment of ML and statistical models a prototype of KETOS, a platform for clinical decision support and machine learning as a service, was conceptualized and implemented. It allows researchers to create ML models based on standardized and harmonized data, while still providing maximum freedom in the model building process. It was conceptualized from the beginning with cross-hospital research in mind and achieved interoperability by relying on the data standardization of the DIC. In the framework of the MII and large research data repositories, a dataset for a specific analysis, model building, and later deployment usually needs to be extracted from the larger dataset requiring methods for data selection and extraction. This was addressed in two further subprojects of this thesis. The first focused on integrating large-throughput genomics data with non-genomic patient data in FHIR format. The second study created a method to select data from a large dataset (30 million FHIR resources) of FHIR formatted patient data, based on inter-criteria relationships. Having established first methods for data extraction, model analysis and deployment, to close the research lifecycle, a method was required to make data across hospitals findable. Therefore, it was investigated how cross-hospital feasibility queries can be created based on harmonized FHIR data and a platform conceptualized. Further, as part of the concept and implementation of the feasibility query platform, it was investigated how the necessary ontology for the user interface could be automatically generated using FHIR implementation guides, profiles, and a terminology server. Within the CODEX project the platform was implemented and distributed to 34 participating university hospitals. The platform performed well with large data volumes, searching through four mil. resources within 30 seconds. While the two attempts for data extraction investigated as part of this thesis were applicable to some data extraction problems, a more wholistic method for data extraction is still missing. To solve this, this thesis’ work on cohort selection of the feasibility query could be combined with a feature selection to create a cohort specific data extraction process. This thesis successfully demonstrates how software can be built, which leverages the data integration efforts of the MII to support large scale cross-hospital data sharing and analysis projects. The innovative software developed here was successfully integrated with the DIC infrastructure and highlights the importance of the standardization and harmonization efforts of the DIC and MII. All developed software was created to not only be applicable to one institution but created with cross-institutional research with the DIC at the core in mind, making a valuable contribution to the MII and the future of medical research infrastructures.