The choice and design of HiGHmed’s three medical use cases were based on the rationale that each of them would prototypically allow HiGHmed to address a variety of specific challenges and requirements in Medical Informatics, as described above. In addition, it reflects the complementary strength of the partner sites, thus providing an ideal basis for collaboration.
The oncology use case will help us to address challenges in Medical Informatics when integrating omics data from genome sequencing and radiology into clinical practice. New, mobile diagnostic devices are expected to change current medical practice and research by contributing to the long-term monitoring of personal health data at an unprecedented level.
Within the cardiology use case, we will systematically explore and address IT challenges related to the integration of data from wearable and connected devices into our IT architecture.
The infection control use case will develop an automated early warning and cluster analysis system to support the algorithmic detection of pathogen clusters. It will include multidrug- resistant organisms within and across university hospitals, the verification of whether clusters represent outbreaks, and the identification of possible causes of outbreaks. This motivates us to address IT challenges when connecting and analyzing multiple clinical and organizational data sources in real time for establishing spatial, temporal, biological and functional associations. Importantly, our technology platforms will be developed in a generic way in order to fully support these three, as well as various other, medical use cases. All the partners will therefore benefit from this integrated technology platform with immediate added value for the three communities of clinicians, patients and researchers.
- Use case 1: Oncology
Oncology, in particular, relies on multidisciplinary collaborations and multimodal treatments that involve different medical specialties. Problems arise because the data necessary for rapid diagnosis or treatment decisions is distributed across an array of heterogeneous IT systems. Data integration thus holds huge promise for improving clinical care, clinical trial planning as well as basic and translational cancer research. At the same time, research findings in multiple cancer entities suggest that certain cancer entities are composed of subgroups that share similar molecular patterns (e.g. Waddell et al., 2015 Nature) and hence require individual treatment strategies.
Multi-center collaborations will be necessary for both molecularly based tumor stratification and the selection of therapeutic strategies, as well as the evaluation of patient outcome, and in order to power clinical trials appropriately.
We chose to demonstrate the potential of data integration for cancers of the pancreas, liver and biliary tract (hepatopancreatobiliary cancer) as well as head-and-neck carcinoma, for which the collaborating centers traditionally hold great expertise in clinical as well as basic research. Liver cancer has already been the topic of a DFG funded Transregio project between UKHD, DKFZ and MHH3, which set the stage for highly relevant aspects in the oncology use case. The tumors selected in the oncology use case represent the complete array of cancer etiologies and all require intense, interdisciplinary diagnostics and treatment, which includes surgery, interventional radiology and gastroenterology, radiotherapy, systemic treatment, as well as multimodal supportive care. Future therapeutic strategies will aim to achieve an IT-based integration of tumor and patient characteristics into personalized therapeutic decisions. In head-and-neck cancer, radio-oncological modalities play an important role; this entity is paradigmatic to show the benefit of integrating imaging, radiation planning and radiotherapeutic data with clinical, pathological and molecular information.
- Use case 2: Cardiology
Heart failure (HF) is the disease requiring the most hospitalizations in Germany and the Western World. Its prevalence is estimated to be 2% in the general population, rising to 10% in the age group over 75 years. Due to the advanced treatment of cardiovascular diseases which cause heart failure and to longer overall life expectancy, the prevalence of HF is steadily increasing, and it is also referred to as the epidemic of the 21st century. HF is therefore of outstanding health-economic relevance. However, despite an improvement in the treatment of HF, the prognosis for HF patients is still poor. Importantly, each hospitalization due to a deterioration of HF by itself markedly worsens the prognosis. Biomedical approaches to detect early or to prevent the decompensation of HF can therefore improve the prognosis and reduce costs.
This use case will apply a) clinical variables from hospital information systems, b) patient-reported data (e.g. symptoms), c) imaging data, d) sensor technology (e.g. sensors for pulmonary pressure or burden of cardiac arrhythmias), and e) cardiovascular biomarkers to improve outcome. Until now, there has been limited evidence from smaller, mostly single-center studies that the aggregate of these measures can further improve the treatment of HF. Hence, there are several major problems to overcome in order to fully utilize the technologies on a broad basis for patient treatment: Collection of valid, interpretable data from multiple centers, integration of that data via international standards into research trials and clinical management, and computational models to predict individual outcomes.
As midterm perspective, we will generate a harmonized dataset from selected clinical variables (e.g. patient-reported symptoms, blood pressure, heart rate), imaging modalities (e.g. LV-EF, global longitudinal strain analysis, LA size) and molecular biomarkers (e.g. NT-proBNP, troponin, ferritin) of all hospitalized patients (retrospective, n>10,000) by retrieving HIS data sets. Existing risk models for predicting a deterioration of HF will be applied and high-risk individuals identified. From the comprehensive data set, pattern recognition techniques and statistical frameworks will be applied to identify (linear) predictors for repeated hospitalization. A clinical trial to modify pharmacological or device treatment in high-risk patients will be designed.
- Use case 3: Infection control
Healthcare-associated infections (HAI) and multidrug-resistant organisms (MDRO) are a major cause of excess patient morbidity and mortality, and cause significant added expenses for hospitals. Failures in the early recognition of pathogen clusters may lead to outbreaks that are difficult to contain. Clusters often remain undetected since transfers of patients occur frequently and the microbiological detection of phenotypically similar pathogens in different patients at a later stage might not be considered to be epidemiologically connected. Although the information required to create epidemiological links may, in principle, be available in different IT subsystems, the lack of connectivity between systems currently does not permit timely cluster recognition. Automated detection systems are urgently needed to facilitate the prevention and containment of nosocomial transmission.
The main aim of this use case is to establish an automated early warning and cluster analysis system to support a) the algorithmic detection of pathogen clusters (including MDRO) within and across university hospitals, b) the verification of whether clusters represent outbreaks, and c) the identification of possible causes of outbreaks. 'Multiple clinical and organizational data sources need to be integrated and analyzed in a timely manner for spatial, temporal, biological and functional associations (challenge: interoperability). Machine learning will be applied to detect clusters of colonization or infection earlier than by conventional means. Furthermore, such an infrastructure will support the quantitative evaluation of the effectiveness of preventive measures, and will thus contribute to improved prevention across institutions.'