Biological cells and tissues consist of thousands of different molecules and processes, and often only one or a few of them are altered in the case of disease.

Smart sensors are used to filter out this disease- or process-relevant information, specific signatures or molecules from a background of countless irrelevant elements, and display it to the treating physician or drug-seeking scientist.

In this context, the distinction is best made in real time to enable rapid decision making.

Typical examples of smart sensors include smart biopsy needles that detect disease-specific tissue signatures and molecular biosensors which, combined with appropriate peripherals, transmit relevant tissue or activity changes in a patient's body or in cellular test models.

As a rule, smart sensors use complex data analysis methods, often from the field of artificial intelligence, to further process the primary data received for interpretation.

More ...

Under construction

Under construction

Under construction

Due to increasing digitization in healthcare, we have a growing pool of structured data on disease progression. In addition to continuous data collection in inpatient care, the electronic patient record (ePA) and self-documentation with apps and wearables are also making data available that allow statements about environmental and risk factors as well as long-term outcomes. UMM's strategic focus on cross-sector, cross-regional collaboration with ambulatory care allows data to be collected across the entire patient journey, harmonized in interoperable formats, and used to optimize healthcare. A major focus here is the development and evaluation of app-based clinical decision support systems. At the Medical Systems Technology research focus, data from outpatient and inpatient care as well as self-documentation and sensor-based measurement with wearables will be processed using artificial intelligence methods to develop disease models that allow conclusions to be drawn about subtypes, expected courses and therapy options. The implementation is carried out in close cooperation between the Center for Preventive Medicine and Digital Health (CPD) and the Mannheim Institute for Intelligent Systems in Medicine (MIiSM) and actively involves the clinical departments of the UMM.

More ...

One major application of artificial intelligence is the extraction of knowledge from clinical data.

Pretherapeutic prognoses can be made with this knowledge, or diagnostic or therapeutic support can be provided.

Starting with clinical data with all its shortcomings such as missing data points, errors or unclear statements from physicians' letters, the challenge is to transform these into structured, quality-controlled data. If these data are available in a structured form, classical or modern methods of machine learning can be used, on the one hand as association, as qualified statement with uncertainty, or as causal inference.

Typical examples are the automatic segmentation and classification of histopathological data, the use of radiomics for diagnostics or the management of missing data by means of multiple imputation in combination with a digital twin.

More ...

Under construction

The essential basis for targeted therapies with innovative drugs is molecular characterization and tumor subclassification in the pharmacological context of action. Among tumor patients, the metastatic state requires minimally invasive tissue sampling via percutaneous, image-guided, needle-based biopsy. The temporal and spatial clonal malignant evolution of a cancer generates molecular heterogeneity between individual metastases as well as within individual lesions. To date, tumor resistance mechanisms cannot be consistently detected in individual tumor areas because the choice of location for tissue biopsy primarily follows only the criteria of optimal technical feasibility, avoidance of patient-burdening multiple biopsies, and best visualization in conventional computer tomography (CT) scans. In addition, metastases require a relevant size of at least approximately 1 cm in routine radiological diagnostics with CT and magnetic resonance imaging before a precise image-guided biopsy can be performed. Thus, despite the availability of innovative targeted drugs, the patient's drug treatment schedule is de facto not adapted to the individual metastatic and resistance behavior of the disease.

With the introduction of intelligent processes, all diagnostic information relevant to the site of metastasis and tumor heterogeneity – from circulating tumor DNA to molecular imaging to image post-processing with radiomics – can be combined and viewed integratively and longitudinally as a digital fingerprint. Using artificial intelligence, proactive metastasis localization and clonal evolution of individual tumor sites can be simulated on the basis of iterative diagnostic and treatment cycles, and thus the location of optimal tissue extraction for individual adaptation of pharmacotherapy can be determined. At the same time, the patient is in a four-dimensional digital coordinate system, based on which state-of-the-art robotic-assisted technologies can simultaneously position biopsy needles precisely, gently, and time-efficiently along radiomics parameter images. By coupling robotic-assisted biopsy with newly developed smart multispectral needles that employ fast spectroscopic techniques at the tip, the exact location of tumor localization can be verified again during the biopsy itself to increase precision and support the metastasis simulation model.

Thus, state-of-the-art robotic assistance systems adaptively respond to changes in tumor spread, tumor composition, and metastatic behavior based on innovative processes supported by artificial intelligence.

More ...

Mannheim Advanced operative & Robotic Science (MARS)

The joint surgical center of the Department of Surgery and the Department of Urology and Urosurgery at the University Hospital Mannheim enables the transfer of new robotic technologies into clinical practice by testing new robotic surgical assistance systems. The digital connection of modern operative telemanipulators to the operating room has several future-oriented potentials. The primary focus is to reduce the surgical risk for a patient and thus improve the outcome. High-precision movements and superhuman perception through the technology of state-of-the-art telemanipulators (OR-robots) aim to make this possible. The development of assisting imaging technologies provides the surgeon with additional information at the right time, thus supporting complicated surgical decisions through machine learning. An example of this is the digital overlay of preoperative imaging (e.g., CT or MRI) with the patient's situs. Additional sensor technology and processing of complex data expand the capabilities of the operating surgeon. They also create increased workplace safety through improved ergonomics. AI-based documentation and analysis of intraoperative findings additionally contribute to individualized control of peri- and postoperative processes. Alongside the purely intraoperative applications, the preoperative and postoperative data management of the patients will also be integrated.

In short, the goal of MARS is to connect the emerging digital AI-driven capabilities with advanced medical devices in order to integrate the future into everyday clinical practice today.

More ...