Verfügbare Projekte
... für Studierende der Informatik
Es sind mehrere Masterprojekte für Studierende der Informatik verfügbar.
- Ein Einstieg in das Projekt ist ab sofort bis einschließlich dem 1. Februar 2023 möglich.
- Die Bewerbung erfolgt über das DigiStrucMed-Programm.
- Bei Interesse an den Projekten sowie bei inhaltlichen Fragen wenden Sie sich bitte direkt an die jeweilige Projektleitung.
P01 - (Semi-) Automated procedures to find arguments for bioethical analyses
Ethical problems are frequently encountered by physicians and other healthcare personnel in clinical practice and in the research context. Technological progress as well as political, legal and social developments further increase the complexity of ethical challenges. Whereas bioethical analyses traditionally draw on ethical theory and socio-empirical research (e.g. surveys or qualitative interviews), only recently methods of advanced data science have been introduced in this domain. For example, decision support systems for, e.g., ethical organ allocation are developed. Furthermore, digital moral games are set up for probing participants’ moral intuitions and behavior. Not least, bioethics ontologies are developed as platforms that can be used by scientific and healthcare communities. Open questions, however, exist regarding the methodological quality of “Digital Bioethics” and also with respect to human-machine-interaction, responsibility and the future formation of the academic field. To provide proper guidance for healthcare personnel’s decisions in clinical practice, the capabilities as well as limitations of “Digital Bioethics” methods must be reliable and consistent.
Apart from abstract ethical theories, it is usually the real world practice determining actual standards that in turn take normative effect in everyday situations. This is reflected in lines of argument in research papers reasoning about methodologies or experiments, but also in factual descriptions of actions performed in analogous cases. Recently, argumentation mining as a new task in information extraction has also received a lot of attention in computer science. “Digital Bioethics” in turn can profit from argumentation mining when it comes to orient oneself in bioethical debates. The idea is to create machines that can reliably discover arguments for and against some topic and thus are enabled to intelligently engage in debates, often referred to as autonomous debating systems (see e.g., IBM‘s debater project, research.ibm.com/interactive/project-debater/). The idea of this project is to mine bioethical arguments from current research publications and to determine the usability of the produced body of moral standards in terms of consistency, completeness, etc. Hence, the research question is whether the results derived from “Digital bioethics” methods are congruent to results derived from established methods.
Projektleitung:
Institute of Ethics, History and Philosophy of Medicine
Hannover Medical School
P03 - Deciphering long-term consequences of HCV infection after viral elimination (Long-HepC) using Omics Integration
Long-term consequences of infectious diseases are an important research topic. Even when the disease itself is cured, risks or impairments remain for patients, as currently demonstrated by Long- COVID, but also by long-term consequences of HCV. However, it is particularly difficult to explore the molecular basis of the long-term consequences of infections. Several studies have demonstrated the importance of metabolomics, epigenetics, and cytokine measurements in understanding HCV infection. Even after the disease is cured, there are persistent molecular changes that are poorly understood. To understand such changes, it is necessary to obtain and analyze molecular data over several time points. Unfortunately, such analyses require more advanced statistical methods then most tools offer to date. While appropriate approaches are available in dedicated statistical programming languages, such as R, they are hard to handle for biomedical experts without intensive training in statistics and computer science. We have access to well-characterized cohorts of HCV patients and a wealth of epigenetic, immunoproteomic (cytokines), and metabolomic data. Healthy individuals as well as inactive HBsAg carriers will serve as controls. We will analyze these data and use them to extend user-friendly analysis tools by the capability to handle advanced statistical models for analyzing repeated-measurement epigenetics data. This will provide access to complex statistical approaches for biomedical experts and allow leveraging the full potential of molecular data available across multiple time points. In particular, we hope to gain novel insights into the epigenetics of long-term HCV infection.
Projektleitung:
Centre for Individualised Infection Medicine (CiiM), Hannover
Dept. of Gastroenterology, Hepatology and Endocrinology
Hannover Medical School
P07 - Development of Translational Scores for a Quantitative and Time-Dependent Assessment of Disease Severity in Patients
Clinical studies and patient monitoring produce a lot of data on various time scales. However, crucial conditional patient information is often hidden in large amounts of correlated variables, requiring decision-makers to rely on macroscopic changes in well-being, aggregated data, and traditional indicators to assess the patients’ status. With the help of Data Science, the project aims to develop multivariate concepts using various techniques from Statistics, Machine Learning, and Computational Modeling to find, assess and analyze the most critical variables in clinical studies and/or real-time data. Further, it aims to develop strategies of merging these findings into a representative metric with variable time-resolution, e.g., to obtain early estimates on condition changes in the kidney.
We hypothesize that strategies of quantitative measurement of well-being can be translated to the human scale. This approach will help with the identification of risk factors and will objectify diagnostics in the assessment of the patients’ status. In addition, it will further monitor patients over time while being sensitive towards changes in the input variables. This requirement is essential for developing medical Apps and monitoring patients with chronic diseases. To achieve this, we plan on translating our recently developed RELSA (Relative Severity Assessment) methodology to a use-case on the human scale. With this algorithm, disease models can be monitored over time, in groups or as individuals, using multiple input variables. Further, different diseases can be compared and ranked in terms of severity.
Therefore, in an interdisciplinary approach at the MHH, and over 10 (+2) months, the successful medical/informatics candidate will retrieve and curate time-series data (e.g., blood pressure plus clinical parameters and covariates) from the children’s hospital. Consequently, data wrangling and medicine skills are required to identify and assess suited clinical models for the project. Finally, the candidate will help develop the score using Data Science Tools like the R software. Therefore, the project is suited for data-centered medical or informatics students interested in the clinical monitoring of patients, mathematical modeling, and artificial intelligence.
Projektleitung:
Institut für Versuchstierkunde
Medizinische Hochschule Hannover
P09 - Next generation Flow-Cytometry to quantify and characterize the leukemic stem cell population in acute myeloid leukemia (Folgeprojekt)
Despite substantial progress in the understanding of the underlying biology of acute myeloid leukemia (AML), treatment of AML remains a major therapeutic challenge with long-term survival under 30%. Responsible for this are residual leukemic stem cells that persist in the bone marrow after chemotherapy and give rise to relapse after attaining morphological complete remission (CR). Finding and tracking these residual leukemic (stem) cells (“minimal residual disease”/MRD) is key to guide AML therapy and has dramatic prognostic impact on relapse incidence and patient survival. MRD can be measured by conventional multicolor flow cytometry (Flow-MRD, 8-10 colors), PCR, or next generation sequencing. Flow-MRD relies on the aberrant expression of surface marker combinations on leukemic blasts (leukemia-associated immunophenotypes - LAIPs), which are not found on normal hematopoietic cells. Hence, the challenge of Flow-MRD is to detect and track small populations of leukemic stem cells against the complex and heterogenous background of human bone marrow. With the development of next-generation Spectral Flow Cytometers, it is now possible to label cells with mixtures of >20 different antibody conjugates at once, which will allow to monitor the cellular composition of leukemia at unprecedented resolution. However, Spectral Flow Cytometry gives rise to large, high-dimensional datasets which are increasingly difficult to analyze by manual gating. Therefore, frameworks for computational flow-cytometry have to be developed to analyze this type of data with the aim to automatically deconvolute and annotate complex mixtures of cells and discover novel cell populations. In this project we will leverage our expertise in flow cytometry, bioinformatics and -omics techniques to develop an integrated pipeline that combines Spectral- and Computational Flow Cytometry for improved detection of minimal residual disease in acute myeloid leukemia. This project represents an exciting new field at the intersection of hematology and computational biology and will be performed as a joint effort between an MD-student, who will perform conventional and spectral flow cytometry on bone marrow samples and an informatics master student who will develop the computational pipeline to analyze high dimensional Flow-MRD datasets.
Projektleitung:
Hematology, Hemostasis, Oncology and Stem Cell Transplantation, Institute of Experimental Hematology
Hannover Medical School
P10 - Towards identifying genetic determinants of antibiotic resistance in clinical Pseudomonas aeruginosa, Escherichia coli and Klebsiella pneumoniae isolates
Antibiotic therapy is a cornerstone of treatment in virtually all medical divisions. However, increasing antibiotic resistance poses a great risk for successful treatment outcomes and patient safety. Today, antimicrobial therapy is mostly based on phenotypic resistance testing (antibiograms). We aim to contribute to the digital revolution in microbiology diagnostics by integrating antimicrobial resistance profiles of a large sample size of clinical isolates with bacterial genomics and transcriptomics data in order to develop new guidance for antibiotic treatment.
We started a study in October 2020, collecting and sequencing every bacterial strain of the species Pseudomonas aeruginosa, Escherichia coli and Klebsiella pneumoniae isolated from patients treated in Hannover Medical School (over 12,000 bacterial isolates already collected). We will concentrate on stratifying clinical isolates into groups of isolates exhibiting distinct phenotypic bacterial resistance profiles and will make use of a graph-based framework to enable genotype-phenotype correlation studies. By using publicly available AMR marker databases for resistance prediction, we expect to generate and validate lists of genetic determinants that explain resistance phenotypes of our clinical isolate collection. We will furthermore link an existing pan-genome graph-database developed in the Institute of Molecular Bacteriology, with the phenotypic resistance data for facilitated and robust identification of novel genomic determinants that reliably predict antibiotic resistance in clinical isolates.
Projektleitung:
Prof. Dr. med. Susanne Häußler
Institute for Molecular Bacteriology, TWINCORE and
Helmholtz Centre for Infection Research (HZI)