Refine
Document Type
- Article (19)
- Working Paper (4)
- Bachelor Thesis (1)
- Conference Proceeding (1)
- Master's Thesis (1)
- Preprint (1)
- Report (1)
Has Fulltext
- yes (28) (remove)
Is part of the Bibliography
- no (28) (remove)
Keywords
- machine learning (28) (remove)
Institute
- Medizin (9)
- Wirtschaftswissenschaften (6)
- Center for Financial Studies (CFS) (4)
- Biochemie, Chemie und Pharmazie (3)
- Buchmann Institut für Molekulare Lebenswissenschaften (BMLS) (2)
- Frankfurt Institute for Advanced Studies (FIAS) (2)
- Psychologie (2)
- Biochemie und Chemie (1)
- Biodiversität und Klima Forschungszentrum (BiK-F) (1)
- Biowissenschaften (1)
We assemble a data set of more than eight million German Twitter posts related to the war in Ukraine. Based on state-of-the-art methods of text analysis, we construct a daily index of uncertainty about the war as perceived by German Twitter. The approach also allows us to separate this index into uncertainty about sanctions against Russia, energy policy and other dimensions. We then estimate a VAR model with daily financial and macroeconomic data and identify an exogenous uncertainty shock. The increase in uncertainty has strong effects on financial markets and causes a significant decline in economic activity as well as an increase in expected inflation. We find the effects of uncertainty to be particularly strong in the first months of the war.
Human functional brain connectivity can be temporally decomposed into states of high and low cofluctuation, defined as coactivation of brain regions over time. Rare states of particularly high cofluctuation have been shown to reflect fundamentals of intrinsic functional network architecture and to be highly subject-specific. However, it is unclear whether such network-defining states also contribute to individual variations in cognitive abilities – which strongly rely on the interactions among distributed brain regions. By introducing CMEP, a new eigenvector-based prediction framework, we show that as few as 16 temporally separated time frames (< 1.5% of 10min resting-state fMRI) can significantly predict individual differences in intelligence (N = 263, p < .001). Against previous expectations, individual’s network-defining time frames of particularly high cofluctuation do not predict intelligence. Multiple functional brain networks contribute to the prediction, and all results replicate in an independent sample (N = 831). Our results suggest that although fundamentals of person-specific functional connectomes can be derived from few time frames of highest connectivity, temporally distributed information is necessary to extract information about cognitive abilities. This information is not restricted to specific connectivity states, like network-defining high-cofluctuation states, but rather reflected across the entire length of the brain connectivity time series.
Analysis of machine learning prediction quality for automated subgroups within the MIMIC III dataset
(2023)
The motivation for this master’s thesis is to explore the potential of predictive data analytics in the field of medicine. For this, the MIMIC-III dataset offers an extensive foundation for the construction of prediction models, including Random Forest, XGBOOST, and deep learning networks. These models were implemented to forecast the mortality of 2,655 stroke patients.
The first part of the thesis involved conducting a comprehensive data analysis of the filtered MIMIC-III dataset.
Subsequently, the effectiveness and fairness of the predictive models were evaluated. Although the performance levels of the developed models did not match those reported in related research, their potential became evident. The results obtained demonstrated promising capabilities and highlighted the effectiveness of the applied methodologies. Moreover, the feature relevance within the XGBOOST model was examined to increase model explainability.
Finally, relevant subgroups were identified to perform a comparative analysis of the prediction performance across these subgroups. While this approach can be regarded as a valuable methodology, it was not possible to investigate underlying reasons for potential unfairness across clusters. Inside the test data, not enough instances remained per subgroup for further fairness or feature relevance analysis.
In conclusion, the implementation of an alternative use case with a higher patient count is recommended.
The code for this analysis is made available via a GitHub repository and includes a frontend to visualize the results.
he most basic behavioural states of animals can be described as active or passive. While high-resolution observations of activity patterns can provide insights into the ecology of animal species, few methods are able to measure the activity of individuals of small taxa in their natural environment. We present a novel approach in which a combination of automatic radiotracking and machine learning is used to distinguish between active and passive behaviour in small vertebrates fitted with lightweight transmitters (<0.4 g).
We used a dataset containing >3 million signals from very-high-frequency (VHF) telemetry from two forest-dwelling bat species (Myotis bechsteinii [n = 52] and Nyctalus leisleri [n = 20]) to train and test a random forest model in assigning either active or passive behaviour to VHF-tagged individuals. The generalisability of the model was demonstrated by recording and classifying the behaviour of tagged birds and by simulating the effect of different activity levels with the help of humans carrying transmitters. The model successfully classified the activity states of bats as well as those of birds and humans, although the latter were not included in model training (F1 0.96–0.98).
We provide an ecological case-study demonstrating the potential of this automated monitoring tool. We used the trained models to compare differences in the daily activity patterns of two bat species. The analysis showed a pronounced bimodal activity distribution of N. leisleri over the course of the night while the night-time activity of M. bechsteinii was relatively constant. These results show that subtle differences in the timing of species' activity can be distinguished using our method.
Our approach can classify VHF-signal patterns into fundamental behavioural states with high precision and is applicable to different terrestrial and flying vertebrates. To encourage the broader use of our radiotracking method, we provide the trained random forest models together with an R package that includes all necessary data processing functionalities. In combination with state-of-the-art open-source automated radiotracking, this toolset can be used by the scientific community to investigate the activity patterns of small vertebrates with high temporal resolution, even in dense vegetation.
The KMT2A (MLL) gene rearrangements (KMT2A-r) are associated with a diverse spectrum of acute leukemias. Although most KMT2A-r are restricted to nine partner genes, we have recently revealed that KMT2A-USP2 fusions are often missed during FISH screening of these genetic alterations. Therefore, complementary methods are important for appropriate detection of any KMT2A-r. Here we use a machine learning model to unravel the most appropriate markers for prediction of KMT2A-r in various types of acute leukemia. A Random Forest and LightGBM classifier was trained to predict KMT2A-r in patients with acute leukemia. Our results revealed a set of 20 genes capable of accurately estimating KMT2A-r. The SKIDA1 (AUC: 0.839; CI: 0.799–0.879) and LAMP5 (AUC: 0.746; CI: 0.685–0.806) overexpression were the better markers associated with KMT2A-r compared to CSPG4 (also named NG2; AUC: 0.722; CI: 0.659–0.784), regardless of the type of acute leukemia. Of importance, high expression levels of LAMP5 estimated the occurrence of all KMT2A-USP2 fusions. Also, we performed drug sensitivity analysis using IC50 data from 345 drugs available in the GDSC database to identify which ones could be used to treat KMT2A-r leukemia. We observed that KMT2A-r cell lines were more sensitive to 5-Fluorouracil (5FU), Gemcitabine (both antimetabolite chemotherapy drugs), WHI-P97 (JAK-3 inhibitor), Foretinib (MET/VEGFR inhibitor), SNX-2112 (Hsp90 inhibitor), AZD6482 (PI3Kβ inhibitor), KU-60019 (ATM kinase inhibitor), and Pevonedistat (NEDD8-activating enzyme (NAE) inhibitor). Moreover, IC50 data from analyses of ex-vivo drug sensitivity to small-molecule inhibitors reveals that Foretinib is a promising drug option for AML patients carrying FLT3 activating mutations. Thus, we provide novel and accurate options for the diagnostic screening and therapy of KMT2A-r leukemia, regardless of leukemia subtype.
Biased auctioneers
(2022)
We construct a neural network algorithm that generates price predictions for art at auction, relying on both visual and non-visual object characteristics. We find that higher automated valuations relative to auction house pre-sale estimates are associated with substantially higher price-to-estimate ratios and lower buy-in rates, pointing to estimates’ informational inefficiency. The relative contribution of machine learning is higher for artists with less dispersed and lower average prices. Furthermore, we show that auctioneers’ prediction errors are persistent both at the artist and at the auction house level, and hence directly predictable themselves using information on past errors.
Background: The categorization of individuals as normosmic, hyposmic, or anosmic from test results of odor threshold, discrimination, and identification may provide a limited view of the sense of smell. The purpose of this study was to expand the clinical diagnostic repertoire by including additional tests. Methods: A random cohort of n = 135 individuals (83 women and 52 men, aged 21 to 94 years) was tested for odor threshold, discrimination, and identification, plus a distance test, in which the odor of peanut butter is perceived, a sorting task of odor dilutions for phenylethyl alcohol and eugenol, a discrimination test for odorant enantiomers, a lateralization test with eucalyptol, a threshold assessment after 10 min of exposure to phenylethyl alcohol, and a questionnaire on the importance of olfaction. Unsupervised methods were used to detect structure in the olfaction-related data, followed by supervised feature selection methods from statistics and machine learning to identify relevant variables. Results: The structure in the olfaction-related data divided the cohort into two distinct clusters with n = 80 and 55 subjects. Odor threshold, discrimination, and identification did not play a relevant role for cluster assignment, which, on the other hand, depended on performance in the two odor dilution sorting tasks, from which cluster assignment was possible with a median 100-fold cross-validated balanced accuracy of 77–88%. Conclusions: The addition of an odor sorting task with the two proposed odor dilutions to the odor test battery expands the phenotype of olfaction and fits seamlessly into the sensory focus of standard test batteries.
Bacteria that are capable of organizing themselves as biofilms are an important public health issue. Knowledge discovery focusing on the ability to swarm and conquer the surroundings to form persistent colonies is therefore very important for microbiological research communities that focus on a clinical perspective. Here, we demonstrate how a machine learning workflow can be used to create useful models that are capable of discriminating distinct associated growth behaviors along distinct phenotypes. Based on basic gray-scale images, we provide a processing pipeline for binary image generation, making the workflow accessible for imaging data from a wide range of devices and conditions. The workflow includes a locally estimated regression model that easily applies to growth-related data and a shape analysis using identified principal components. Finally, we apply a density-based clustering application with noise (DBSCAN) to extract and analyze characteristic, general features explained by colony shapes and areas to discriminate distinct Bacillus subtilis phenotypes. Our results suggest that the differences regarding their ability to swarm and subsequently conquer the medium that surrounds them result in characteristic features. The differences along the time scales of the distinct latency for the colony formation give insights into the ability to invade the surroundings and therefore could serve as a useful monitoring tool.
Bayesian inference is ubiquitous in science and widely used in biomedical research such as cell sorting or “omics” approaches, as well as in machine learning (ML), artificial neural networks, and “big data” applications. However, the calculation is not robust in regions of low evidence. In cases where one group has a lower mean but a higher variance than another group, new cases with larger values are implausibly assigned to the group with typically smaller values. An approach for a robust extension of Bayesian inference is proposed that proceeds in two main steps starting from the Bayesian posterior probabilities. First, cases with low evidence are labeled as “uncertain” class membership. The boundary for low probabilities of class assignment (threshold 𝜀
) is calculated using a computed ABC analysis as a data-based technique for item categorization. This leaves a number of cases with uncertain classification (p < 𝜀
). Second, cases with uncertain class membership are relabeled based on the distance to neighboring classified cases based on Voronoi cells. The approach is demonstrated on biomedical data typically analyzed with Bayesian statistics, such as flow cytometric data sets or biomarkers used in medical diagnostics, where it increased the class assignment accuracy by 1–10% depending on the data set. The proposed extension of the Bayesian inference of class membership can be used to obtain robust and plausible class assignments even for data at the extremes of the distribution and/or for which evidence is weak.
Phenotypical screening is a widely used approach in drug discovery for the identification of small molecules with cellular activities. However, functional annotation of identified hits often poses a challenge. The development of small molecules with narrow or exclusive target selectivity such as chemical probes and chemogenomic (CG) libraries, greatly diminishes this challenge, but non-specific effects caused by compound toxicity or interference with basic cellular functions still pose a problem to associate phenotypic readouts with molecular targets. Hence, each compound should ideally be comprehensively characterized regarding its effects on general cell functions. Here, we report an optimized live-cell multiplexed assay that classifies cells based on nuclear morphology, presenting an excellent indicator for cellular responses such as early apoptosis and necrosis. This basic readout in combination with the detection of other general cell damaging activities of small molecules such as changes in cytoskeletal morphology, cell cycle and mitochondrial health provides a comprehensive time-dependent characterization of the effect of small molecules on cellular health in a single experiment. The developed high-content assay offers multi-dimensional comprehensive characterization that can be used to delineate generic effects regarding cell functions and cell viability, allowing an assessment of compound suitability for subsequent detailed phenotypic and mechanistic studies.