616 Krankheiten
Refine
Document Type
- Article (18)
- Conference Paper (10)
- Article in a Periodical of the TH Wildau (8)
- Bachelor Thesis (1)
- Part of a Book (1)
Year of publication
Institute
Has Fulltext
- yes (38)
Keywords
- automatic speech recognition (3)
- deep learning (3)
- Aphasia (2)
- C60 fullerene (2)
- apoptosis (2)
- copy number alteration (2)
- digital health (2)
- machine learning (2)
- photoplethysmography (2)
- prostate cancer (2)
Neurologische Erkrankungen gehören weltweit zu den häufigsten Ursachen für Behinderungen und Tod. Laut dem Globalen Statusbericht der Neurologie der WHO sind mehr als 40 Prozent der Weltbevölkerung von neurologischen Erkrankungen betroffen. Um dieser wachsenden Belastung zu begegnen, sind effizientere Datenverarbeitungssysteme und eine verbesserte Überwachung erforderlich, um evidenzbasierte Entscheidungsfindung und Rechenschaftspflicht im Gesundheitswesen zu ermöglichen.
Gleichzeitig eröffnen Fortschritte an der Schnittstelle von Quantencomputing und Gesundheitswesen neue Möglichkeiten für medizinische Technologien. Quantenphysikalische Prinzipien werden genutzt, um diagnostische Werkzeuge, Therapieplanung und Patientenversorgung durch höhere Recheneffizienz und Präzision zu verbessern. Klassische Machine-Learning-Modelle stoßen bei hochdimensionalen und komplexen medizinischen Datensätzen, insbesondere im Bereich der Neurologie, häufig an ihre Grenzen. Diese Herausforderung hat das Interesse an Quantum Machine Learning geweckt, das potenziell Vorteile bei der Verarbeitung und Analyse solcher Daten bietet und in bestimmten Anwendungen klassische Methoden übertreffen könnte.
In dieser Studie werden klassische Machine-Learning-Klassifikatoren mit Quantum-Machine-Learning-Klassifikatoren, mit Fokus auf Klassifikation neurologischer Erkrankungen, verglichen. Die Ergebnisse zeigen, dass Quantenmodelle weiter verfeinert und problemspezifisch angepasst werden müssen, um konkurrenzfähig zu sein. Obwohl frühere Studien mögliche Vorteile nahelegen, sind maßgeschneiderte Quantenalgorithmen und weitere Forschung erforderlich, um das Potenzial des Quantencomputings in der Medizin vollständig auszuschöpfen.
Leishmaniose ist eine parasitäre Erkrankung, die durch Sandmücken übertragen wird und vor allem in tropischen Regionen verbreitet ist. Die klassische Diagnose erfolgt mittels mikroskopischer Untersuchung gefärbter Gewebeproben, erfordert jedoch viel Expertise und ist anfällig für subjektive Fehler. Zur Verbesserung dieses Verfahrens entwickelten und evaluierten wir ein Deep-Learning-basiertes Echtzeit-Detektionssystem zur Identifikation von Leishmanien in Hellfeld-Mikroskopiebildern. Verwendet wurden Gewebeproben von PCR-bestätigten Fällen kutaner Leishmaniasis, aufgenommen sowohl mit einem hochauflösenden Labormikroskop als auch mit einem kostengünstigen mobilen Gerät, um unterschiedliche Bildqualitäten abzudecken. Die Parasiten wurden mittels Bounding-Boxen annotiert und ein YOLOv8-Modell trainiert. Das Modell erreichte eine Genauigkeit von 84 %, Sensitivität von 84 %, Spezifität von 82 % und Präzision von 87 %. Eine grafische Benutzeroberfläche ermöglicht die Echtzeitanalyse von Live-Mikroskopbildern.
Post hoc implementation of non-standard phonetic features in the context of aphasic speech analysis
(2025)
Despite current progress, automatic speech recognition (ASR) often struggles with non-standard speech, for example, influenced by dialectal or pathological features. (Re)training ASR models to accommodate these variations is not always possible due to limited data. This paper proposes applying the knowledge about non-standard (aphasic and dialectal) phonetic features to the ASR transcription post hoc. Using speech data from German speakers with aphasia who speak the Thuringian-Upper Saxon dialect, this study evaluates the impact of these modifications on an ASR-based error analysis pipeline. The approach helps to reduce automatic error rates on the recordings manually labelled as error-free. The performance of the pipeline also improves both in general acceptance or rejection of the responses and error attribution. General acceptance/rejection accuracy reaches the mean of 83.3%, which is considered sufficient to be used in a digital application for speech and language therapy support.
Those who suffer from aphasia benefit from digital speech and language therapy solutions, and automatic speech recognition (ASR) has been already used for giving feedback on the correctness of the answers in naming exercises. AphaDIGITAL application is to provide German-speaking users with detailed feedback on phonemic/phonetic and semantic errors, based on automatic speech and language processing. For this purpose, open-source ASR solutions for German were evaluated on different corpora of atypical speech, including two small datasets with aphasic speech samples. Character error rate, the number of precisely recognized items and empty outputs served as evaluation metrics. The four selected models are generally robust to the deteriorated condition of speech and audio quality and consistently outperform commercial models in atypical speech recognition. Applying error acceptance threshold, additional use of phonemic error rate, and other valuable insights for ASR implementation in aphaDIGITAL are discussed.
Background
Accurate and rapid diagnosis of rheumatic diseases is essential for further treatment decision. Different rheumatic diseases present characteristic patterns (image features) in fluorescence optical imaging (FOI). We developed an atlas of FOI image features and tested its ability to differentiate various rheumatic diseases.
Methods
FOI images from patients with rheumatoid arthritis (RA), psoriatic arthritis (PsA), connective tissue diseases (CTD) and osteoarthritis (OA) were analysed by two readers blinded for diagnosis and calibrated against each other, using the prima vista mode (PVM) and an automated 5-phase model. Twenty-six different reoccurring typical signal enhancement patterns (features) indicating inflamed joints, nail or skin were defined and all FOI images were scored accordingly. The feature frequency in each patient cohort and phase (PVM, 5-phase) was counted. Contingency tables were created with categorical variable counts and diagnosis using common formulae.
Findings
Four hundred thirty-eight patients with RA (n=117), PsA (n=110), CTD (n=121) and OA (n=90) were included. Once the data had been categorised, a two-step diagnostic pathway was developed: in the first step, OA was best distinguished from the other diseases with high specificity by five patterns (specificity >0.9, diagnostic OR between 2.34 and 8.24). In a second step, the remaining autoimmune diseases were differentiated from each other by a certain number of features (five for RA, 12 for PsA and four for CTD).
Interpretation
This was the first study to show that feature analysis in FOI helps to differentiate typical rheumatic diseases from each other, potentially simplifying and speeding up the diagnostic process. Therefore, FOI could be considered an additional component of a wider range of imaging techniques used in rheumatology.
In mammals, proper storage and distribution of lipids in and between tissues is essential for the maintenance of energy homeostasis. In contrast, aberrantly high levels of triglycerides in the blood ("hypertriglyceridemia") represent a hallmark of the metabolic syndrome and type 2 diabetes. As hypertriglyceridemia has been identified as an important risk factor for cardiovascular complications, in this study we aimed to identify molecular mechanisms in aberrant triglyceride elevation under these conditions.
To determine the importance of hepatic lipid handling for systemic dyslipidemia, we profiled the expression patterns of various hepatic lipid transporters and receptors under healthy and type 2 diabetic conditions. A differentially expressed lipoprotein receptor was functionally characterized by generating acute, liver-specific loss- and gain-of-function animal models.
We show that the hepatic expression of lipid transporter lipolysis-stimulated lipoprotein receptor (LSR) is specifically impaired in mouse models of obesity and type 2 diabetes and can be restored by leptin replacement. Experimental imitation of this pathophysiological situation by liver-specific knockdown of LSR promotes hypertriglyceridemia and elevated apolipoprotein (Apo)B and E serum levels in lean wild-type and ApoE knockout mice. In contrast, genetic restoration of LSR expression in obese animals to wild-type levels improves serum triglyceride levels and serum profiles in these mice.
The dysregulation of hepatic LSR under obese and diabetic conditions may provide a molecular rationale for systemic dyslipidemia in type 2 diabetes and the metabolic syndrome and represent a novel target for alternative treatment strategies in these patients.
Linguistic and extralinguistic factors in automatic speech recognition of German atypical speech
(2024)
Automatic speech recognition (ASR) has been already used in speech and language therapy, including diagnostic tasks and practice exercises for people with aphasia (PWA). The lack of relevant data makes it difficult to evaluate the algorithms’ suitability for German-speaking PWA. For the current project, four open-source ASR models were selected based on their performance on other types of atypical speech, and the details of their evaluation are presented in this paper. The four selected models are generally robust to speakers’ gender and age. The one-word recognition yields better results for words of moderate length. Speech rate should be neither too slow nor too quick for lower error rates both in words and phrases, and the latter should be also of moderate length.
Stress is an important factor affecting human health, and its timely detection can significantly improve quality of life. This study addresses the current issue of identifying stress states using photoplethysmography signals obtained from the wearable Shimmer 3 sensor. The purpose of the study was to establish an effective approach for stress detection, specifically through the analysis of heart rate variability changes during different states: rest and stress. A distinctive feature of the research is the use of the air raid siren sound as a stress factor. The research methodology includes the collection and analysis of biosignals, allowing for the assessment of cardiovascular system functioning under the impact of stress factors. The results of the study demonstrated significant differences in heart rhythm indicators depending on the person's state, highlighting the potential of these metrics as a tool for health monitoring and stress detection. The paper proposes a new approach to using photoplethysmography for assessing stress responses, which may contribute to the development of personalized stress management methods.
This study investigates stress recognition using galvanic skin response (GSR) and photoplethysmography (PPG) data and machine learning, with a new focus on air raid sirens as a stressor. It bridges laboratory and real-world conditions and highlights the reliability of wearable sensors in dynamic, high-stress environments such as war and conflict zones. The study involves 37 participants (20 men, 17 women), aged 20–30, who had not previously heard an air raid siren. A 70 dB “S-40 electric siren” (400–450 Hz) was delivered via headphones. The protocol included a 5 min resting period, followed by 3 min “no-stress” phase, followed by 3 min “stress” phase, and finally a 3 min recovery phase. GSR and PPG signals were recorded using Shimmer 3 GSR+ sensors on the fingers and earlobes. A single session was conducted to avoid sensitization. The workflow includes signal preprocessing to remove artifacts, feature extraction, feature selection, and application of different machine learning models to classify the “stress “and “no-stress” states. As a result, the best classification performance was shown by the k-Nearest Neighbors model, achieving 0.833 accuracy. This was achieved by using a particular combination of heart rate variability (HRV) and GSR features, which can be considered as new indicators of siren-induced stress.
Identification of a bone density is crucial in all branches of medicine. Assessing bone density is a labor-intensive process. Bone density is quite unstable and depends on many factors, both physiological (aging, hormonal balance) and pathological (medication use, various underlying conditions).
The aim of our study is to determine bone density in patients with malignant breast tumors undergoing anti-cancer therapy.
Material and Methods. The study included 50 women aged 60-70 years who were diagnosed with infiltrating intraductal carcinoma. According to established protocols, MSCT is recommended for this category of women at intervals of once every six months. The first MSCT scan was performed immediately after the diagnosis was made, before the start of treatment, and the second scan was conducted six months later.
Results. The maximum radiological density was 75.8954±37.9477 Hu in the group of women who had been receiving treatment for six months, compared to 93.9388±46.9694 Hu in the group of patients who did not take the drug. Meanwhile, the minimum density showed a slight increase from 29.7295±14.8647 Hu to 38.6919±19.3460 Hu, which can be attributed to the compensatory mechanisms of the body.
Conclusions. In the course of this study, bone density in patients with infiltrating intraductal carcinoma undergoing anti-cancer therapy was determined using uncertainty estimation. It was found that after six months, the first to respond to changes in density was the maximum bone density.
