004 Datenverarbeitung; Informatik
Refine
Document Type
- Part of a Book (31)
- Article (19)
- Conference Publication (13)
- Book (6)
- Report (4)
- Other (1)
- Recension (1)
- Working Paper (1)
Institute
Is part of the Bibliography
- yes (76)
Objective: Electrocardiogramms (ECG) interpretation is increasingly supported by algorithms trained on large-scale datasets. Here we present a first manually annotated open-source ECG-database of children and adults with congenital heart defect (CHD) with arrhythmia annotations.
Approach: Signal data (ECG and intracardiac electrograms) were extracted from children and patients with CHD undergoing electrophysiological studies (EPS). The dataset is provided in the wave form database format. Annotations were manually annotated using the LightWAVE® software.
Main results: The dataset comprises 39 ECGs from 39 patients (duration ranging from 00:01:17 h to 02:30:31 h). A total of 113 924 beats were annotated. A large part of beats (33%) are normal sinus rhythm. However, the database also includes sinus rhythm beats with preexcitation (12%) or complete right bundle branch block (11%). Supraventricular tachycardias (23%) are further categorized into atrioventricular reentrant tachycardias (AVRT; 15.5%) and atrioventricular nodal reentrant tachycardias (AVNRT; 7.5%). Also, ventricular tachycardias (1.7%) and very rare conditions as aberrant conducted AVRT or AVNRT or AVNRT with 2nd degree heart block are represented. Additionally paced beats are included (12%), thereof 7% atrial paced beats and 5% ventricular paced beats. Premature beats (premature ventricular or atrial beats) make up 7% of the dataset.
Significance: This open-source database (https://physionet.org/content/leipzig-heart-center-ecg) containing many pathologies in children and patients with CHD may provide a foundation for ECG algorithms aimed at detecting ECG pathologies in this special patient population.
From Scenario Farming to Learning: A Modular Low-Code Framework for Decision Support in Scheduling
(2025)
This paper presents a relational integration of the TSXor compression algorithm for regularly sampled time-series data. We design and implement a schema and compression pipeline within TimescaleDB, allowing lossless compression and in-database decompression using SQL and PL/pgSQL. Our approach replaces full timestamps with delta-based offsets and encodes floating-point values using reference-based XOR encoding. We evaluate the system on multiple real-world datasets, showing moderate compression ratios (up to 23%) and fast decompression performance. While the SQL-based implementation introduces storage and execution overhead, the integration enables transparent querying and lays the foundation for further database-native optimizations of time-series compression.
Heart failure with preserved ejection fraction (HFpEF) remains challenging to diagnose due to the complexity of diastolic function assessment during stress echocardiography, where multiple hemodynamic parameters must be evaluated under time pressure. Explainable artificial intelligence, specifically rule-based Clinical Decision Support Systems (CDSS), offers promising improvements in reproducibility and interpretability.
Methods: A rule-based CDSS was developed and clinically validated to automate left ventricular diastolic function assessment during semi-supine bicycle stress echocardiography. A prospective cohort of 134 patients (mean age 61.3 ± 8.7 years) with exertional dyspnea and preserved left ventricular ejection fraction (LVEF >50%) was enrolled, excluding individuals with significant valvular pathologies, arrhythmias, or unstable ischemia. Echocardiographic and Doppler data were collected using Toshiba Aplio500 and Esaote MyLabSIGMA systems. The algorithm incorporated manual input of measurements, computed derived indices (e.g., diastolic reserve index, myocardial stiffness, vascular resistance), and applied rule-based logic in accordance with ASE/EACVI (2016/2022) guidelines and the ESC HFpEF consensus.
Results: The CDSS generated diagnostic conclusions within 3 min per case, matching expert assessments in 93% of cases and correctly identifying stress-induced diastolic dysfunction in 85%. It demonstrated high diagnostic agreement (ICC &gt; 0.94) and discrimination (AUC = 0.92). Rule-based outputs, such as “Impaired diastolic reserve” or “Right ventricular dysfunction under load,” were based on combinations of parameters (e.g., E/e′ > 15, Δe′ ≤ 0, TAPSE < 17 mm, PCWR > 12 mmHg).
Conclusion: The explainable, guideline-compliant CDSS enables real-time, transparent analysis of diastolic function, supporting improved diagnostic consistency and augmented physician decision-making in cardiovascular care.
The number of people in Germany requiring care has risen steadily, increasing the importance of informal care. This form of care is often associated with considerable psychological and physical strain. The aim of this study is to systematically categorize and qualitatively analyze the free-text responses from a survey on home care using an artificial intelligence-based approach in order to identify key challenges and support needs in home care from the perspective of informal caregivers and non-caregiving relatives. The study used data from a 2019 survey on home care in Saxony. Free-text responses were categorized and analyzed using GPT-4 Turbo within a hybrid human-AI workflow. All AI outputs were subsequently validated and corrected by researchers. Respondents reported substantial financial burdens for both care recipients and informal caregivers. They also highlighted structural barriers to accessing services and insufficient support from the care system. Improving home care requires structural measures, including the expansion of low-threshold counseling services, more flexible leave regulations, stronger financial security for informal caregivers, and the sustainable strengthening of care infrastructures. Given an AI error rate of 36.45%, the study emphasizes the need for human post-processing to ensure analytical accuracy.
The rapid development of Artificial Intelligence (AI) has profoundly transformed translation practices and poses new challenges for higher education. This article presents a multi-stage teaching project designed for MA students that critically explores the potentials and limitations of AI-based translation tools. Through the comparative analysis of literary and contemporary texts—most notably Franz Kafka’s short prose piece “Gib’s auf ”—students examine outputs from tools such as DeepL, Google Translate, ChatGPT, and Matecat. The project combines text analysis, comparison of machine translations, post-editing, and collaborative translation workshops, including direct interaction with a contemporary author. Results show that while AI tools provide efficient and often accurate support, they remain limited in conveying stylistic nuance, pragmatics, and cultural meaning. The study demonstrates that translation quality depends on human interpretation, creativity, and responsibility, highlighting AI as a didactic catalyst rather than a substitute for professional translational competence.
This article deals with the use of Artificial Intelligence (AI) in teaching German for specific purposes. It examines the potential of AI for the didactic design of exercises and teaching materials, for providing personalized feedback, and for supporting differentiated instruction. As a case study, a project conducted during the summer semester 2025 in Kyrgyzstan is presented, in which ChatGPT-supported exercises were developed and evaluated. Additionally, the perspectives of teachers regarding the opportunities and risks of using AI in specialized language teaching are discussed.
In this article, I present changes to the content of the Bachelor’s degree programme German Business Communication that I consider necessary in the light of AI-driven developments both in science and economic fields. I focus on developing critical thinking when students use Generative Artificial Intelligence (GenAI) in their scientific work. This fits in well with the programme’s critical and engaged academic approach. The article sets out with an empirical consideration of AI in academic settings and GenAI application within the framework of the degree programme. Based upon this, a teaching/learning unit on GenAI use in a module of the programme is presented, which was designed and implemented in the winter semester 2025/26. The preliminary results reveal that further considerations are required to expand teaching units on AI tools for specific areas of business communication.
