Hanna Huebner

Die Abkürzung für „künstliche Intelligenz“ ist KI. Damit ist gemeint, dass Computer Aufgaben übernehmen können, die normalerweise Menschen durchführen. Für die Entwicklung dieser Fähigkeiten benötigen Computer oder andere technische Geräte wie Smartphones, Smartwatches oder Wearables spezielle Programme oder sogenannten Code. Dieser Code wird von Menschen geschrieben [1]. So entsteht der Eindruck, dass der Computer selbstständig denken und handeln kann. Tatsächlich handelt es sich hier um künstlich erzeugte Intelligenz und nicht um echtes Denken. Ein bedeutender Unterschied zwischen KI und Menschen besteht bei Gefühlen. Menschen empfinden Freude, Trauer oder Überraschung. Eine KI besitzt solche Gefühle nicht. Damit eine KI das kann, was sie kann, muss die KI zuvor lernen. Dazu sagt man auch, dass die KI trainiert werden muss. Für das Training der KI sind die Daten entscheidend, aus denen Muster und Regeln abgeleitet werden. Die Zuverlässigkeit einer KI hängt ausschlaggebend von der Qualität und Relevanz der Daten ab, mit denen sie trainiert wird [2]. Werden beispielsweise geeignete medizinische Daten verwendet, kann KI dazu beitragen, die Medizin von morgen zu unterstützen und voranzubringen [3]. Für das Training von KI-Systemen werden große Datenmengen benötigt. Dazu zählen unter anderem Anamnesen, Diagnosen, Therapien, Medikamentenpläne, Labor- und Bilddaten sowie Befundberichte aus der elektronischen Patientenakte. Diese Daten enthalten sehr persönliche Informationen. Deshalb ist es wichtig, sie gut zu schützen. Kliniken, Universitäten und Forschungseinrichtungen speichern Daten verschlüsselt, damit Unbefugte diese nicht lesen können. Außerdem wird genau festgelegt, wer Zugriff hat. Häufig werden persönliche Angaben entfernt (Anonymisierung), sodass keine Rückschlüsse mehr auf einzelne Personen möglich sind. Die Datenschutz-Grundverordnung (DSGVO) schützt in Europa das Recht auf Datenschutz [4]. Sie ist entscheidend dafür, dass mit KI Fortschritt ermöglicht wird, ohne die Privatsphäre des Einzelnen zu gefährden. KI benötigt viel Rechenleistung – vor allem, wenn große Modelle mit riesigen Datenmengen trainiert werden. Ein normaler Laptop oder Computer stößt dabei sehr schnell an seine Grenzen. Deshalb kommen spezielle Hochleistungsrechner zum Einsatz. Diese bestehen aus vielen leistungsstarken Prozessoren (CPUs) und Grafikkarten (GPUs), die zusammenarbeiten. Die schnellsten und mächtigsten dieser Systeme nennt man Supercomputer. Sie ermöglichen es, komplexe KI-Modelle in vertretbarer Zeit zu entwickeln und zu verbessern – Aufgaben, für die ein normaler Computer Jahre brauchen würde. Damit KI-Algorithmen in der Medizin eingesetzt werden können, müssen sie zuverlässig, nachvollziehbar und sicher funktionieren. Patientinnen und Patienten, Angehörige, Ärztinnen und Ärzte und das medizinische Fachpersonal müssen darauf vertrauen können, dass die Ergebnisse korrekt sind und die notwendigen medizinischen Entscheidungen unterstützen. Hier ist ein zentraler Aspekt die Transparenz [5]. Für jede und jeden muss nachvollziehbar sein, wie eine KI zu einem Ergebnis kommt. Hierfür relevant wiederum sind die Daten. Nur mit qualitativ hochwertigen, repräsentativen Daten lassen sich faire und genaue Vorhersagen treffen. KI soll medizinisches Personal unterstützen, nicht ersetzen, und ist immer in ein verantwortungsvolles menschliches Handeln eingebettet. Article published online: 17 June 2026 © 2026. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany

Background: Mobile health (mHealth) apps promise to enhance patient empowerment, enable real-time health monitoring, and support self-management. Patients with breast cancer stand to benefit particularly from these capabilities, given the demanding pre- and posttreatment procedures they face. However, the effective adoption of these tools is challenged by issues including accessibility, usability limitations, and privacy concerns. Objective: This study aimed to investigate the current usage behavior and attitudes toward mHealth apps among patients with breast cancer in Germany, focusing on acceptance, usefulness, empowerment, and data-sharing. Specifically, we aimed to identify the incentives and barriers influencing use and the intention to use mHealth apps at the time of this writing to guide their further implementation and use. Methods: A cross-sectional online survey was conducted between November 2023 and June 2024 using a structured questionnaire developed in German. Participants with a self-reported breast cancer diagnosis were recruited via university hospital channels, patient associations, and social media. Topics included technology and health app usage, perceived usefulness, personal empowerment, and attitudes toward data sharing. The survey comprised both closed- and open-ended questions to capture quantitative patterns and qualitative insights into participants’ experiences and perspectives. Results: We collected responses from 90 participants with breast cancer. Participants demonstrated high engagement with digital technology; 97% (87/90) reported daily smartphone use, and 70% (63/90) owned a smartwatch or fitness tracker. Approximately 67% (60/90) of respondents had used health apps in the past year, with many relying on general fitness and lifestyle apps to support their health management. Most participants (62/90, 69%) indicated that health apps contribute to a better understanding of their personal health. Health apps were generally perceived as useful, with users highlighting their ability to provide continuous support and timely, reliable health information (n=31). This, in turn, was seen as instrumental in promoting self-management and empowerment. Respondents also showed a conditional willingness to share personal health data for research, particularly when clear benefits for treatment advancements were evident (n=82). Nonetheless, concerns were raised regarding data privacy, app usability, the need for multiple apps to cover all necessary features, and the potential increase in fear of illness (cyberchondria). Conclusions: The findings underscore the potential of mHealth apps to empower patients with breast cancer by improving health literacy and supporting self-management. To maximize their impact, future digital health tools must align closely with patient needs by incorporating adaptive, interactive features and integrating seamlessly with clinical care. Addressing key barriers—particularly data privacy and usability issues—is essential for broader adoption. Longitudinal studies are warranted to evaluate the long-term effects of health app usage on patient empowerment, quality of life, and clinical outcomes.

Background: Wearables and smart devices could complement face-to-face prenatal care appointments by monitoring pregnant women's health, especially since pregnancy may be a vulnerable time when mental health issues and pregnancy-related anxiety may arise. Aim: The aim of this study was to analyze the extent whether the willingness of women to use smart devices for pregnancy care monitoring differs between those with pregnancy-related anxiety and those without. Methods: A survey was conducted to ascertain participant's general characteristics, attitudes toward smart devices, the willingness to use them, and the level of pregnancy-related anxiety (PRAQ-R2). Associations between the willingness and pregnancy-related anxiety parameters were analyzed. Results: Completed questionnaires from 210 women were included in the analysis. A significant difference between women showing high and low levels of pregnancy-related anxiety was observed in terms of their willingness to use a smartwatch (median score 4.00; interquartile range (IQR) 4.00–5.00 vs. 4.00; IQR 3.00–4.00; P = .02), smart contraction counter (median score 4.00; IQR 3.00–5.00 vs. 3.00; IQR 3.00–4.00; P = .02), smart blood pressure monitor (median score 5.00; IQR 3.00–5.00 vs. 4.00; IQR 3.00–4.00; P = .003) or sleep tracker (median score 4.00; IQR 3.00–5.00 vs. 3.00; IQR 2.00–4.00; P = .007). Overall, anxious women showed significantly higher willingness to use smart devices in the context of prenatal care. Conclusions: The data suggest that pregnant women are receptive to using smart devices to enhance their prenatal care, particularly those experiencing higher levels of anxiety. This study serves as an initial step in evaluating attitudes toward these devices. As a follow-up, it is recommended that acceptance and feasibility studies are conducted alongside the further development of existing and new pregnancy-specific wearables.

Die Abkürzung für „künstliche Intelligenz“ ist KI. Damit ist gemeint, dass Computer Aufgaben übernehmen können, die normalerweise Menschen durchführen. Für die Entwicklung dieser Fähigkeiten benötigen Computer oder andere technische Geräte wie Smartphones, Smartwatches oder Wearables spezielle Programme oder sogenannten Code. Dieser Code wird von Menschen geschrieben [1]. So entsteht der Eindruck, dass der Computer selbstständig denken und handeln kann. Tatsächlich handelt es sich hier um künstlich erzeugte Intelligenz und nicht um echtes Denken. Ein bedeutender Unterschied zwischen KI und Menschen besteht bei Gefühlen. Menschen empfinden Freude, Trauer oder Überraschung. Eine KI besitzt solche Gefühle nicht. Damit eine KI das kann, was sie kann, muss die KI zuvor lernen. Dazu sagt man auch, dass die KI trainiert werden muss. Für das Training der KI sind die Daten entscheidend, aus denen Muster und Regeln abgeleitet werden. Die Zuverlässigkeit einer KI hängt ausschlaggebend von der Qualität und Relevanz der Daten ab, mit denen sie trainiert wird [2]. Werden beispielsweise geeignete medizinische Daten verwendet, kann KI dazu beitragen, die Medizin von morgen zu unterstützen und voranzubringen [3]. Für das Training von KI-Systemen werden große Datenmengen benötigt. Dazu zählen unter anderem Anamnesen, Diagnosen, Therapien, Medikamentenpläne, Labor- und Bilddaten sowie Befundberichte aus der elektronischen Patientenakte. Diese Daten enthalten sehr persönliche Informationen. Deshalb ist es wichtig, sie gut zu schützen. Kliniken, Universitäten und Forschungseinrichtungen speichern Daten verschlüsselt, damit Unbefugte diese nicht lesen können. Außerdem wird genau festgelegt, wer Zugriff hat. Häufig werden persönliche Angaben entfernt (Anonymisierung), sodass keine Rückschlüsse mehr auf einzelne Personen möglich sind. Die Datenschutz-Grundverordnung (DSGVO) schützt in Europa das Recht auf Datenschutz [4]. Sie ist entscheidend dafür, dass mit KI Fortschritt ermöglicht wird, ohne die Privatsphäre des Einzelnen zu gefährden. KI benötigt viel Rechenleistung – vor allem, wenn große Modelle mit riesigen Datenmengen trainiert werden. Ein normaler Laptop oder Computer stößt dabei sehr schnell an seine Grenzen. Deshalb kommen spezielle Hochleistungsrechner zum Einsatz. Diese bestehen aus vielen leistungsstarken Prozessoren (CPUs) und Grafikkarten (GPUs), die zusammenarbeiten. Die schnellsten und mächtigsten dieser Systeme nennt man Supercomputer. Sie ermöglichen es, komplexe KI-Modelle in vertretbarer Zeit zu entwickeln und zu verbessern – Aufgaben, für die ein normaler Computer Jahre brauchen würde. Damit KI-Algorithmen in der Medizin eingesetzt werden können, müssen sie zuverlässig, nachvollziehbar und sicher funktionieren. Patientinnen und Patienten, Angehörige, Ärztinnen und Ärzte und das medizinische Fachpersonal müssen darauf vertrauen können, dass die Ergebnisse korrekt sind und die notwendigen medizinischen Entscheidungen unterstützen. Hier ist ein zentraler Aspekt die Transparenz [5]. Für jede und jeden muss nachvollziehbar sein, wie eine KI zu einem Ergebnis kommt. Hierfür relevant wiederum sind die Daten. Nur mit qualitativ hochwertigen, repräsentativen Daten lassen sich faire und genaue Vorhersagen treffen. KI soll medizinisches Personal unterstützen, nicht ersetzen, und ist immer in ein verantwortungsvolles menschliches Handeln eingebettet. Article published online: 14 January 2026 © 2026. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany

Immunotherapy has become standard of care in the treatment of diffuse large B-cell lymphoma (DLBCL). Changes in immunophenotypes observed at first diagnosis predict therapy outcome but little is known about the resolution of these alterations in remission. Comprehensive characterization of immune changes from fresh, peripheral whole blood revealed a functionally relevant increase of myeloid-derived suppressor cells, reduced naïve T cells, and an increase of activated and terminally differentiated T cells before treatment, which aggravated after therapy. Suggesting causal relation, injection of lymphoma in mice induced similar changes in the murine T cells. Distinct immune imprints were found in those who have survived breast cancer and acute myeloid leukemia. Identified alterations persisted beyond 5 years of ongoing complete remission and correlated with increased proinflammatory markers such as interleukin-6, β2-microglobulin, or soluble CD14 in DLBCL. The chronic inflammation was associated with functionally blunted T-cell immunity against severe acute respiratory syndrome coronavirus 2–specific peptides, and reduced responses correlated with reduced naïve T cells. Persisting inflammation was confirmed by deep sequencing and by cytokine profiles, together pointing toward a compensatory activation of innate immunity. The persisting, lymphoma-induced immune alterations in remission may explain long-term complications, have implications for vaccine strategies, and are likely relevant for immunotherapies.

Background: The introduction of oral anticancer therapies has, at least partially, shifted treatment from clinician-supervised hospital care to patient-managed home regimens. However, patients with breast cancer receiving oral cyclin-dependent kinase 4/6 inhibitor therapy still require regular hospital visits to monitor side effects. Telemonitoring has the potential to reduce hospital visits while maintaining quality care. Objective: This study aims to develop a digital home-based health care center (DHHC) for acquiring electrocardiograms (ECGs), white blood cell (WBC) counts, side effect photo documentation, and patient-reported quality of life (QoL) data. Methods: The DHHC was set up using an Apple Watch Series 6 (ECG measurements), a HemoCue WBC DIFF Analyzer (WBC counts), an iPhone SE (QoL assessments and photo documentation), a TP-Link M7350-4G Wi-Fi router, and a Raspberry Pi 4 Model B. A custom-built app stored and synchronized remotely collected data with the clinic. The feasibility and acceptance of the DHHC among patients with breast cancer undergoing cyclin-dependent kinase 4/6 inhibitor therapy were evaluated in a prospective, single-arm, monocentric study. Patients (n=76) monitored side effects—ECGs, WBC counts, photo documentation, and QoL—at 3 predefined time points: study inclusion (on-site), day 14 (remote), and day 28 (remote). After the study completion, patients completed a comprehensive questionnaire on user perception and feasibility. Adherence to scheduled visits, the success rate of the data transfer, user perception and feasibility, and the clinical relevance of remote measurements were evaluated. Results: Mean adherence to the planned remote visits was 63% on day 14 and 37% on day 28. ECG measurements were performed most frequently (day 14: 57/76, 75%; day 28: 31/76, 41%). The primary patient-reported reason for nonadherence was device malfunction. The expected versus the received data transfer per patient was as follows: ECGs: 3 versus 3.04 (SD 1.9); WBC counts: 3 versus 2.14 (SD 1.14); QoL questionnaires: 3 versus 2.5 (SD 1.14); and photo documentation: 6 versus 4.4 (SD 3.36). Among patients, 81% (55/68) found ECG measurements easy, 82% (55/67) found photo documentation easy, and 48% (33/69) found WBC measurements easy. Additionally, 61% (40/66) of patients felt comfortable with self-monitoring and 79% (54/68) were willing to integrate remote monitoring into their future cancer care. Therapy-induced decreased neutrophil count was successfully detected (P<.001; mean baseline: 4.3, SD 2.2, ×10⁹/L; on-treatment: 1.8, SD 0.8, ×10⁹/L). All-grade neutropenia and corrected QT interval prolongations were detected in 80% (55/68) and 2% (1/42) of patients, respectively. Conclusions: Adherence to scheduled remote visits was moderate, with nonadherence primarily attributed to device-related complications, which may have also affected the success rate of data transfer. Overall, patients considered remote monitoring useful and feasible. The prevalence of reported adverse events was comparable to existing literature, suggesting clinical potential. This initial feasibility study highlights the potential of the DHHC.

Objective: This study aimed to assess patients’ preferences regarding the content and features of a digital patient navigator app. A secondary objective was to explore how patient characteristics influence the perceived importance and anticipated frequency of use of different app components. Methods: The study was a monocentric cross-sectional survey conducted at the University Hospital Erlangen. A questionnaire consisting of 20 questions was designed to inquire about patients’ preferences for a digital patient navigator. Participants had breast diseases. Descriptive analyses were conducted to identify key app features and their correlation with patient characteristics. Results: Questionnaires from 243 patients were analyzed. Preferred key app features, in order of preference from highest to lowest, were as follows: a communication feature to chat with treating physicians, a personalized treatment plan, real-time notifications, mediation services for providing support, 3D navigation within the hospital, and personal documents. Patients’ age, native language, and education significantly influenced the assessment of importance and frequency of use of some app features. A personalized treatment plan was considered more important by younger individuals (≤60 years, p < 0.001), whereas real-time notifications were considered more important by participants with a lower educational level ( p = 0.003) and younger individuals ( p = 0.036). Increased frequency of use of a personalized treatment plan tool was also associated with younger age ( p < 0.001) and lower education levels ( p = 0.025). Conclusion: These findings suggest that a patient navigator app could be a valuable tool for a broad range of patients, potentially complementing in-person patient navigators. To ensure broad usability and acceptance, future development should account for varying needs across patient subgroups—particularly in terms of personalization, language accessibility, and communication features.

Background: Peripheral blood mononuclear cells (PBMCs) are valuable biomarkers, providing crucial insights into the patients’ immune system. Reliable biobanking of PBMCs is essential to minimize heterogeneity. In multicenter trials, blood sample transportation to central laboratories can increase the time between blood collection and PBMC isolation. This study evaluated the effect of prolonged blood hold time on PBMC viability and cytotoxicity. Methods: From July 2021 to May 2023, 104 patients with early HER2-positive breast cancer were enrolled in the NeoOn trial, of whom 49 patients were included in this subproject. PBMCs were isolated ≤ 6 hours (h) or ≥ 20 h after blood collection. PBMC yield and viability were determined using the LUNA-II Automated Cell Counter. Flow cytometry was used to quantify in vitro cytotoxicity, the percentage of natural killer (NK) and T cells, as well as apoptotic and necrotic cells. Results: Isolating PBMCs ≥ 20 h resulted in a higher cell yield, but lower NK cell viability compared to PBMCs ≤ 6 h. PBMCs ≥ 20 h were less robust to thawing and showed higher loss during recovery. Compared to PMBCS ≤ 6 h, PBMCs ≥ 20 h exhibited lower antibody-mediated cytotoxicity (p ≤ 0.0001) and antibody-dependent phagocytosis (p < 0.0051). While the percentage of T and NK cells and the T cell viability remained unaffected by hold time, the percentage of apoptotic NK cells was higher for PBMCs ≥ 20 h (41.0 ± 12.9% vs. 23.8 ± 13.4%; p = 0.0364). Conclusions: Extended blood storage time caused increased apoptosis and necrosis of NK cells, adversely affecting PBMC quality and reducing NK cell related functionality. Hence, blood hold time should be minimized to maintain PBMC integrity and NK cell functionality for in vitro biomarker assays. Trial registration: Trial registration number: EudraCT 2020-001943-21. Date of registration: December 29th 2020