Search Results (23)

Background/Objectives: Chest X-ray (CXR) images are a widely used first-line screening tool for pulmonary tuberculosis (TB) detection but are difficult to interpret, which has increased demand for an automated screening tool. Deep-learning-based computer-aided diagnosis systems have demonstrated a classification performance comparable to that of trained radiologists, but they rely on dense annotations such as lesion-level or pixel-level labels, which are costly and difficult to obtain in routine clinical workflows. We developed CPM-XNet, an annotation-efficient framework for lesion-annotation-free downstream TB classification in CXR images. Methods: CPM-XNet incorporates a compressing–projecting mask (CPM) to provide soft lung-aware modulation while preserving global contextual information. The CPM-modulated images are then used for downstream classification with multiple convolutional neural network backbones and a vision transformer baseline. Results: Experiments were conducted using an internal hospital dataset and public TB datasets, and CPM-XNet showed improved performance compared with baseline models trained on unmodulated images. In a repeated-seed evaluation of the main ResNet-101 configuration on the Tung cohort, CPM-ResNet101 showed higher and more stable performance than the non-CPM counterpart and demonstrated significant paired improvement using McNemar’s exact test. An ablation analysis indicated that CPM modulation was the main contributor to performance improvement while data augmentation and the classifier architecture further influenced the overall robustness. Conclusions: CPM-XNet provides an annotation-efficient strategy for lesion-annotation-free downstream TB classification in CXR images. The findings support preliminary technical feasibility, although larger, naturally imbalanced, cross-institutional validation is required before clinical deployment can be inferred. Full article

Background/Objectives: Cardiopulmonary resuscitation (CPR) skill assessments are susceptible to evaluator subjectivity, cognitive fatigue, and observational limitations. Although recent advances in multimodal artificial intelligence have increased the possibility of automated video-based assessment, its validity for clinical skill evaluation remains insufficiently examined. Methods: In this cross-sectional study, we enrolled 130 laypersons who underwent Basic Life Support training and skill testing. Twenty recordings were used for prompt development and 110 recordings were analyzed. Expert evaluators and GPT-4o independently assessed participants’ skills using a 12-item checklist. The manikin sensor data were the reference standard for the four chest compression metrics. Agreement was evaluated using Gwet’s agreement coefficient 1 (AC1) and intraclass correlation coefficient (2,1). Diagnostic accuracy, sensitivity, and specificity were compared using McNemar’s test. Results: Procedural items such as confirming cardiac arrest, calling 119, and requesting an automated external defibrillator showed a near-perfect agreement between experts and GPT-4o (AC1 > 0.8). However, the agreement was poor for the compression depth (AC1 = 0.374) and full chest recoil (AC1 = 0.355). Experts demonstrated high sensitivity (77.8–84.3%) but low specificity (24.6–47.8%), whereas GPT-4o showed low sensitivity (35.6–40.6%) but high specificity (69.2–76.1%). Conclusions: GPT-4o cannot serve as a standalone evaluator because of its inherent limitations in inferring three-dimensional spatial information from two-dimensional videos. However, its high agreement on procedural items and complementary error patterns with that of human evaluators on compression metrics suggests its potential as a decision support tool to mitigate expert leniency bias in CPR education. Full article

Objectives: The aim of this study was to determine differences in injury types and frequencies between piston-based and band-based automated chest compression devices in patients with non-traumatic out-of-hospital cardiac arrest (OHCA) at a German cardiac arrest center. Methods: This retrospective single-center study assessed resuscitation-related injuries in OHCA patients using protocol-based early whole-body CT scans at hospital admission. CT scans were reviewed independently by two reviewers blinded to the compression device used. Between May 2015 and September 2021, all patients resuscitated from non-traumatic OHCA, treated with a mechanical chest compression device, and showing stable return of spontaneous circulation (ROSC) until CT examination according to the institutional standard operating procedure for all OHCA patients were included. Patients were categorized by compression device type, and group differences were analyzed using the Chi-square test and Mann–Whitney U test. In addition, patient-level incidences of rib fracture types were calculated, and risk ratios with corresponding 95% confidence intervals were used to compare rib fracture patterns between groups. A p-value of <0.05 was considered statistically significant. Results: Among 71 patients, 32 received band-based and 39 piston-based treatment. Both groups were comparable in resuscitation duration, body constitution, and gender ratio, although the band-based group was older. Thoracic injuries predominated, with rib fractures representing the most frequent injury pattern (64/71, 90.1%). The median number of rib fractures per patient was 10 (IQR 8–12) in the band-based group and 9 (IQR 7–12) in the piston-based group. The band-based group had significantly more liver lacerations (5/32, 15.6% vs. 0/39, 0%; p = 0.01) and displaced rib fractures (117 vs. 87; p = 0.046; patient-level RR = 1.43, 95% CI 1.06–1.93). Conclusions: In this observational study of a CT-based cohort of OHCA patients with stable ROSC, the band-based device was associated with significantly higher frequencies of liver lacerations and displaced rib fractures than the piston-based device. These findings should be interpreted as hypothesis-generating and may support further evaluation of device-specific injury profiles in future studies. Full article

Background: Early cardiopulmonary resuscitation (CPR) and timely use of automated external defibrillators (AEDs) are critical determinants of survival following out-of-hospital cardiac arrest (OHCA). University students enrolled in healthcare degree programs represent a strategic target population for the dissemination of basic life support and defibrillation (BLS-D) skills. However, evidence on their level of knowledge, attitudes, and perceived preparedness remains limited in Southern Italy. Methods: A cross-sectional observational study was conducted between mid-December 2025 and 15 January 2026 among undergraduate healthcare students at the Magna Graecia University of Catanzaro (Italy). Data were collected using a structured, self-administered questionnaire assessing socio-demographic characteristics, CPR/AED knowledge, attitudes, and perceived confidence. Composite knowledge scores were calculated and categorized as poor, sufficient, good, or excellent. Statistical analyses included chi-square tests, Cramér’s V, and Spearman’s rank correlation. Results: A total of 604 students were included (mean age 24.4 ± 6.7 years; 69.9% female), of whom 46.4% reported prior BLS-D training. Knowledge levels were heterogeneous: myocardial infarction was widely recognized as a cause of cardiac arrest (81.1%), whereas recognition of non-shockable rhythms, including asystole and pulseless electrical activity, remained low (<25%). Procedural knowledge, particularly regarding the chain of survival and chest compression rate, improved with academic year and prior BLS-D training. Conversely, ventilation skills and correct AED pad placement were consistently inadequate. Attitudes toward CPR were largely positive; however, perceived confidence in performing resuscitation was moderate to low, especially in complex scenarios. More than 80% of students expressed strong interest in further training and supported mandatory BLS-D education. Conclusions: Healthcare students demonstrated favorable attitudes toward CPR but insufficient and uneven knowledge, particularly in rhythm recognition, ventilation, and AED use. Academic progression and structured BLS-D training were associated with improved competencies, although critical gaps persisted. Integrating mandatory, hands-on BLS-D training with regular refresher sessions into healthcare curricula should enhance preparedness and potentially reduce OHCA-related mortality, especially in high-risk regions such as Calabria. Full article

Traditional cardiopulmonary resuscitation (CPR) training faces limitations such as instructor dependency, low efficiency, and subjective assessment. To address these issues, this study proposes a novel computer vision-based method for the automation and objective evaluation of artificial respiration, shifting focus to the long-overlooked ventilation component. We developed an evaluation framework integrating human pose estimation and spatio-temporal graph convolution network (ST-GCN): first, OpenPose is utilized to extract skeletal keypoints of the rescuer, followed by action classification and recognition-including chest compressions, airway opening, and artificial breathing via a ST-GCN. Based on the American Heart Association (AHA) guidelines, this research defines and implements five quantitative metrics for ventilation quality, including CPR operation procedure, chin-frontal angle, interruption time, ventilation time, and ventilation frequency. An automated scoring model was established accordingly. Validated on a self-constructed dataset containing multi-source videos, the model achieved an accuracy of 87.64% in recognizing artificial respiration actions and 84.47% in evaluating action standardization. Experimental results demonstrate that the system can effectively and objectively evaluate the quality of artificial respiration. Compared with traditional instructor-dependent approaches, this study provides a low-cost, scalable technical solution, offering a new pathway for promoting high-quality CPR training. Full article

Background: Artificial intelligence (AI) and machine learning (ML) have rapidly expanded across the continuum of cardiopulmonary resuscitation (CPR), with growing evidence of their contribution to improving early recognition, intervention quality, and post-cardiac arrest outcomes. This narrative review synthesizes the current advancements and challenges in AI/ML-enhanced resuscitation science. Methods: A targeted literature search was conducted in Web of Science for the period 2018–2025 using the keywords “artificial intelligence” and “cardiopulmonary resuscitation”. The search identified studies addressing AI/ML applications across the resuscitation pathway, which were reviewed and categorized according to the American Heart Association’s Chain of Survival—prevention and preparedness, activation of the emergency response system, high-quality CPR including early defibrillation, advanced resuscitation interventions, post-cardiac arrest care, and recovery. Results: The literature demonstrates substantial promise for AI/ML in several domains: (1) early recognition and timely activation of emergency medical services through real-time detection algorithms; (2) optimization of high-quality CPR, including feedback systems, automated assessment of chest compressions, and prediction of defibrillation success; (3) support for advanced resuscitation interventions, such as rhythm classification, prognostication, and intra-arrest decision support; (4) post-cardiac arrest care, including outcome prediction and neuroprognostication; and (5) integrative and cross-domain approaches that link multiple phases of resuscitation into end-to-end AI-supported systems. Emerging work also highlights the role of AI in education and training, with applications in simulation, assessment, and skill reinforcement. Conclusions: AI/ML technologies hold significant potential to augment clinical performance across all links of the Chain of Survival. Their effective implementation requires attention to ethical considerations, data representativeness, and real-world validation. Future research should prioritize multicenter datasets, transparency, bias mitigation, and clinically embedded evaluation frameworks to ensure that AI/ML systems support safe, equitable, and high-impact resuscitation care. Full article

Background/Objectives: The objective of this study was to clinically validate the performance of the Nanox.AI HealthOST software in detecting incidental vertebral compression fractures (VCFs) on outpatient chest and abdomen CT scans using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). A secondary aim was to assess the rate of missed VCFs using initial radiologist reports. Methods: A retrospective analysis was performed on 590 outpatient CT scans. HealthOST, an artificial intelligence solution from Nanox.AI that allows for automated spine analysis using CT images was evaluated against a consensus ground truth established by two radiologists, including a senior musculoskeletal radiologist. Two vertebral body height reduction thresholds were tested: mild (>20%) and moderate (>25%). Original radiologist reports were reviewed to identify missed VCFs. Results: At the 20% threshold, the AI achieved a sensitivity of 92.0%, a specificity of 52.7%, a PPV of 16.5%, and an NPV of 98.5%. At the 25% threshold, sensitivity decreased to 78.0%, while specificity improved to 94.2%, with a PPV of 51.1% and an NPV of 98.2%. The AI identified 88% and 92% of fractures missed by radiologists at the 20% and 25% thresholds, respectively. Conclusions: The Nanox HealthOST AI solution demonstrates potential as an effective screening tool, with threshold selection adaptable to clinical needs with a secondary review by a radiologist that is advisable to ensure diagnostic accuracy. The study further indicates that radiologists often overlook VCFs in reporting non-indicated cases and that AI has a role in enhancing the detection and reporting of vertebral compression fractures in routine clinical practice. Full article

Out-of-hospital cardiac arrest (OHCA) mortality remains alarmingly high in most countries. The majority of pharmacological attempts to improve outcomes have failed. Randomized trials have shown limited survival benefits with vasopressin, fibrinolysis, amiodarone, or lidocaine. Even the benefits of adrenaline remain a matter of debate. In this context, relying on technology may seem appealing. However, technological strategies have also yielded disappointing results. This is exemplified by automated external chest compression devices. When first introduced, theoretical models, animal studies, and early clinical trials suggested they could improve survival. Yet, randomized trials failed to confirm this benefit. Similarly, to date, extracorporeal membrane oxygenation (ECMO), therapeutic hypothermia, and primary angioplasty have demonstrated inconsistent survival advantage. Other technological innovations continue to be explored, such as artificial intelligence to improve the diagnosis of cardiac arrest during emergency calls, mobile applications to dispatch citizen responders to patients in cardiac arrest, geolocation of defibrillators, and even the delivery of defibrillators via drones. Nevertheless, it is clear that the focus and investment should prioritize the initial links in the chain of survival: early alerting, chest compressions, and defibrillation. Significant improvements in these critical steps can be achieved through the education of children. Modern technological tools must be leveraged to enhance this training by incorporating gamification and democratizing access to education. These strategies hold the potential to fundamentally improve the management of cardiac arrest. Full article

Background: There is growing interest in the quality of manual ventilation during cardiopulmonary resuscitation (CPR), but accurate assessment of ventilation parameters remains a challenge. Waveform capnography is currently the reference for monitoring ventilation rate in intubated patients, but fails to provide information on tidal volumes and inspiration–expiration timing. Moreover, the capnogram is often distorted when chest compressions (CCs) are performed during ventilation compromising its reliability during CPR. Our main purpose was to characterize manual ventilation during CPR and to assess how CCs may impact on ventilation quality. Methods: Retrospective analysis were performed of CPR recordings fromtwo databases of adult patients in cardiac arrest including capnogram, compression depth, and airway flow, pressure and volume signals. Using automated signal processing techniques followed by manual revision, individual ventilations were identified and ventilation parameters were measured. Oscillations on the capnogram plateau during CCs were characterized, and its correlation with compression depth and airway volume was assessed. Finally, we identified events of reversed airflow caused by CCs and their effect on volume and capnogram waveform. Results: Ventilation rates were higher than the recommended 10 breaths/min in 66.7% of the cases. Variability in ventilation rates correlated with the variability in tidal volumes and other ventilatory parameters. Oscillations caused by CCs on capnograms were of high amplitude (median above 74%) and were associated with low pseudo-volumes (median 26 mL). Correlation between the amplitude of those oscillations with either the CCs depth or the generated passive volumes was low, with correlation coefficients of −0.24 and 0.40, respectively. During inspiration and expiration, reversed airflow events caused opposed movement of gases in 80% of ventilations. Conclusions: Our study confirmed lack of adherence between measured ventilation rates and the guideline recommendations, and a substantial dispersion in manual ventilation parameters during CPR. Oscillations on the capnogram plateau caused by CCs did not correlate with compression depth or associated small tidal volumes. CCs caused reversed flow during inspiration, expiration and in the interval between ventilations, sufficient to generate volume changes and causing oscillations on capnogram. Further research is warranted to assess the impact of these findings on ventilation quality during CPR. Full article

Background: Chest compression and defibrillation are essential components of cardiac arrest treatment. Mechanical chest compression devices (MCCD) and automated external defibrillators (AED) are used separately in clinical practice. We developed an automated compression–defibrillation apparatus (ACDA) that performs mechanical chest compression and automated defibrillation. We investigated the performance of cardiopulmonary resuscitation (CPR) with automatic CPR (A-CPR) compared to that with MCCD and AED (conventional CPR: C-CPR). Methods: Pigs were randomized into A-CPR or C-CPR groups: The A-CPR group received CPR+ACDA, and the C-CPR group received CPR+MCCD+AED. Hemodynamic parameters, outcomes, and time variables were measured. During a simulation study, healthcare providers performed a basic life support scenario for manikins with an ACDA, MCCD, and AED, and time variables and chest compression parameters were measured. Results: The animals showed no significant in hemodynamic effects, including aortic pressures, coronary perfusion pressure, carotid blood flow, and end-tidal CO₂, and resuscitation outcomes between the two groups. In both animal and simulation studies, the time to defibrillation, time to chest compression, and hands-off time were significantly shorter in the A-CPR group than those in the C-CPR group. Conclusions: CPR using ACDA showed similar hemodynamic effects and resuscitation outcomes as CPR using AED and MCCD separately, with the advantages of a reduction in the time to compression, time to defibrillation, and hands-off time. Full article

Background: The Lund University Cardiopulmonary Assist System-2/-3 was developed for auto-matic chest compressions during cardiopulmonary resuscitation (mechanical CPR or MCPR) and often allows a patient suffering from cardiac arrest to be taken to the cardiac catheterization room. We report the clinical outcomes of percutaneous coronary interventions (PCI) performed in cardiac arrest patients under automatic MCPR devices. Methods: We retrieved all patients with cardiac arrest who were referred to PCI under MCPR devices from the Cardio-FR database (003-REP-CER-FR) from January 2016 to December 2021. Patients who were hemodynamically stable at the time of coronary examination/intervention (even those who had been resuscitated immediately before) were excluded from the analysis. Baseline patient and procedure characteristics were collected. The primary outcome was the return of spontaneous circulation (ROSC). Results: Of all patients who were on MCPR at the cardiac catheterization room, eleven still required active CPR during coronary examination/intervention and were included in the analysis. Mean age was 67.9 ± 10 years, nine were male. The MCPR device was initiated on average after 8.5 ± 8.1 minutes. All patients had ventricular defibrillation and received an average of 3.4 ± 3.6 shocks and 82% adrenaline boluses. The MCPR was used for an average of 51.1 ± 34.4 minutes. Total resuscitation time was on average 59.6 ± 38.3 minutes. Of the eleven patients, nine underwent ad hoc PCI. ROSC was achieved in four patients after 36.5 ± 49.8 minutes. The survival was 36% (four patients) at 24 hours and 27% (three patients) at three months. Only one of the patients resuscitated for >25 minutes survived. Patients with in-hospital cardiac arrest were associated with shorter ROSC (p <0.01), shorter resuscitation time (p = 0.009) and better survival (p = 0.03) than patients with out-of-hospital cardiac arrest. Conclusions: MCPR allows patients in cardiac arrest to reach the cardiac catheterization room. However, the prognosis is grim with high mortality. Only one patient survived after >25 minutes of mechanical resuscitation. Full article

Objective: We assessed the effectiveness of automated chest compression devices depending on the time of admission based on the frequency of iatrogenic chest injuries, the duration of in-hospital resuscitation efforts, and clinical outcomes among out-of-hospital cardiac arrest (OHCA) patients. Methods: We conducted a retrospective historical control study of OHCA patients in Japan between 2015–2022. The patients were divided according to time of admission, where day-time was considered 07:00–22:59 and night-time 23:00–06:59. These patients were then divided into two categories based on the in-hospital cardiopulmonary resuscitation (IHCPR) device: manual chest compression (mCC) group and automatic chest compression devices (ACCD) group. We used univariate and multivariate ordered logistic regression models adjusted for pre-hospital confounders to evaluate the impact of ACCD use during IHCPR on outcomes (IHCPR duration, CPR-related chest injuries, and clinical outcomes) in the day-time and night-time groups. Results: Among 1101 patients with OHCA (day-time, 809; night-time, 292), including 215 patients who underwent ACCD during IHCPR in day-time (26.6%) and 104 patients in night-time group (35.6%), the multivariate model showed a significant association of ACCD use with the outcomes of in-hospital resuscitation and higher rates of return in spontaneous circulation, lower incidence of CPR-related chest injuries, longer in-hospital resuscitation durations, greater survival to Emergency Department and hospital discharge, and greater survival with good neurological outcome to hospital discharge, though only in the night-time group. Conclusions: Patients who underwent ACCD during in-hospital resuscitation at night had a significantly longer duration of in-hospital resuscitation, a lower incidence of CPR-related chest injuries, and better outcomes. Full article

The spread of infectious diseases has accelerated the transition from face-to-face (F2F) to non-F2F (NF2F) education. To maintain the effect of successful NF2F education in cardiopulmonary resuscitation, reorganizing the curriculum to suit the NF2F educational environment is necessary. We propose an appropriate learning curriculum for NF2F basic life support (BLS) training for laypersons based on expert surveys and learners’ performance outcomes. This study included three stages and used multiple methods. A draft curriculum was created through a literature review and three-round Delphi approach, and then applied as a test for actual education. After the training, the final curriculum of the NF2F BLS training for laypersons was proposed by reflecting on the performance outcomes of learners and expert opinions. NF2F theoretical education was simplified into five content items: concept of chain of survival, legal protection for first aiders, importance of bystander cardiopulmonary resuscitation, how to recognize a patient in cardiac arrest and activate the emergency medical services system, and reduced training time. In the hands-on skills session, it was recommended to practice chest compressions using a simple intuitive feedback device and to use automated external defibrillators step-by-step more than in F2F training. In conclusion, NF2F training is a suitable option for BLS training methods in situations where F2F training is difficult. Full article

This study aims to present a novel deep learning algorithm for a sliding shock advisory decision during cardiopulmonary resuscitation (CPR) and its performance evaluation as a function of the cumulative hands-off time. We retrospectively used 13,570 CPR episodes from out-of-hospital cardiac arrest (OHCA) interventions reviewed in a period of interest from 30 s before to 10 s after regular analysis of automated external defibrillators (AEDs). Three convolutional neural networks (CNNs) with raw ECG input (duration of 5, 10, and 15 s) were applied for the shock advisory decision during CPR in 26 sequential analyses shifted by 1 s. The start and stop of chest compressions (CC) can occur at arbitrary times in sequential slides; therefore, the sliding hands-off time (sHOT) quantifies the cumulative CC-free portion of the analyzed ECG. An independent test with CPR episodes in 393 ventricular fibrillations (VF), 177 normal sinus rhythms (NSR), 1848 other non-shockable rhythms (ONR), and 3979 asystoles (ASYS) showed a substantial improvement of VF sensitivity when increasing the analysis duration from 5 s to 10 s. Specificity was not dependent on the ECG analysis duration. The 10 s CNN model presented the best performance: 92–94.4% (VF), 92.2–94% (ASYS), 96–97% (ONR), and 98.2–99.5% (NSR) for sliding decision times during CPR; 98–99% (VF), 98.2–99.8% (ASYS), 98.8–99.1 (ONR), and 100% (NSR) for sliding decision times after end of CPR. We identified the importance of sHOT as a reliable predictor of performance, accounting for the minimal sHOT interval of 2–3 s that provides a reliable rhythm detection satisfying the American Heart Association (AHA) standards for AED rhythm analysis. The presented technology for sliding shock advisory decision during CPR achieved substantial performance improvement in short hands-off periods (>2 s), such as insufflations or pre-shock pauses. The performance was competitive despite 1–2.8% point lower ASYS detection during CPR than the standard requirement (95%) for non-noisy ECG signals. The presented deep learning strategy is a basis for improved CPR practices involving both continuous CC and CC with insufflations, associated with minimal CC interruptions for reconfirmation of non-shockable rhythms (minimum hands-off time) and early treatment of VF (minimal pre-shock pauses). Full article

Background: Police first responder systems also including automated external defibrillation (AED) has in the past shown considerable impact on favourable outcomes after out-of-hospital cardiac arrest (OHCA). While short hands-off times in chest compressions are known to be beneficial, various AED models use different algorithms, inducing longer or shorter durations of crucial timeframes along basic life support (BLS). Yet, data on details of these differences, and also of their potential impact on clinical outcomes are scarce. Methods: For this retrospective observational study, patients with OHCA of presumed cardiac origin and initially shockable rhythm treated by police first responders in Vienna, Austria, between 01/2013 and 12/2021 were included. Data from the Viennese Cardiac Arrest Registry and AED files were extracted, and exact timeframes were analyzed. Results: There were no significant differences in the 350 eligible cases in demographics, return of spontaneous circulation, 30-day survival, or favourable neurological outcome between the used AED types. However, the Philips HS1 and -FrX AEDs showed immediate rhythm analysis after electrode placement (0 [0–1] s) and almost no shock loading time (0 [0–1] s), as opposed to the LP CR Plus (3 [0–4] and 6 [6–6] s, respectively) and LP 1000 (3 [2–10] and 6 [5–7] s, respectively). On the other hand, the HS1 and -FrX had longer analysis times of 12 [12–16] and 12 [11–18] s than the LP CR Plus (5 [5–6] s) and LP 1000 (6 [5–8] s). The duration from when the AED was turned on until the first defibrillation were 45 [28–61] s (Philips FrX), 59 [28–81] s (LP 1000), 59 [50–97] s (HS1), and 69 [55–85] s (LP CR Plus). Conclusion: In a retrospective analysis of OHCA-cases treated by police first responders, we could not find significant differences in clinical patient outcomes concerning the respective used AED model. However, various differences in time durations (e.g., electrode placement to rhythm analysis, analysis duration, or AED turned on until first defibrillation) along the BLS algorithm were seen. This opens up the question of AED-adaptations and tailored training methods for professional first responders. Full article