Earlier reviews of machine learning for mortality after cardiac surgery pooled deep learning (DL) with shallow and non-neural methods. We asked whether models meeting an explicit definition of DL discriminate short-term mortality better than EuroSCORE II, the Society of Thoracic Surgeons Predicted Risk of Mortality (STS-PROM) or logistic regression in the same patients. This re-conducted review (PROSPERO CRD420261354546) supersedes an earlier preprint. MEDLINE, Europe PMC, Lens.org and IEEE Xplore were searched on 27–28 September 2026, with backward citation searching; Embase, CINAHL and CENTRAL were not searched, and forward citation searching was not performed. DL required at least two hidden layers, or a named architecture of established depth, shown in the source. A large language model pre-screened records and extracted data; two reviewers verified every decision and value. Risk of bias was assessed with PROBAST + AI, and results were synthesised without meta-analysis as within-cohort differences in the area under the curve (AUC). Of 1,845 records identified, 515 duplicates were removed and 1,330 were screened; of 76 reports sought, 54 were assessed in full and 22 could not be retrieved. Four studies (five cohort-level comparisons) met all criteria. A chest-radiograph convolutional network had lower AUCs than STS-PROM in temporal (0.829 vs. 0.884) and external (0.738 vs. 0.803) cohorts with about two dozen deaths each; two-hidden-layer perceptrons did not outperform logistic regression (0.761 vs. 0.762; 0.77 vs. 0.80); one study reported AUCs only graphically. All four studies had high or unclear risk of bias. Fifteen reports described networks of undetermined depth, including a tabular transformer that outperformed EuroSCORE II internally (0.834 vs. 0.807); in the 12 with extractable comparisons, the network had a higher AUC than the best comparator in 4 and lower in 8 (median difference − 0.004). Evidence that DL improves on established scores for short-term mortality after adult cardiac surgery is sparse, inconsistent and at high or unclear risk of bias, and the search was incomplete relative to the registered protocol. Differences in discrimination were small or absent, and calibration and clinical utility were rarely evaluated. Superiority cannot be established. This review was registered on PROSPERO, the International Prospective Register of Systematic Reviews, on 29 March 2026 (CRD420261354546; https://www.crd.york.ac.uk/prospero/). At registration, the record stated that searching and screening for an earlier version of the review had been completed and that data extraction had not started. That earlier version was posted as a preprint on 31 March 2026 (Preprints.org, doi:10.20944/preprints202603.2465.v1); it is superseded by this report. Its withdrawal was requested from Preprints.org on 18 October 2026 and had not been confirmed at the time of resubmission. The review reported here was re-conducted in full after registration: the searches were run on 27–28 September 2026, and screening, data extraction and risk-of-bias assessment followed. The registered eligibility criteria were not changed in substance; the registered deep-learning criterion was operationalised before full-text assessment and, after peer review, applied more strictly (a post hoc change reported in the manuscript). The search sources, risk-of-bias tool and synthesis method differ from the registered protocol. Every difference, with its timing and reason, is reported in Table 1 of the manuscript, and an amendment describing them has been submitted to PROSPERO.
We aimed to perform a meta-analysis to evaluate the effect of intraoperative hemoadsorption (HA) with cardiopulmonary bypass (CPB) on postoperative inflammatory biomarkers in patients undergoing cardiac surgery. We searched PubMed, Web of Science, and Scopus databases for randomized controlled trials (RCTs) that compared intraoperative HA with CPB versus CPB without intraoperative HA in patients undergoing cardiac surgery in terms of postoperative inflammatory biomarkers. Twelve RCTs involving 830 patients (420 in HA group and 410 control group) were included. HA was associated with significantly lower interleukin-6 (IL-6) levels at 6 h (SMD = − 3.33, 95
Abstract Background Non-tracheally intubated (NTI) robot-assisted thoracic surgery (RATS) for lung cancer resection has been only anecdotally reported, and evidence regarding feasibility is still lacking. This study aims at evaluating feasibility of NTI-RATS anatomic lung resection for early-stage lung cancer. Results This retrospective single-institution feasibility study of prospectively collected data included patients undergoing anatomic NTI-RATS lung resections. The study was approved by our institutional ethical committee. All procedures included careful patient selection, thoracic epidural anesthesia, temporary phrenic and vagal nerve blocks and 4-port RATS approach. Feasibility was defined as the possibility to accomplish the whole procedure by RATS without need of conversion to open-surgery or trachealintubation with double lumen intubation according to predefined criteria. Perioperative blood gases and cerebral oxygen saturation were evaluated at predefined time points whereas recovery time was calculated according to standardized criteria. Post-operative protocol allowed to start feeding and mobilizing within 4 h after surgery. Between January and July 2026, 15 consecutive NTI-RATS anatomic lung resections were performed (12 lobectomies and 3 segmentectomies) with only one conversion to general anesthesia due to excessive mediastinal movement. We documented a median intra-operative PaCO₂ increment (+56.1% p<0.001), which returned close to baseline values 30 minutes after awakening. No patient met the criteria for cerebral desaturation throughout the procedure despite a slight median decrement (-10.1%, p<0.001) at the onset of surgical pneumothorax. Resection with free margins was achieved in all cases. Early feeding and mobilization were possible in all instances. There was no 30-day mortality and only one minor postoperative morbidity due to atrial fibrillation. Conclusions NTI-RATS anatomic lung resection proved feasible in the vast majority (93.3%, 14/15, 95% confidence interval 68.1%-99.8%) in a highly selected population of patients with early-stage NSCLC.
Abstract Background The clamp-on technique for acute type A aortic dissection (ATAAD) repair has gained traction as an alternative to open distal anastomosis (ODA), but procedure-specific evidence comparing these strategies remains limited. This study aimed to compare surgical safety, postoperative morbidity, and organ protection between the clamp-on technique and conventional ODA. Methods PubMed, Embase, and Cochrane Library were systematically searched through May 2026 for studies comparing clamp-on versus ODA in adults with ATAAD. Random-effects models with Mantel-Haenszel weighting were used to estimate pooled risk ratios (RRs) for binary endpoints. Confidence intervals were adjusted using the Hartung-Knapp-Sidik-Jonkman (HKSJ) method, and 95% prediction intervals were calculated. Results Nine observational studies (2,032 patients) were included, of which eight contributed crude event data to the quantitative synthesis, whereas the propensity-matched IRAD analysis (Pupovac et al.) was addressed narratively because its event counts derive from multiply imputed matched cohorts of varying size (94–101 pairs). No significant differences were observed between groups in hospital mortality (RR 1.26; 95% CI 0.83 to 1.92) or postoperative stroke (RR 0.98; 95% CI 0.41 to 2.35). Risks did not differ significantly for distal reoperation (RR 1.41; 95% CI 0.34 to 5.80) or new-onset dialysis (RR 0.78; 95% CI 0.42 to 1.45). The clamp-on strategy was associated with a non-significant trend toward lower acute kidney injury (RR 0.76; 95% CI 0.52 to 1.11; three studies) and sepsis (RR 0.64; 95% CI 0.33 to 1.26; three studies). Conclusions In patients undergoing surgical repair for acute type A aortic dissection, clamp-on repair and open distal anastomosis showed no detectable difference in hospital mortality or stroke, although confidence intervals remain compatible with clinically important benefit or harm in either direction. The association of clamp-on with lower acute kidney injury and sepsis did not remain statistically significant with Hartung-Knapp-Sidik-Jonkman-adjusted confidence intervals and should be regarded as hypothesis-generating. Given the very low certainty of evidence and inherent selection biases, these findings support a patient-tailored approach over a standardized strategy and underscore the need for prospective multicenter registries.
Abstract Background Reported incidences of postoperative diaphragmatic dysfunction vary widely, largely because diagnostic criteria differ and preoperative imaging is seldom available; whether anatomical elevation, contractile impairment, or their combination carries clinical consequence is undefined, and the reproducibility of any composite definition has not been established. We prospectively studied 169 adults undergoing cardiac surgery, pairing standardised pre- and postoperative chest radiography with quantitative diaphragmatic ultrasonography performed in every patient, to determine the incidence, reliability, predictors and clinical consequences of dysfunction defined as new anatomical elevation combined with contractile failure. Results Combined dysfunction occurred in 18.3% of patients, isolated radiographic elevation in 3.6% and isolated ultrasonographic impairment in 10.7%. Measurement reliability was high (intra-observer ICC 0.886–0.952; inter-observer 0.832–0.901), and the combined criterion reproduced the phenotype far more consistently than ultrasonography alone (intra- and inter-observer agreement 100% and 96.7% versus 83.3% and 70.0%). Redefining dysfunction against each patient’s own preoperative baseline reproduced the classification in 98.2% of patients (κ = 0.938), supporting the literature-derived thresholds in this population. Coronary grafting with left internal mammary artery harvesting — a combined exposure these data cannot partition — was independently associated with dysfunction (adjusted odds ratio 3.95, 95% confidence interval 1.32–11.82), as was diabetes mellitus (3.09, 1.26–7.56); dysfunction was left-sided in 28 of 31 cases. After adjustment, affected patients required 1.45 times longer mechanical ventilation (95% CI 1.06–1.98) and 1.41 times longer hospitalisation (1.13–1.76), with a smaller, non-significant difference in intensive care stay; in absolute terms these amounted to approximately one additional hour of ventilation and one additional day in hospital. Prolonged ventilation and hospitalisation were concentrated in the combined-criterion group, and 89.7% of affected survivors regained normal function by three months. Conclusions Anchoring radiography to an individual preoperative baseline and requiring ultrasonographic confirmation identifies diaphragmatic injury more specifically, and more reproducibly, than either modality alone, and its thresholds are concordant with a within-patient definition. Mammary-artery grafting and diabetes mark mechanical–ischaemic and metabolic susceptibility, supporting phrenic-protective technique and perioperative glycaemic optimisation as testable interventions.
