Retrospective Dataset Research Articles

Multiregional dynamic contrast-enhanced MRI-based integrated system for predicting pathological complete response of axillary lymph node to neoadjuvant chemotherapy in breast cancer: multicentre study

The accurate evaluation of axillary lymph node (ALN) response to neoadjuvant chemotherapy (NAC) in breast cancer holds great value. This study aimed to develop an artificial intelligence system utilising multiregional dynamic contrast-enhanced MRI (DCE-MRI) and clinicopathological characteristics to predict axillary pathological complete response (pCR) after NAC in breast cancer. This study included retrospective and prospective datasets from six medical centres in China between May 2018 and December 2023. A fully automated integrated system based on deep learning (FAIS-DL) was built to perform tumour and ALN segmentation and axillary pCR prediction sequentially. The predictive performance of FAIS-DL was assessed using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. RNA sequencing analysis were conducted on 45 patients to explore the biological basis of FAIS-DL. 1145 patients (mean age, 50 years±10 [SD]) were evaluated. Among these patients, 506 were in the training and validation sets (axillary pCR rate of 40.3%), 127 in the internal test set (axillary pCR rate of 37.8%), 414 in the pooled external test set (axillary pCR rate of 48.8%), and 98 in the prospective test set (axillary pCR rate of 43.9%). For predicting axillary pCR, FAIS-DL achieved AUCs of 0.95, 0.93, and 0.94 in the internal test set, pooled external test set, and prospective test set, respectively, which were also significantly higher than those of the clinical model and deep learning models based on single-regional DCE-MRI (all P<0.05, DeLong test). In the pooled external and prospective test sets, the FAIS-DL decreased the unnecessary axillary lymph node dissection rate from 47.9% to 6.8%, and increased the benefit rate from 52.2% to 86.5%. RNA sequencing analysis revealed that high FAIS-DL scores were associated with the upregulation of immune-mediated genes and pathways. FAIS-DL has demonstrated satisfactory performance in predicting axillary pCR, which may guide the formulation of personalised treatment regimens for patients with breast cancer in clinical practice. This study was supported by the National Natural Science Foundation of China (82371933), National Natural Science Foundation of Shandong Province of China (ZR2021MH120), Mount Taishan Scholars and Young Experts Program (tsqn202211378), Key Projects of China Medicine Education Association (2022KTM030), China Postdoctoral Science Foundation (314730), and Beijing Postdoctoral Research Foundation (2023-zz-012).

Applying Machine Learning for Prescriptive Support: A Use Case with Unfractionated Heparin in Intensive Care Units.

Continuous unfractionated heparin is widely used in intensive care, yet its complex pharmacokinetic properties complicate the determination of appropriate doses. To address this challenge, we developed machine learning models to predict over- and under-dosing, based on anti-Xa results, using a monocentric retrospective dataset. The random forest model achieved a mean AUROC of 0.80 [0.77-0.83], while the XGB model reached a mean AUROC of 0.80 [0.76-0.83]. Feature importance was employed to enhance the interpretability of the model, a critical factor for clinician acceptance. After prospective validation, machine learning models such as those developed in this study could be implemented within a computerized physician order entry (CPOE) as a clinical decision support system (CDSS).

Machine Learning-Based Identification of Risk Factors of Keratoconus Progression Using Raw Corneal Tomography Data.

The purpose of this study was to identify early indicators of keratoconus progression in Pentacam data using machine learning (ML) techniques. A retrospective Pentacam tabular data set was created by retrieving 11,760 tomography tests performed in patients with keratoconus. Data for eyes labeled unstable based on their referral for cross-linking were differentiated from data for eyes labeled stable and not referred for follow-up procedures. A boosted decision tree was trained on the final data set using a cross-validation method. The final labeled data set included 1218 tomography tests. Training a ML model on a single test for each eye did not accurately predict disease progression, as indicated by the mean receiver-operating characteristic area under the curve of 0.59 ± 0.1, with precision of 0.27, recall of 0.3, and F1 score of 0.28. Training on serial tests for each eye included 819 tomography scans and yielded good prognostic abilities: a receiver-operating characteristic area under the curve of 0.75 ± 0.07, precision of 0.32, recall of 0.67, and F1 score of 0.43. In addition, 4 of the 55 Pentacam raw data parameters predominantly used the algorithm decision: age, central keratoconus index, Rs B, and D10 mm pachy. This study revealed specific dominant parameters attributing to the classification of stability, which are not routinely assessed in determining progression in common practice. Using ML techniques, keratoconus deterioration was evaluated algorithmically with training on multiple tests, yet was not predicted by a single tomography test. Hence, our study highlights novel factors to the current consideration of cross-linking referral and may serve as a supportive tool for clinicians.

Pooled analysis of multiple sclerosis findings on multisite 7 Tesla MRI: Protocol and initial observations.

Although 7 T MRI research has contributed much to our understanding of multiple sclerosis (MS) pathology, most prior data has come from small, single-center studies with varying methods. In order to truly know if such findings have widespread applicability, multicenter methods and studies are needed. To address this, members of the North American Imaging in MS (NAIMS) Cooperative worked together to create a multicenter collaborative study of 7 T MRI in MS. In this manuscript, we describe the methods we have developed for the purpose of pooling together a large, retrospective dataset of 7 T MRIs acquired in multiple MS studies at five institutions. To date, this group has contributed five-hundred and twenty-eight 7 T MRI scans from 350 individuals with MS to a common data repository, with plans to continue to increase this sample size in the coming years. We have developed unified methods for image processing for data harmonization and lesion identification/segmentation. We report here our initial observations on intersite differences in acquisition, which includes site/device differences in brain coverage and image quality. We also report on the development of our methods and training of image evaluators, which resulted in median Dice Similarity Coefficients for trained raters' annotation of cortical and deep gray matter lesions, paramagnetic rim lesions, and meningeal enhancement between 0.73 and 0.82 compared to final consensus masks. We expect this publication to act as a resource for other investigators aiming to combine multicenter 7 T MRI datasets for the study of MS, in addition to providing a methodological reference for all future analysis projects to stem from the development of this dataset.

A stochastic model-based control methodology for glycemic management in the intensive care unit

Introduction: Intensive care unit (ICU) patients exhibit erratic blood glucose (BG) fluctuations, including hypoglycemic and hyperglycemic episodes, and require exogenous insulin delivery to keep their BG in healthy ranges. Glycemic control via glycemic management (GM) is associated with reduced mortality and morbidity in the ICU, but GM increases the cognitive load on clinicians. The availability of robust, accurate, and actionable clinical decision support (CDS) tools reduces this burden and assists in the decision-making process to improve health outcomes. Clinicians currently follow GM protocol flow charts for patient intravenous insulin delivery rate computations.Methods: We present a mechanistic model-based control algorithm that estimates the optimal intravenous insulin rate to keep BG within a target range; the goal is to develop this approach for eventual use within CDS systems. In this control framework, we employed a stochastic model representing BG dynamics in the ICU setting and used the linear quadratic Gaussian control methodology to develop a controller.Results: We designed two experiments, one using virtual (simulated) patients and one using a real-world retrospective dataset. Using these, we evaluated the safety and efficacy of this model-based glycemic control methodology. The presented controller avoids hypoglycemia and hyperglycemia in virtual patients, maintaining BG levels in the target range more consistently than two existing GM protocols. Moreover, this methodology could theoretically prevent a large proportion of hypoglycemic and hyperglycemic events recorded in a real-world retrospective dataset.Discussion: The current version of the methodology shows potential usefulness in GM of ICU patients. However, it is limited to a subgroup of the ICU patient population, who are fed through and enteral tube and delivered intravenous insulin. After extending to a broader ICU patient population who can consume oral nutrition and are delivered subcutaneous insulin for GM, the methodology could be tested with pilot studies and clinical trials for eventual use as a CDS tool.

Potential anatomical triggers for plan adaptation of cervical cancer external beam radiotherapy.

This study aimed to identify potential anatomical variation triggers using magnetic resonance imaging for plan adaption of cervical cancer patients to ensure dose requirements were met over an external beam radiotherapy course. Magnetic resonance images (MRIs) acquired before and during treatment were rigidly registered to a pre-treatment computerised tomography (CT) image for 11 retrospective cervix cancer datasets. Target volumes (TVs) and organs at risk (OARs) were delineated on both MRIs and propagated onto the CT. Treatment plans were generated based on the pre-treatment contours and applied to the mid-treatment contours. Anatomical and dosimetric changes between each timepoint were assessed. The anatomical changes included the change in centroid position and volume size. Dosimetric changes included the V30Gy and V40Gy for the OARs, and V95%, V100%, D95% and D98% for the TVs. Correlation with dosimetric and anatomical changes were assessed to determine potential replan triggers. Changes in the bowel volume and position in the superior-inferior direction, and the high-risk CTV anterior posterior position were highly correlated with a change in dose to the bowel and target, respectively. Hence changes in bowel and high-risk CTV could be used as a potential replan triggers.

Real-life experience with sorafenib for advanced and refractory desmoid-type fibromatosis.

In recent years, there has been a change in the therapeutic landscape of desmoid-type fibromatosis (DF). Watchful waiting is now preferred over initial local treatments such as surgery and radiotherapy. Systemic treatment is considered for progressive or symptomatic disease. The aim of this study is to review real-life data on the use of sorafenib in DF. We established a retrospective dataset of patients treated with sorafenib in our centre, Ghent University Hospital, for progressive DF. Patient demographics, disease characteristics, response to therapy using Response Evaluation Criteria in Solid Tumours 1.1 criteria and toxicity according to CTCAE v5.0 were assessed. Eleven patients with DF were treated with sorafenib between 2020 and 2024. Median treatment duration was 20.4 months (95% confidence interval [CI], 10.0-NR). 36.4% achieved partial response, 54.5% stable disease and 9.1% progressive disease. For three patients, the treatment is ongoing. The median time to objective response rate is 15.0 months (95% CI, 8.8-NR). The majority (81.8%) experienced grade 2 toxicity, and one third of patients grade 3 toxicity (36.4%). The most common all-grade adverse event was skin toxicity (hand-foot syndrome, pruritus and rash) (90.9%). Nine patients (81.8%) needed dose reduction with a median time to first reduction of 1.1 months (95% CI, 0.5-NR). One patient stopped treatment due to toxicity. Real-life data on the use of sorafenib in the treatment of DF is consistent with published data in clinical trial setting. Sorafenib is an effective treatment option for progressive DF although associated with significant toxicity and the need for rapid dose reduction.

Liver vessel MRI image segmentation based on dual-path diffusion model

Accurate segmentation of liver blood vessels in magnetic resonance imaging (MRI) images is a challenging task due to the complex tree-like structure and anisotropic diffusion properties of blood vessels. To solve this problem, we propose a new Dual-Path Diffusion Model (DPDM) framework. The framework consists of two collaborative diffusion paths: a local feature learning path based on convolution operations and a global context modeling path based on transform blocks. Local path encodes rich shape priors and preserve spatial details, while global path captures long-distance dependencies and enhance representation. In the decoding phase, the boundary features from the boundary path are fused with the features of the ordinary path decoding, which further enhances the shape sensitivity. In addition, we leverage a multi-task learning scheme to jointly optimize vascular segmentation and boundary prediction tasks in an end-to-end manner. Experiments on retrospective clinical dataset demonstrate that the proposed DPDM framework achieves excellent performance on the liver vessel segmentation task. Compared with state-of-the-art methods, our approach achieved a 6.0% and 7.3% performance improvement in Dice coefficient and IoU index, respectively. Our approach offers a promising solution for automated blood vessel segmentation in precision medicine.

Implementing AI in Hospitals to Achieve a Learning Health System: Systematic Review of Current Enablers and Barriers.

Efforts are underway to capitalize on the computational power of the data collected in electronic medical records (EMRs) to achieve a learning health system (LHS). Artificial intelligence (AI) in health care has promised to improve clinical outcomes, and many researchers are developing AI algorithms on retrospective data sets. Integrating these algorithms with real-time EMR data is rare. There is a poor understanding of the current enablers and barriers to empower this shift from data set-based use to real-time implementation of AI in health systems. Exploring these factors holds promise for uncovering actionable insights toward the successful integration of AI into clinical workflows. The first objective was to conduct a systematic literature review to identify the evidence of enablers and barriers regarding the real-world implementation of AI in hospital settings. The second objective was to map the identified enablers and barriers to a 3-horizon framework to enable the successful digital health transformation of hospitals to achieve an LHS. The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines were adhered to. PubMed, Scopus, Web of Science, and IEEE Xplore were searched for studies published between January 2010 and January 2022. Articles with case studies and guidelines on the implementation of AI analytics in hospital settings using EMR data were included. We excluded studies conducted in primary and community care settings. Quality assessment of the identified papers was conducted using the Mixed Methods Appraisal Tool and ADAPTE frameworks. We coded evidence from the included studies that related to enablers of and barriers to AI implementation. The findings were mapped to the 3-horizon framework to provide a road map for hospitals to integrate AI analytics. Of the 1247 studies screened, 26 (2.09%) met the inclusion criteria. In total, 65% (17/26) of the studies implemented AI analytics for enhancing the care of hospitalized patients, whereas the remaining 35% (9/26) provided implementation guidelines. Of the final 26 papers, the quality of 21 (81%) was assessed as poor. A total of 28 enablers was identified; 8 (29%) were new in this study. A total of 18 barriers was identified; 5 (28%) were newly found. Most of these newly identified factors were related to information and technology. Actionable recommendations for the implementation of AI toward achieving an LHS were provided by mapping the findings to a 3-horizon framework. Significant issues exist in implementing AI in health care. Shifting from validating data sets to working with live data is challenging. This review incorporated the identified enablers and barriers into a 3-horizon framework, offering actionable recommendations for implementing AI analytics to achieve an LHS. The findings of this study can assist hospitals in steering their strategic planning toward successful adoption of AI.

A Deep Learning-Based Framework for Predicting Intracerebral Hemorrhage Hematoma Expansion Using Head Non-contrast CT Scan

Rationale and ObjectivesHematoma expansion (HE) in intracerebral hemorrhage (ICH) is a critical factor affecting patient outcomes, yet effective clinical tools for predicting HE are currently lacking. We aim to develop a fully automated framework based on deep learning for predicting HE using only clinical non-contrast CT (NCCT) scans. Materials and MethodsA large retrospective dataset (n=2,484) was collected from 84 centers, while a prospective dataset (n=500) was obtained from 26 additional centers. Baseline NCCT scans and follow-up NCCT scans were conducted within 6 hours and 48 hours from symptom onset, respectively. HE was defined as a volume increase of more than 6 mL on the follow-up NCCT. The retrospective dataset was divided into a training set (n=1,876) and a validation set (n=608) by patient inclusion time. A two-stage framework was trained to predict HE, and its performance was evaluated on both the validation and prospective sets. Receiver operating characteristics area under the curve (AUC), sensitivity, and specificity were leveraged. ResultsOur two-stage framework achieved an AUC of 0.760 (95% CI 0.724-0.799) on the retrospective validation set and 0.806 (95% CI 0.750-0.859) on the prospective set, outperforming the commonly used BAT score, which had AUCs of 0.582 and 0.699, respectively. ConclusionOur framework can automatically and robustly identify ICH patients at high risk of HE using admission head NCCT scans, providing more accurate predictions than the BAT score.

Improving lung nodule segmentation in thoracic CT scans through the ensemble of 3D U-Net models.

The current study explores the application of 3D U-Net architectures combined with Inception and ResNet modules for precise lung nodule detection through deep learning-based segmentation technique. This investigation is motivated by the objective of developing a Computer-Aided Diagnosis (CAD) system for effective diagnosis and prognostication of lung nodules in clinical settings. The proposed method trained four different 3D U-Net models on the retrospective dataset obtained from AIIMS Delhi. To augment the training dataset, affine transformations and intensity transforms were utilized. Preprocessing steps included CT scan voxel resampling, intensity normalization, and lung parenchyma segmentation. Model optimization utilized a hybrid loss function that combined Dice Loss and Focal Loss. The model performance of all four 3D U-Nets was evaluated patient-wise using dice coefficient and Jaccard coefficient, then averaged to obtain the average volumetric dice coefficient (DSCavg) and average Jaccard coefficient (IoUavg) on a test dataset comprising 53 CT scans. Additionally, an ensemble approach (Model-V) was utilized featuring 3D U-Net (Model-I), ResNet (Model-II), and Inception (Model-III) 3D U-Net architectures, combined with two distinct patch sizes for further investigation. The ensemble of models obtained the highest DSCavg of 0.84 ± 0.05 and IoUavg of 0.74 ± 0.06 on the test dataset, compared against individual models. It mitigated false positives, overestimations, and underestimations observed in individual U-Net models. Moreover, the ensemble of models reduced average false positives per scan in the test dataset (1.57 nodules/scan) compared to individual models (2.69-3.39 nodules/scan). The suggested ensemble approach presents a strong and effective strategy for automatically detecting and delineating lung nodules, potentially aiding CAD systems in clinical settings. This approach could assist radiologists in laborious and meticulous lung nodule detection tasks in CT scans, improving lung cancer diagnosis and treatment planning.

Diagnosing Solid Lesions in the Pancreas With Multimodal Artificial Intelligence

Diagnosing solid lesions in the pancreas via endoscopic ultrasonographic (EUS) images is challenging. Artificial intelligence (AI) has the potential to help with such diagnosis, but existing AI models focus solely on a single modality. To advance the clinical diagnosis of solid lesions in the pancreas through developing a multimodal AI model integrating both clinical information and EUS images. In this randomized crossover trial conducted from January 1 to June 30, 2023, from 4 centers across China, 12 endoscopists of varying levels of expertise were randomly assigned to diagnose solid lesions in the pancreas with or without AI assistance. Endoscopic ultrasonographic images and clinical information of 439 patients from 1 institution who had solid lesions in the pancreas between January 1, 2014, and December 31, 2022, were collected to train and validate the joint-AI model, while 189 patients from 3 external institutions were used to evaluate the robustness and generalizability of the model. Conventional or AI-assisted diagnosis of solid lesions in the pancreas. In the retrospective dataset, the performance of the joint-AI model was evaluated internally and externally. In the prospective dataset, diagnostic performance of the endoscopists with or without the AI assistance was compared. The retrospective dataset included 628 patients (400 men [63.7%]; mean [SD] age, 57.7 [27.4] years) who underwent EUS procedures. A total of 130 patients (81 men [62.3%]; mean [SD] age, 58.4 [11.7] years) were prospectively recruited for the crossover trial. The area under the curve of the joint-AI model ranged from 0.996 (95% CI, 0.993-0.998) in the internal test dataset to 0.955 (95% CI, 0.940-0.968), 0.924 (95% CI, 0.888-0.955), and 0.976 (95% CI, 0.942-0.995) in the 3 external test datasets, respectively. The diagnostic accuracy of novice endoscopists was significantly enhanced with AI assistance (0.69 [95% CI, 0.61-0.76] vs 0.90 [95% CI, 0.83-0.94]; P < .001), and the supplementary interpretability information alleviated the skepticism of the experienced endoscopists. In this randomized crossover trial of diagnosing solid lesions in the pancreas with or without AI assistance, the joint-AI model demonstrated positive human-AI interaction, which suggested its potential to facilitate a clinical diagnosis. Nevertheless, future randomized clinical trials are warranted. ClinicalTrials.gov Identifier: NCT05476978.

Deriving Automated Device Metadata From Intracranial Pressure Waveforms: A Transforming Research and Clinical Knowledge in Traumatic Brain Injury ICU Physiology Cohort Analysis.

Treatment for intracranial pressure (ICP) has been increasingly informed by machine learning (ML)-derived ICP waveform characteristics. There are gaps, however, in understanding how ICP monitor type may bias waveform characteristics used for these predictive tools since differences between external ventricular drain (EVD) and intraparenchymal monitor (IPM)-derived waveforms have not been well accounted for. We sought to develop a proof-of-concept ML model differentiating ICP waveforms originating from an EVD or IPM. We examined raw ICP waveform data from the ICU physiology cohort within the prospective Transforming Research and Clinical Knowledge in Traumatic Brain Injury multicenter study. Nested patient-wise five-fold cross-validation and group analysis with bagged decision trees (BDT) and linear discriminant analysis were used for feature selection and fair evaluation. Nine patients were kept as unseen hold-outs for further evaluation. ICP waveform data totaling 14,110 hours were included from 82 patients (EVD, 47; IPM, 26; both, 9). Mean age, Glasgow Coma Scale (GCS) total, and GCS motor score upon admission, as well as the presence and amount of midline shift, were similar between groups. The model mean area under the receiver operating characteristic curve (AU-ROC) exceeded 0.874 across all folds. In additional rigorous cluster-based subgroup analysis, targeted at testing the resilience of models to cross-validation with smaller subsets constructed to develop models in one confounder set and test them in another subset, AU-ROC exceeded 0.811. In a similar analysis using propensity score-based rather than cluster-based subgroup analysis, the mean AU-ROC exceeded 0.827. Of 842 extracted ICP features, 62 were invariant within every analysis, representing the most accurate and robust differences between ICP monitor types. For the nine patient hold-outs, an AU-ROC of 0.826 was obtained using BDT. The developed proof-of-concept ML model identified differences in EVD- and IPM-derived ICP signals, which can provide missing contextual data for large-scale retrospective datasets, prevent bias in computational models that ingest ICP data indiscriminately, and control for confounding using our model's output as a propensity score by to adjust for the monitoring method that was clinically indicated. Furthermore, the invariant features may be leveraged as ICP features for anomaly detection.

Sequence comparison of spoiled gradient echo and balanced steady-state free precession for pulmonary free-breathing proton MRI in patients and healthy volunteers: Correspondence, repeatability, and validation with dynamic contrast-enhanced MRI.

Phase-resolved functional lung (PREFUL) MRI is a proton-based, contrast agent-free technique derived from the Fourier decomposition approach to measure regional ventilation and perfusion dynamics during free-breathing. Besides the necessity of extensive PREFUL postprocessing, the utilized MRI sequence must fulfill specific requirements. This study investigates the impact of sequence selection on PREFUL-MRI-derived functional parameters by comparing the standard spoiled gradient echo (SPGRE) sequence with a lung-optimized balanced steady-state free precession (bSSFP) sequence, thereby facilitating PREFULs clinical application in pulmonary disease assessment. This study comprised a prospective dataset of healthy volunteers and a retrospective dataset of patients with suspected chronic thromboembolic pulmonary hypertension. Both cohorts underwent PREFUL-MRI with both sequences to assess the correspondence of PREFUL ventilation and perfusion parameters (A). Additionally, healthy subjects were scanned a second time to evaluate repeatability (B), whereas patients received dynamic contrast-enhanced (DCE)-MRI, considered the perfusion gold standard for comparison with PREFUL-MRI (C). Signal-to-noise ratio (SNR), calculated from the unprocessed images, was compared alongside median differences of PREFUL-MRI-derived parameters using a paired Wilcoxon signed rank test. Further evaluations included calculation of the Pearson correlation, intraclass-correlation coefficient for repeatability assessment, and spatial overlap (SO) for regional comparison of PREFUL-MRI and DCE-MRI. bSSFP showed a clear SNR advantage over SPGRE (median: 23 vs. 9, p< 0.001). (A) Despite significant differences, parameter values were strongly correlated (r≥0.75). After thresholding, binary maps showed high healthy overlap across both cohorts (SOHealthy > 86%) and high defect overlap in the patient cohort (SODefect≥48%). (B) bSSFP demonstrated slightly higher repeatability across most parameters. (C) Both sequences demonstrated comparable correspondence to DCE-MRI, with SPGRE excelling in absolute quantification and bSSFP in spatial agreement. Although bSSFP showed superior SNR results, both sequences displayed spatial defect concordance and highly correlated PREFUL parameters with deviations regarding repeatability and alignment with DCE-MRI.

Deep survival analysis for interpretable time-varying prediction of preeclampsia risk

ObjectiveSurvival analysis is widely utilized in healthcare to predict the timing of disease onset. Traditional methods of survival analysis are usually based on Cox Proportional Hazards model and assume proportional risk for all subjects. However, this assumption is rarely true for most diseases, as the underlying factors have complex, non-linear, and time-varying relationships. This concern is especially relevant for pregnancy, where the risk for pregnancy-related complications, such as preeclampsia, varies across gestation. Recently, deep learning survival models have shown promise in addressing the limitations of classical models, as the novel models allow for non-proportional risk handling, capturing nonlinear relationships, and navigating complex temporal dynamics. MethodsWe present a methodology to model the temporal risk of preeclampsia during pregnancy and investigate the associated clinical risk factors. We utilized a retrospective dataset including 66,425 pregnant individuals who delivered in two tertiary care centers from 2015 to 2023. We modeled the preeclampsia risk by modifying DeepHit, a deep survival model, which leverages neural network architecture to capture time-varying relationships between covariates in pregnancy. We applied time series k-means clustering to DeepHit’s normalized output and investigated interpretability using Shapley values. ResultsWe demonstrate that DeepHit can effectively handle high-dimensional data and evolving risk hazards over time with performance similar to the Cox Proportional Hazards model, achieving an area under the curve (AUC) of 0.78 for both models. The deep survival model outperformed traditional methodology by identifying time-varied risk trajectories for preeclampsia, providing insights for early and individualized intervention. K-means clustering resulted in patients delineating into low-risk, early-onset, and late-onset preeclampsia groups—notably, each of those has distinct risk factors. ConclusionThis work demonstrates a novel application of deep survival analysis in time-varying prediction of preeclampsia risk. Our results highlight the advantage of deep survival models compared to Cox Proportional Hazards models in providing personalized risk trajectory and demonstrating the potential of deep survival models to generate interpretable and meaningful clinical applications in medicine.

MR radiomics to predict microvascular invasion status and biological process in combined hepatocellular carcinoma-cholangiocarcinoma

ObjectivesTo establish an MRI-based radiomics model for predicting the microvascular invasion (MVI) status of cHCC-CCA and to investigate biological processes underlying the radiomics model.MethodsThe study consisted of a retrospective dataset (82 in the training set, 36 in the validation set) and a prospective dataset (25 patients in the test set) from two hospitals. Based on the training set, logistic regression analyses were employed to develop the clinical-imaging model, while radiomic features were extracted to construct a radiomics model. The diagnosis performance was further validated in the validation and test sets. Prognostic aspects of the radiomics model were investigated using the Kaplan–Meier method and log-rank test. Differential gene expression analysis and gene ontology (GO) analysis were conducted to explore biological processes underlying the radiomics model based on RNA sequencing data.ResultsOne hundred forty-three patients (mean age, 56.4 ± 10.5; 114 men) were enrolled, in which 73 (51.0%) were confirmed as MVI-positive. The radiomics model exhibited good performance in predicting MVI status, with the area under the curve of 0.935, 0.873, and 0.779 in training, validation, and test sets, respectively. Overall survival (OS) was significantly different between the predicted MVI-negative and MVI-positive groups (median OS: 25 vs 18 months, p = 0.008). Radiogenomic analysis revealed associations between the radiomics model and biological processes involved in regulating the immune response.ConclusionA robust MRI-based radiomics model was established for predicting MVI status in cHCC-CCA, in which potential prognostic value and underlying biological processes that regulate immune response were demonstrated.Critical relevance statementMVI is a significant manifestation of tumor invasiveness, and the MR-based radiomics model established in our study will facilitate risk stratification. Furthermore, underlying biological processes demonstrated in the radiomics model will offer valuable insights for guiding immunotherapy strategies.Key PointsMVI is of prognostic significance in cHCC-CCA, but lacks reliable preoperative assessment.The MRI-based radiomics model predicts MVI status effectively in cHCC-CCA.The MRI-based radiomics model demonstrated prognostic value and underlying biological processes.The radiomics model could guide immunotherapy and risk stratification in cHCC-CCA.Graphical

P-252 Non-invasive artificial intelligence (AI) prediction of blastocyst development from mature oocytes unveils correlation with blastocyst ploidy for enhanced reproductive insight at earliest stage of development

Abstract Study question Can MAGENTA identify mature oocytes more likely to develop into a blastocyst of euploid status? Summary answer Although built to predict blastocyst development, MAGENTA analysis of mature oocytes additionally displays the ability to highlight those with higher potential to become euploid embryos. What is known already MAGENTA (AI image-analysis tool) has been trained to assess and predict blastocyst stage embryo development from mature, denuded oocytes. MAGENTA scores consistently identify oocytes of better quality – more likely to develop into a blastocyst, including blastocysts of higher morphological quality. However, in addition to achieving a blastocyst, the ploidy status of that embryo is crucial in dictating overall chances of implantation and cycle success. Although MAGENTA was not trained to predict blastocyst ploidy from oocytes, most chromosomal errors are of maternal meiotic origin. A correlation between MAGENTA and ploidy status would ignite research possibilities and provide valuable insights. Study design, size, duration A retrospective dataset of 1,324 blastocyst embryos that underwent PGT-A via NGS sequencing, between the years 2021-2023, was assessed. All embryos that were tested for ploidy from a patient cohort were included. Embryos reported as mosaic or ‘no diagnosis’ were excluded from analysis (N = 169). All embryos were cultured in a time-lapse incubator immediately post-ICSI. Of the biopsied blastocysts, 81% were of good morphological quality (expansion grade 1-6, ICM and TE grade A/B). Participants/materials, setting, methods The data includes 292 patients with a mean age of 38.5 (range 19-46) and an average number of tested blastocysts of 4 (range 1-14). Of the 1173 blastocysts, 847 were aneuploid (72.2%) and 326 were euploid (27.8%). 83% of the blastocysts were from patients of advanced maternal age (AMA, ≥38years old). The first time-lapse image, capturing the mature oocyte stage, was extracted, and scored by MAGENTA (scale 0-10). MAGENTA scores were compared by Welch’s t-tests. Main results and the role of chance Impressively, oocytes that developed into euploid blastocysts (n = 326) were identified by MAGENTA with a higher average score of 6.8 compared to those that developed into aneuploid blastocysts (n = 847) with a score of 6.4; a statistically significant differentiation (p &amp;lt; 0.05). 80% of the dataset were high quality blastocysts (expansion 4-6, ICM and TE grade A/B) with an imbalance of euploid (n = 295; mean score 6.8) and aneuploid (n = 643; mean score 6.6) blastocysts—likely a result of AMA. Therefore, subgroup analysis was focused on assessing the ploidy status relationships amongst several patient age groups within these AMA cases. Due to our clinic’s indication for PGT-A, the smallest subset of oocytes was from patients &amp;lt;38 years of age (N = 373, 48% euploid). In this age group there was no significant difference in MAGENTA scores between the oocytes of euploid and aneuploid embryo status (6.9 vs 6.7; p = 0.522). However, in oocytes from patients aged ≥38 (N = 800, 19% euploid), the distinction between those that became euploid and aneuploid blastocysts is evident with a statistically significant difference in MAGENTA scores (6.8 vs 6.3; p &amp;lt; 0.05). This relationship with ploidy status additionally remains with further stratification into the ≥41 age group (N = 467 oocytes, 13% euploid; 6.7 vs 5.9; p &amp;lt; 0.01). Limitations, reasons for caution Majority of the patients in this dataset underwent PGT-A due to AMA, further research is required to include a wider age demographic, particularly in patients &amp;lt;35 years old. Greater representation of varying blastocyst qualities is required to further assess MAGENTA correlation to ploidy outcomes amongst quality groups. Wider implications of the findings This study presents the first non-invasive assessment of mature oocyte quality that indicates oocytes with greater potential for euploid blastocyst development. This application of MAGENTA provides a novel method of gaining ploidy status potential at the mature oocyte stage, which is invaluable to both IVF and cryopreservation cycles. Trial registration number NOT APPLICABLE

Association of Dynamic Trajectories of Time-Series Data and Life-Threatening Mass Effect in Large Middle Cerebral Artery Stroke.

Life-threatening, space-occupying mass effect due to cerebral edema and/or hemorrhagic transformation is an early complication of patients with middle cerebral artery stroke. Little is known about longitudinal trajectories of laboratory and vital signs leading up to radiographic and clinical deterioration related to this mass effect. We curated a retrospective data set of 635 patients with large middle cerebral artery stroke totaling 95,463 data points for 10 longitudinal covariates and 40 time-independent covariates. We assessed trajectories of the 10 longitudinal variables during the 72h preceding three outcomes representative of life-threatening mass effect: midline shift ≥ 5mm, pineal gland shift (PGS) > 4mm, and decompressive hemicraniectomy (DHC). We used a "backward-looking" trajectory approach. Patients were aligned based on outcome occurrence time and the trajectory of each variable was assessed before that outcome by accounting for cases and noncases, adjusting for confounders. We evaluated longitudinal trajectories with Cox proportional time-dependent regression. Of 635 patients, 49.0% were female, and the mean age was 69years. Thirty five percent of patients had midline shift ≥ 5mm, 24.3% of patients had PGS > 4mm, and 10.7% of patients underwent DHC. Backward-looking trajectories showed mild increases in white blood cell count (10-11K/UL within 72h), temperature (up to half a degree within 24h), and sodium levels (1-3mEq/L within 24h) before the three outcomes of interest. We also observed a decrease in heart rate (75-65 beats per minute) 24h before DHC. We found a significant association between increased white blood cell count with PGS > 4mm (hazard ratio 1.05, p value 0.007). Longitudinal profiling adjusted for confounders demonstrated that white blood cell count, temperature, and sodium levels appear to increase before radiographic and clinical indicators of space-occupying mass effect. These findings will inform the development of multivariable dynamic risk models to aid prediction of life-threatening, space-occupying mass effect.

A Clinical-Radiomics Nomogram for Predicting Early Ischemic Stroke Risk in Patients with Transient Ischemic Attack

To develop and validate a clinical-radiomics nomogram for predicting early ischemic stroke risk in patients who sustain a transient ischemic attack (TIA). A retrospective training dataset (n= 76) and a prospective validation dataset (n= 34) of patients with TIA were studied. Image processing was performed using ITK-snap and Artificial Intelligent Kit. Radiomics features were selected in R. A nomogram predicting recurrent TIA/stroke in 90 days as a recurrent ischemic event was established. Model performance was assessed by computing the receiver operating characteristic curve and decision curve analysis (DCA). We found a higher proportion of diabetes and hypertension in the patients with recurrent TIA compared with the stable patients in both the training and validation datasets (P < 0.05). Recurrent patients had significantly higher ABCD2 scores and plaque scores compared to stable patients. ABCD2 score and necrotic/lipid core area were independent risk factors for recurrent ischemic events (odds ratio [OR], 2.75; 95% confidence interval [CI], 1.47-6.40; and OR, 1.20; 95% CI, 1.07-1.41, respectively). The radiomics model had area under the curve values of 0.737 (95% CI, 0.715-0.878) in the training dataset and 0.899 (95% CI, 0.706-0.936) in the validation dataset, which was superior to the ABCD2 score and plaque model for predicting stroke recurrence (P < 0.05). The nomogram predicting recurrent ischemic events was 0.923 (95% CI, 0.895-0.978) in the training dataset and 0.935 (95% CI, 0.830-0.959) in the validation dataset. DCA confirmed the clinical value of this nomogram. The nomogram, based on clinical ABCD2 score, carotid plaque components and radiomics score, shows good performance in predicting the risk of recurrent ischemic events in patients with TIA.

Retinal peri-arteriolar versus peri-venular amyloidosis, hippocampal atrophy, and cognitive impairment: exploratory trial

The relationship between amyloidosis and vasculature in cognitive impairment and Alzheimer’s disease (AD) pathogenesis is increasingly acknowledged. We conducted a quantitative and topographic assessment of retinal perivascular amyloid plaque (AP) distribution in individuals with both normal and impaired cognition. Using a retrospective dataset of scanning laser ophthalmoscopy fluorescence images from twenty-eight subjects with varying cognitive states, we developed a novel image processing method to examine retinal peri-arteriolar and peri-venular curcumin-positive AP burden. We further correlated retinal perivascular amyloidosis with neuroimaging measures and neurocognitive scores. Our study unveiled that peri-arteriolar AP counts surpassed peri-venular counts throughout the entire cohort (P < 0.0001), irrespective of the primary, secondary, or tertiary vascular branch location, with a notable increase among cognitively impaired individuals. Moreover, secondary branch peri-venular AP count was elevated in the cognitively impaired (P < 0.01). Significantly, peri-venular AP count, particularly in secondary and tertiary venules, exhibited a strong correlation with clinical dementia rating, Montreal cognitive assessment score, hippocampal volume, and white matter hyperintensity count. In conclusion, our exploratory analysis detected greater peri-arteriolar versus peri-venular amyloidosis and a marked elevation of amyloid deposition in secondary branch peri-venular regions among cognitively impaired subjects. These findings underscore the potential feasibility of retinal perivascular amyloid imaging in predicting cognitive decline and AD progression. Larger longitudinal studies encompassing diverse populations and AD-biomarker confirmation are warranted to delineate the temporal-spatial dynamics of retinal perivascular amyloid deposition in cognitive impairment and the AD continuum.