Early Prediction Of Outcome Research Articles

Early Prediction of Poststroke Rehabilitation Outcomes Using Wearable Sensors.

Inpatient rehabilitation represents a critical setting for stroke treatment, providing intensive, targeted therapy and task-specific practice to minimize a patient's functional deficits and facilitate their reintegration into the community. However, impairment and recovery vary greatly after stroke, making it difficult to predict a patient's future outcomes or response to treatment. In this study, the authors examined the value of early-stage wearable sensor data to predict 3 functional outcomes (ambulation, independence, and risk of falling) at rehabilitation discharge. Fifty-five individuals undergoing inpatient stroke rehabilitation participated in this study. Supervised machine learning classifiers were retrospectively trained to predict discharge outcomes using data collected at hospital admission, including patient information, functional assessment scores, and inertial sensor data from the lower limbs during gait and/or balance tasks. Model performance was compared across different data combinations and was benchmarked against a traditional model trained without sensor data. For patients who were ambulatory at admission, sensor data improved the predictions of ambulation and risk of falling (with weighted F1 scores increasing by 19.6% and 23.4%, respectively) and maintained similar performance for predictions of independence, compared to a benchmark model without sensor data. The best-performing sensor-based models predicted discharge ambulation (community vs household), independence (high vs low), and risk of falling (normal vs high) with accuracies of 84.4%, 68.8%, and 65.9%, respectively. Most misclassifications occurred with admission or discharge scores near the classification boundary. For patients who were nonambulatory at admission, sensor data recorded during simple balance tasks did not offer predictive value over the benchmark models. These findings support the continued investigation of wearable sensors as an accessible, easy-to-use tool to predict the functional recovery after stroke. Accurate, early prediction of poststroke rehabilitation outcomes from wearable sensors would improve our ability to deliver personalized, effective care and discharge planning in the inpatient setting and beyond.

The Efficacy of LRINEC Scoring in Patients with Necrotizing Fasciitis and its Correlation with the Outcomes.

<b><br>Introduction:</b> Necrotizing fasciitis (NF) is an expeditiously escalating inflammatory infection of the fascia, with secondary necrosis of the subcutaneous tissues. To aid diagnosis, the Laboratory Risk Indicator for Necrotizing Fasciitis (LRINEC) scoring system was devised to stratify the likelihood of infection in patients on presentation.</br> <b><br>Aim:</b> The study aimed to investigate and quantify the relationships between the LRINEC scoring system and the outcomes such as the need for amputation, length of hospital stay, and mortality in patients with necrotizing fasciitis.</br> <b><br>Material and methods:</b> A total number of 90 patients were admitted with a provisional diagnosis of necrotizing fasciitis. LRINEC score was calculated for each case based on six laboratory variables at the time of presentation. Enrolled patients were divided into three groups on the basis of the LRINEC score. The differences in mortality, length of hospitalization, number of debridement procedures, and need for amputation between these groups were compared.</br> <b><br>Results:</b> Increasing age, male gender, and DM have a poor prognosis in necrotizing fasciitis patients. The amputation and mortality rates are better correlated with higher LRINEC scores.</br> <b><br>Conclusions:</b> The LRINEC score helps in stratifying the patients into three risk categories (low risk, moderate risk, and high risk) according to the severity in a much more organized way, and thus the appropriate management like surgical debridement can be introduced on time. LRINEC score is a robust index that is capable of detecting early cases of necrotizing fasciitis and is simple enough for routine use. It is a simplified bedside diagnostic tool for early diagnosis and prediction of outcomes in patients with necrotizing fasciitis.</br&gt.

Clinical Features and a Prediction Model for Early Prediction of Composite Outcome in Chlamydia psittaci Pneumonia: A Multi-Centre Retrospective Study in China.

C. psittaci pneumonia has atypical clinical manifestations and is often ignored by clinicians. This study analyzed the clinical characteristics, explored the risk factors for composite outcome and established a prediction model for early prediction of composite outcome among C. psittaci pneumonia patients. A multicenter, retrospective, observational cohort study was conducted in ten Chinese tertiary hospitals. Patients diagnosed with C. psittaci pneumonia were included, and their clinical data were collected and analyzed. The composite outcome of C. psittaci pneumonia included death during hospitalization, ICU admission, and mechanical ventilation. Univariate and multivariable logistic regression analyses were conducted to determine the significant variables. A ten-fold cross-validation was performed to internally validate the model. The model performance was evaluated using various methods, including receiver operating characteristics (ROC), C-index, sensitivity, specificity, positive/negative predictive value (PPV/NPV), decision curve analysis (DCA), and clinical impact curve analysis (CICA). In total, 83 patients comprised training cohorts and 36 patients comprised validation cohorts. CURB-65 was used to establish predictive Model 1. Multivariate logistic regression analysis identified three independent prognostic factors, including serum albumin, CURB-65, and white blood cells. These factors were employed to construct model 2. Model 2 had acceptable discrimination (AUC of 0.898 and 0.825 for the training and validation sets, respectively) and robust internal validity. The specificity, sensitivity, NPV, and PPV for predicting composite outcome in the nomogram model were 91.7%, 84.5%, 50.0%, and 98.4% in the training sets, and 100.0%, 64.7%, 14.2%, and 100.0% in the validation sets. DCA and CICA showed that the nomogram model was clinically practical. This study constructs a refined nomogram model for predicting the composite outcome in C. psittaci pneumonia patients. This nomogram model enables early and accurate C. psittaci pneumonia patients' evaluation, which may improve clinical outcomes.

Radiomics analysis of baseline computed tomography to predict oncological outcomes in patients treated for resectable colorectal cancer liver metastasis.

The purpose of this study was to determine and compare the performance of pre-treatment clinical risk score (CRS), radiomics models based on computed (CT), and their combination for predicting time to recurrence (TTR) and disease-specific survival (DSS) in patients with colorectal cancer liver metastases. We retrospectively analyzed a prospectively maintained registry of 241 patients treated with systemic chemotherapy and surgery for colorectal cancer liver metastases. Radiomics features were extracted from baseline, pre-treatment, contrast-enhanced CT images. Multiple aggregation strategies were investigated for cases with multiple metastases. Radiomics signatures were derived using feature selection methods. Random survival forests (RSF) and neural network survival models (DeepSurv) based on radiomics features, alone or combined with CRS, were developed to predict TTR and DSS. Leveraging survival models predictions, classification models were trained to predict TTR within 18 months and DSS within 3 years. Classification performance was assessed with area under the receiver operating characteristic curve (AUC) on the test set. For TTR prediction, the concordance index (95% confidence interval) was 0.57 (0.57-0.57) for CRS, 0.61 (0.60-0.61) for RSF in combination with CRS, and 0.70 (0.68-0.73) for DeepSurv in combination with CRS. For DSS prediction, the concordance index was 0.59 (0.59-0.59) for CRS, 0.57 (0.56-0.57) for RSF in combination with CRS, and 0.60 (0.58-0.61) for DeepSurv in combination with CRS. For TTR classification, the AUC was 0.33 (0.33-0.33) for CRS, 0.77 (0.75-0.78) for radiomics signature alone, and 0.58 (0.57-0.59) for DeepSurv score alone. For DSS classification, the AUC was 0.61 (0.61-0.61) for CRS, 0.57 (0.56-0.57) for radiomics signature, and 0.75 (0.74-0.76) for DeepSurv score alone. Radiomics-based survival models outperformed CRS for TTR prediction. More accurate, noninvasive, and early prediction of patient outcome may help reduce exposure to ineffective yet toxic chemotherapy or high-risk major hepatectomies.

Enhanced feature selection and ensemble learning for cardiovascular disease prediction: hybrid GOL2-2 T and adaptive boosted decision fusion with babysitting refinement.

Global Cardiovascular disease (CVD) is still one of the leading causes of death and requires the enhancement of diagnostic methods for the effective detection of early signs and prediction of the disease outcomes. The current diagnostic tools are cumbersome and imprecise especially with complex diseases, thus emphasizing the incorporation of new machine learning applications in differential diagnosis. This paper presents a new machine learning approach that uses MICE for mitigating missing data, the IQR for handling outliers and SMOTE to address first imbalance distance. Additionally, to select optimal features, we introduce the Hybrid 2-Tier Grasshopper Optimization with L2 regularization methodology which we call GOL2-2T. One of the promising methods to improve the predictive modelling is an Adaboost decision fusion (ABDF) ensemble learning algorithm with babysitting technique implemented for the hyperparameters tuning. The accuracy, recall, and AUC score will be considered as the measures for assessing the model. On the results, our heart disease prediction model yielded an accuracy of 83.0%, and a balanced F1 score of 84.0%. The integration of SMOTE, IQR outlier detection, MICE, and GOL2-2T feature selection enhances robustness while improving the predictive performance. ABDF removed the impurities in the model and elaborated its effectiveness, which proved to be high on predicting the heart disease. These findings demonstrate the effectiveness of additional machine learning methodologies in medical diagnostics, including early recognition improvements and trustworthy tools for clinicians. But yes, the model's use and extent of work depends on the dataset used for it really. Further work is needed to replicate the model across different datasets and samples: as for most models, it will be important to see if the results are generalizable to populations that are not representative of the patient population that was used for the current study.

Advances in Computational Biology for Diagnosing Neurodegenerative Diseases: A Comprehensive Review.

The numerous and varied forms of neurodegenerative illnesses provide a considerable challenge to contemporary healthcare. The emergence of artificial intelligence has fundamentally changed the diagnostic picture by providing effective and early means of identifying these crippling illnesses. As a subset of computational intelligence, machine-learning algorithms have become very effective tools for the analysis of large datasets that include genetic, imaging, and clinical data. Moreover, multi-modal data integration, which includes information from brain imaging (MRI, PET scans), genetic profiles, and clinical evaluations, is made easier by computational intelligence. A thorough knowledge of the course of the illness is made possible by this consolidative method, which also facilitates the creation of predictive models for early medical evaluation and outcome prediction. Furthermore, there has been a great deal of promise shown by the use of artificial intelligence to neuroimaging analysis. Sophisticated image processing methods combined with machine learning algorithms make it possible to identify functional and structural anomalies in the brain, which often act as early indicators of neurodegenerative diseases. This chapter examines how computational intelligence plays a critical role in improving the diagnosis of neurodegenerative diseases such as Parkinson's, Alzheimer's, etc. To sum up, computational intelligence provides a revolutionary approach for improving the identification of neurodegenerative illnesses. In the battle against these difficult disorders, embracing and improving these computational techniques will surely pave the path for more individualized therapy and more therapies that are successful.

Dual-purposing disulfiram for enhanced chemotherapy and afterglow imaging using chlorin e6 and semiconducting polymer combined strategy.

Rationale: One of the main challenges in chemotherapy is achieving high treatment efficacy while minimizing adverse events. Fully exploiting the therapeutic potential of an old drug and monitoring its effects in vivo is highly valuable, but often difficult to achieve. Methods: In this study, by encapsulating disulfiram (DSF) approved by US Food and Drug Administration, semiconducting polymer nanocomplex (MEHPPV), and Chlorin e6 into a polymeric matrix F127 via nanoprecipitation method, a nanosystem (FCMC) was developed for afterglow imaging guided cancer treatment. The characteristics, stability as well as the ability of singlet oxygen (1O2) production of FCMC were first carefully examined. Then, we studied the mechanism for enhanced anti-cancer efficiency and afterglow luminescence in vitro. For experiments in vivo, 4T1 subcutaneous xenograft tumor mice were injected with FCMC via the tail vein every three days and the antitumor effect of FCMC was evaluated by monitoring tumor volume and body weight every three day. Results: The nanosystem, which combines DSF with Ce6, can generates abundant 1O2 that enhances the antitumor activity of DSF. The in vivo results show that FCMC-treated group exhibits an obviously higher tumor-growth inhibition rate of 67.89% after 15 days of treatment, compared to the control group of F127@MEHPPV-CuET. Moreover, Ce6 also greatly enhances the afterglow luminescence intensity of MEHPPV and promotes the redshift of the afterglow emission towards the ideal near-infrared imaging window, thereby enabling efficient afterglow tumor imaging in vivo. Conclusions: This multifunctional nanoplatform not only improves the anticancer efficacy of DSF, but also enables monitoring tumor via robust afterglow imaging, thus exhibiting great potential for cancer therapy and early therapeutic outcome prediction.

Neutrophil-to-lymphocyte ratio at admission for early diagnosis, severity assessment, and prognosis of acute traumatic spinal cord injury.

A retrospective study. This study examined the value of neutrophil-to-lymphocyte ratio at admission for early diagnosis, severity assessment, and prognosis of acute traumatic SCI. The First People's Hospital of Neijiang, China. This was a single-center, retrospective, cohort study of patients treated within 12 h of acute SCI between January 2018 and October 2022. Ninety-four SCI patients were selected as the Observation group, including 26 with complete injury (AIS grade A) and 68 with incomplete injury (AIS grade B-D), while 94 patients with simple spinal fracture were randomly selected as the Control group. Eighty-one observation group patients underwent surgical treatment, of which 33 had a higher AIS grade (Good prognosis subgroup) and 48 a lower or equal grade post-surgery (Poor prognosis subgroup). Univariate and multivariate analyses were performed to assess predictors of early diagnosis, severity, and 6-month outcome. Initial white blood cell count, neutrophil count, monocyte count, and NLR were higher in the Observation group than the Control group, while lymphocyte count was lower in the Observation group. Multivariate logistic regression analysis identified NLR as an independent predictor of early diagnosis. Spinal canal encroachment ≥50%, neutrophil count, and NLR were higher in the complete injury subgroup, and spinal canal encroachment ≥50% was an independent predictor of complete injury, while NLR was not. The NLR was higher in the poor prognosis subgroup and was an independent risk factor. Peripheral blood NLR is useful for early diagnosis of acute SCI and is predictive of clinical outcome.

Novel alternate (Alt) response criteria for early outcome prediction in pancreatic (P) neuroendocrine tumors (NETs): utilizing banked imaging and radiomics

Novel alternate (Alt) response criteria for early outcome prediction in pancreatic (P) neuroendocrine tumors (NETs): utilizing banked imaging and radiomics

The prognostic value of the phosphate-to-albumin ratio in patients with OHCA: A multicenter observational study

PurposeIn patients with out-of-hospital cardiac arrest (OHCA), early and accurate outcome prediction is crucial for making treatment decisions and informing their relatives. A previous study reported an association between high phosphate levels and unfavorable neurological outcomes after return of spontaneous circulation (ROSC); however, its prognostic value was insufficient when used independently. Therefore, this study aimed to validate the usefulness of the phosphate-to-albumin ratio (PAR) in predicting neurological outcomes and in-hospital mortality by incorporating albumin, another known prognostic indicator. Materials and methodsThis multicenter observational study included adult OHCA survivors from October 2015 to June 2021. The primary endpoint was an unfavorable neurological outcome at hospital discharge, defined as a cerebral performance category score of 3–5. The in-hospital mortality rates were also evaluated. ResultsOf the 2397 adult OHCA survivors, PAR differed significantly between the unfavorable and favorable neurological outcome groups, as well as between the non-survival and survival to hospital discharge groups (2.4 vs 1.4, 2.5 vs 1.6, respectively). The area under the receiver operating characteristic curve (AUROC) value of the PAR for predicting unfavorable neurological outcome was 0.81 (95% confidence interval [CI], 0.79–0.83), and the AUROC value for predicting in-hospital mortality was 0.76 (95% CI, 0.74–0.78). In multivariable analysis, the PAR was independently associated with unfavorable neurological outcome (odds ratio [OR] 1.30, 95% CI 1.15–1.37; p < 0.001) and in-hospital mortality (OR 1.24, 95% CI 1.12–1.38; p < 0.001). ConclusionThe PAR is a readily obtainable and independent prognostic indicator for patients with ROSC after OHCA, helping healthcare providers in predicting outcomes.

Clinical Outcome Prediction of Early Brain Injury in Aneurysmal Subarachnoid Hemorrhage: the SHELTER-Score

BackgroundDespite intensive research on preventing and treating vasospasm and delayed cerebral ischemia in aneurysmal subarachnoid hemorrhage (aSAH), mortality and morbidity rates remain high. Early brain injury (EBI) has emerged as possibly the major significant factor in aSAH pathophysiology, emphasizing the need to investigate EBI-associated clinical events for improved patient management and decision-making. This study aimed to identify early clinical and radiological events within 72 h after aSAH to develop a conclusive predictive EBI score for clinical practice.MethodsThis retrospective analysis included 561 consecutive patients with aSAH admitted to our neurovascular center between 01/2014 and 09/2022. Fourteen potential predictors occurring within the initial 72 h after hemorrhage were analyzed. The modified Rankin Scale (mRS) score at 6 months, discretized to three levels (0–2, favorable; 3–5, poor; 6, dead), was used as the outcome variable. Univariate ordinal regression ranked predictors by significance, and forward selection with McFadden’s pseudo-R2 determined the optimal set of predictors for multivariate proportional odds logistic regression. Collinear parameters were excluded, and fivefold cross-validation was used to avoid overfitting.ResultsThe analysis resulted in the Subarachnoid Hemorrhage Associated Early Brain Injury Outcome Prediction score (SHELTER-score), comprising seven clinical and radiological events: age (0–4 points), World Federation of Neurosurgical Societies (0–2.5 points), cardiopulmonary resuscitation (CPR) (2 points), mydriasis (1–2 points), midline shift (0.5–1 points), early deterioration (1 point), and early ischemic lesion (2 points). McFadden’s pseudo-R2 = 0.339, area under the curve for death or disability 0.899 and 0.877 for death. A SHELTER-score below 5 indicated a favorable outcome (mRS 0–2), 5–6.5 predicted a poor outcome (mRS 3–5), and ≥ 7 correlated with death (mRS 6) at 6 months.ConclusionsThe novel SHELTER-score, incorporating seven clinical and radiological features of EBI, demonstrated strong predictive performance in determining clinical outcomes. This scoring system serves as a valuable tool for neurointensivists to identify patients with poor outcomes and guide treatment decisions, reflecting the great impact of EBI on the overall outcome of patients with aSAH.

Early Prediction of Treatment Response By Circulating Tumor DNA Profiling in Patients with Diffuse Large B-Cell Lymphoma Receiving CAR T-Cell Therapy

Introduction Patients with relapsed or refractory diffuse large B-cell lymphoma (r/r DLBCL) are characterized by a particularly poor prognosis. Anti-CD19 chimeric antigen receptor (CAR) T-cell therapy has improved outcomes of patients with r/r DLBCL, yet a substantial proportion of patients still experiences relapse or progression after CAR T-cell treatment. Identifying patients at high risk of CAR T-cell therapy resistance or future lymphoma progression remains challenging. Here, we explored the value of circulating tumor DNA (ctDNA) as a prognostic biomarker at baseline and early into treatment in DLBCL patients undergoing standard-of-care CAR T-cell therapy, without the need for matched tumor genotypes. We further investigated associations between ctDNA and other known clinical risk factors. Methods We applied targeted next-generation sequencing (NGS) to a total of 53 samples from r/r DLBCL patients receiving CAR T-cell therapy (axicabtagene ciloleucel and tisagenlecleucel, n=16) at the University Medical Center Freiburg (Germany), using a modified version of the AVENIO ctDNA analysis workflow (Roche; Research Use Only) that covered ~314kb and targeted 466 distinct genes recurrently mutated in DLBCL (based on the CAPP-Seq workflow, Kurtz DM et al. J Clin Oncol 2018). Tumor genotypes were assessed noninvasively from plasma samples obtained at baseline before lymphodepletion with matched germline controls. Concentrations of ctDNA were quantified at baseline and early after CAR T-cell therapy at day 7-10. Initial treatment response was assessed by conventional PET-CT obtained 4-6 weeks after treatment based on the Lugano criteria. Results At baseline prior to lymphodepletion, we detected somatic mutations in 100% of plasma samples by noninvasive genotyping, with a median of 19.5 mutations per patient (range: 3-97). The most frequently mutated genes were IGHV, IGLL5, BCL2, MYC, BTG1, CREBBP, and TP53. Patients with high baseline ctDNA concentrations (mean allele frequency [AF] &amp;gt;= 1%) showed shorter progression-free survivial (PFS) and overall survival (OS) than patients with ctDNA levels below this threshold (median PFS: 86.5 days vs. not reached, p=0.04, HR: 6.3, 95% CI: 1.7-23.6; median OS: 219 days vs. not reached, p=0.048, HR: 4.1, 95% CI 1.0-16.6, Figure 1). Pretreatment ctDNA levels were significantly correlated with LDH concentrations ( p=0.049, r=0.5). We further found significantly higher ctDNA concentrations in patients with transformed DLBCL (vs. non-transformed, p=0.008) and in those with 3 or more prior treatment lines (vs. 1-2, p=0.04). At day 7-10 after CAR T-cell therapy, ctDNA levels dropped a median of 10-fold compared to the pretreatment time point ( p=0.008). Patients achieving a complete (CMR) or partial metabolic response (PMR) by PET-CT after 4-6 weeks showed a significant decrease of ctDNA ( p=0.001), while those with a stable (SD) or progressive disease (PD) had no significant ctDNA reduction at day 7-10 after CAR T-cell infusion. Vice versa, while 100% of patients with a &amp;gt;1.5-log-fold drop of ctDNA level between the pretreatment and interim time point revealed a CMR or PMR in early PET-CT scans, only 62% of patients with a smaller ctDNA decline responded to treatment according to PET-CT. Conclusions Our data suggests, together with previously published studies (Sworder BJ et al., Cancer Cell, 2023; Frank MJ et al. J Clin Oncol, 2021), that ctDNA profiling may allow robust noninvasive genotyping, accurate risk-stratification, and early prediction of clinical outcomes in r/r DLBCL patients undergoing CAR T-cell therapy. Although limited by small sample size, this pilot study highlights a potential future role of ctDNA measurements for personalized treatment selection and guidance in clinical trials.

Circulating Tumor DNA Dynamics as Early Outcome Predictors for Lisocabtagene Maraleucel as Second-Line Therapy for Large B-Cell Lymphoma from the Phase 3 TRANSFORM Study

Background: While the prognostic value of MRD using ultrasensitive circulating tumor DNA (ctDNA) assays after first-line immunochemotherapy for DLBCL has been shown (Roschewski M, et al. Blood 2022;140[suppl 1]:785), its relevance to outcomes after second-line (2L) treatment with CD19-directed CAR T cell therapy remains unclear. The phase 3 TRANSFORM study (NCT03575351) showed statistically significant and clinically meaningful improvements in event-free survival (EFS), CR rate, and PFS for lisocabtagene maraleucel (liso-cel) compared with standard of care (SOC) for patients (pts) with primary refractory or early relapsed large B-cell lymphoma (LBCL) eligible for autologous HSCT (Kamdar M, et al. Lancet 2022; Abramson JS, et al. Blood 2023). Here, we report longitudinally measured ctDNA levels from TRANSFORM as a prospective measure of disease burden and describe the predictive value of ctDNA after 2L liso-cel therapy for LBCL. Methods: From the original randomized population (n = 184), 551 plasma samples from 160 pts (liso-cel, n = 79; SOC, n = 81) were used for ctDNA assessment in a blinded manner by Phased Variant (PV) Enrichment and Detection Sequencing (PhasED-Seq) at Foresight Diagnostics. Tumor-derived PVs were identified directly from baseline plasma samples taken at screening without use of tumor tissue; matched DNA from peripheral blood mononuclear cells was used to censor germline variants and clonal hematopoiesis. PVs were used to longitudinally assess ctDNA on the day of liso-cel infusion; at Day 15; and Months 1, 2, 3, and 12 after infusion. For the SOC arm, only pretreatment samples were evaluated. ctDNA levels were compared with baseline characteristics, responses by Lugano 2014 criteria, and EFS. Pts/samples were reported as having detectable MRD when ctDNA levels exceeded a detection threshold (≈1:10 6 cell-free DNA molecules) corresponding to 98% specificity. Results: PVs were identified from 136 of 160 pts (85%) with evaluable pretreatment samples from the liso-cel and SOC treatment arms. The remaining samples had lower disease burden (sum of product of diameters[SPD]; 13%) or failed quality check (2%). Pretreatment ctDNA levels varied widely (median [IQR]: 136 [35‒696] haploid genome equivalents/mL) and were correlated with disease indices portraying higher-risk disease, including stage ( P = 0.02), age-adjusted International Prognostic Index ( P &amp;lt; 0.0001), disease burden (SPD; P &amp;lt; 0.0001), and LDH ( P &amp;lt; 0.0001). Serial ctDNA levels were evaluated after liso-cel infusion in 63 pts. Achieving undetectable ctDNA was associated with achieving CR and longer EFS at all time points after liso-cel infusion. ctDNA levels decreased rapidly in pts who achieved CR at Month 3, with 50% (17/34) having undetectable ctDNA by Day 15 after liso-cel infusion. Conversely, 89% of pts (16/18) with PD at Month 3 still had detectable ctDNA at Day 15 with lower levels of ctDNA clearance. These results highlight significant enrichment of undetectable ctDNA observed as early as Day 15 after infusion between pts with Month 3 CR versus PD ( P = 0.006). ctDNA clearance in pts who achieved CR at Month 3 continued to deepen further over time, with undetectable ctDNA rates of 63%, 74%, and 86% at 1, 3, and 12 months after liso-cel infusion, respectively. Furthermore, most pts in CR by PET/CT had undetectable ctDNA at paired time points (71%), confirming that pts achieving CR also achieved durable molecular remission. In the cases with discordant CR and ctDNA results, detection of residual ctDNA was significantly prognostic for shorter EFS at all respective response assessments (fraction of CR cases with residual ctDNA at 1 month and Log-rank test P value for EFS: 47% [15/32], P = 0.011; 3 months: 26% [9/34], P = 0.025; 12 months: 13% [4/31], P = 0.003). Conclusions: Achieving undetectable ctDNA in the first 3 months and as early as 15 days after liso-cel infusion was significantly associated with durable clinical benefit. Conversely, detectable ctDNA captured relapse risk unmeasured by imaging. These results demonstrate the potential clinical value of ctDNA monitoring as a biomarker for disease surveillance and as an early predictor of clinical benefit of liso-cel treatment for LBCL, which could guide and accelerate future clinical trials.

Evaluation of novel biomarkers for early pregnancy outcome prediction†.

To assess performance and discriminatory capacity of commercially available enzyme-linked immunosorbent assays of biomarkers for predicting first trimester pregnancy outcome in a multi-center cohort. In a case-control study at three academic centers of women with pain and bleeding in early pregnancy, enzyme-linked immunosorbent assays of biomarkers were screened for assay performance. Performance was assessed via functional sensitivity, assay reportable range, recovery/linearity, and intra-assay precision (%Coefficient of Variation). Top candidates were analyzed for discriminatory capacity for viability and location among 210 women with tubal ectopic pregnancy, viable intrauterine pregnancy, or miscarriage. Assay discrimination was assessed by visual plots, area under the curve with 95% confidence intervals, and measures of central tendency with two-sample t-tests. Of 25 biomarkers evaluated, 22 demonstrated good or acceptable assay performance. Transgelin-2, oviductal glycoprotein, and integrin-linked kinase were rejected due to poor performance. The best biomarkers for discrimination of pregnancy location were pregnancy-specific beta-1-glycoprotein 9, pregnancy-specific beta-1-glycoprotein 1, insulin-like growth factor binding protein 1, kisspeptin (KISS1), pregnancy-specific beta-1-glycoprotein 3, and beta parvin (PARVB). The best biomarkers for discrimination of pregnancy viability were pregnancy-specific beta-1-glycoprotein 9, pregnancy-specific beta-1-glycoprotein 3, EH domain-containing protein 3, KISS1, WAP four-disulfide core domain protein 2 (HE4), quiescin sulfhydryl oxidase 2, and pregnancy-specific beta-1-glycoprotein 1. Performance of commercially available enzyme-linked immunosorbent assays was acceptable for a panel of novel biomarkers to predict early pregnancy outcome. Of these, six and seven candidates demonstrated good discriminatory capacity of pregnancy location and viability, respectively, when validated in a distinct external population. Four markers demonstrated good discrimination for both location and viability.

Improving the efficacy of enuresis alarm treatment through early prediction of treatment outcome: a machine learning approach

Improving the efficacy of enuresis alarm treatment through early prediction of treatment outcome: a machine learning approach

Node embedding-based graph autoencoder outlier detection for adverse pregnancy outcomes

Adverse pregnancy outcomes, such as low birth weight (LBW) and preterm birth (PTB), can have serious consequences for both the mother and infant. Early prediction of such outcomes is important for their prevention. Previous studies using traditional machine learning (ML) models for predicting PTB and LBW have encountered two important limitations: extreme class imbalance in medical datasets and the inability to account for complex relational structures between entities. To address these limitations, we propose a node embedding-based graph outlier detection algorithm to predict adverse pregnancy outcomes. We developed a knowledge graph using a well-curated representative dataset of the Emirati population and two node embedding algorithms. The graph autoencoder (GAE) was trained by applying a combination of original risk factors and node embedding features. Samples that were difficult to reconstruct at the output of GAE were identified as outliers considered representing PTB and LBW samples. Our experiments using LBW, PTB, and very PTB datasets demonstrated that incorporating node embedding considerably improved performance, achieving a 12% higher AUC-ROC compared to traditional GAE. Our study demonstrates the effectiveness of node embedding and graph outlier detection in improving the prediction performance of adverse pregnancy outcomes in well-curated population datasets.

The Validity of the ROX Index and APACHE II in Predicting Early, Late, and Non-Responses to Non-Invasive Ventilation in Patients with COVID-19 in a Low-Resource Setting.

The use of the Ratio of Oxygen Saturation (ROX) index to predict the success of high-flow nasal oxygenation (HFNO) is well established. The ROX can also predict the need for intubation, mortality, and is easier to calculate compared with APACHE II. In this prospective study, the primary aim is to compare the ROX (easily administered in resource limited setting) to APACHE II for clinically relevant outcomes such as mortality and the need for intubation. Our secondary aim was to identify thresholds for the ROX index in predicting outcomes such as the length of ICU stay and failure of non-invasive respiratory support therapies and to assess the effectiveness of using the ROX (day 1 at admission, day 2, and day 3) versus Acute physiology and chronic health evaluation (APACHE) II scores (at admission) in patients with Coronavirus Disease 2019 (COVID-19) pneumonia and Acute Respiratory Distress Syndrome (ARDS) to predict early, late, and non-responders. After screening 208 intensive care unit patients, a total of 118 COVID-19 patients were enrolled, who were categorized into early (n = 38), late (n = 34), and non-responders (n = 46). Multinomial logistic regression, receiver operating characteristic (ROC), Multivariate Cox regression, and Kaplan-Meier analysis were conducted. Multinomial logistic regressions between late and early responders and between non- and early responders were associated with reduced risk of treatment failures. ROC analysis for early vs. late responders showed that APACHE II on admission had the largest area under the curve (0.847), followed by the ROX index on admission (0.843). For responders vs. non-responders, we found that the ROX index on admission had a slightly better AUC than APACHE II on admission (0.759 vs. 0.751). A higher ROX index on admission [HR (95% CI): 0.29 (0.13-0.52)] and on day 2 [HR (95% CI): 0.55 (0.34-0.89)] were associated with a reduced risk of treatment failure. The ROX index can be used as an independent predictor of early response and mortality outcomes to HFNO and NIV in COVID-19 pneumonia, especially in low-resource settings, and is non-inferior to APACHE II.

Platelet aggregates detected using quantitative phase imaging associate with COVID-19 severity

BackgroundThe clinical spectrum of acute SARS-CoV-2 infection ranges from an asymptomatic to life-threatening disease. Considering the broad spectrum of severity, reliable biomarkers are required for early risk stratification and prediction of clinical outcomes. Despite numerous efforts, no COVID-19-specific biomarker has been established to guide further diagnostic or even therapeutic approaches, most likely due to insufficient validation, methodical complexity, or economic factors. COVID-19-associated coagulopathy is a hallmark of the disease and is mainly attributed to dysregulated immunothrombosis. This process describes an intricate interplay of platelets, innate immune cells, the coagulation cascade, and the vascular endothelium leading to both micro- and macrothrombotic complications. In this context, increased levels of immunothrombotic components, including platelet and platelet-leukocyte aggregates, have been described and linked to COVID-19 severity.MethodsHere, we describe a label-free quantitative phase imaging approach, allowing the identification of cell-aggregates and their components at single-cell resolution within 30 min, which prospectively qualifies the method as point-of-care (POC) testing.ResultsWe find a significant association between the severity of COVID-19 and the amount of platelet and platelet-leukocyte aggregates. Additionally, we observe a linkage between severity, aggregate composition, and size distribution of platelets in aggregates.ConclusionsThis study presents a POC-compatible method for rapid quantitative analysis of blood cell aggregates in patients with COVID-19.

Abstract 203: Predicting Outcomes After Mechanical Thrombectomy: Early vs Delayed Neurological Improvement in the SELECT Database

Introduction Understanding of patient benefit and improvement following mechanical thrombectomy in ischemic stroke has made tremendous strides in recent years, however clinicians still struggle to accurately predict patient outcomes and long‐term neurological functioning following large vessel occlusion and endovascular therapy. These limitations are further complicated by a highly variable time course of improvement following intervention that spans from minutes to months. Prior literature has shown that successful reperfusion is associated with better outcomes even in the absence of early improvement. Delayed neurological improvement may be the result of microvasculature effects and inflammatory changes, and is likely multifactorial. Improved understanding of predictors of early versus delayed neurological improvement in this population may help inform goals of care, aid in clinical decision making, and manage patient and family expectations. Methods Patient data from the SELECT database was retrospectively analyzed. The SELECT study was a prospective, non‐randomized, multicenter cohort study that enrolled patients with acute ischemic stroke due to anterior circulation large vessel occlusion at nine US centers between 1/2016 and 2/2018, with patients falling into either the mechanical thrombectomy or medical management arm. Only patients in the mechanical thrombectomy arm (n=149) were included in this analysis. Possible predictors of early vs late neurological improvement were segregated into pre‐procedure (age, NIHSS at baseline, IV thrombolytic therapy, transfer status, occlusion location, ischemic core, ischemic penumbra, hypoperfusion intensity ratio, successful reperfusion, type of anesthesia, time from last known well to puncture, time from puncture to reperfusion) and post‐procedure (NIHSS at 24‐hour follow‐up, ASPECTS on follow‐up imaging, ASPECTS point loss from baseline, infarct volume on follow‐up imaging, infarct growth from baseline ischemic core, neurological worsening, midline shift, hemorrhagic transformation and type). Backwards stepwise regression was used to identify significant predictors of early and delayed improvement. Results Pre‐procedure predictors of early neurological improvement (defined as mRS 0‐2 at discharge) were age (p=0.001), NIHSS score (p=0.002), ischemic core volume (p=0.001), transfer status (p=0.019), and successful reperfusion (defined as mTICI 2b or better, p=0.001). Post‐procedure predictor of early neurological improvement was NIHSS score at 24 hours (p&lt;0.001). Pre‐procedure predictor of delayed neurological improvement (defined as mRS 0‐2 at 90 days among those who did not achieve early neurological improvement) was ischemic core volume (p=0.003). Post‐procedure predictors of delayed neurological improvement were final infarct volume (p=0.012), change in ASPECTS on follow‐up imaging (p=0.001), and growth in infarct size (p=0.030). Pre‐procedure predictors of early excellent outcome (defined as mRS 0‐1 at discharge) were age (p=0.001), NIHSS score (p&lt;0.001), successful reperfusion (p=0.001), ischemic core volume (p=0.004), and infarct growth rate (p=0.034). Pre‐procedure predictor of delayed excellent outcome (defined as mRS 0‐1 at 90 days among those who did not achieve early excellent outcome) was NIHSS score (p=0.002). Additionally, midline shift of 5mm or more at 24 hours was associated with poor outcome at 90 days. Conclusion Several demographic, imaging, and clinical factors were found to correlate with either early or delayed neurological improvement. Further investigation may help delineate how these predictors can be utilized to inform clinical decisions and discussions regarding stroke patients receiving mechanical thrombectomy.

P-280 - Molecular Radiation Resistance Signature Induced by Alpha Emitters in Metastatic Castration-resistant Prostate Cancer: Early Clinical Outcome Prediction

P-280 - Molecular Radiation Resistance Signature Induced by Alpha Emitters in Metastatic Castration-resistant Prostate Cancer: Early Clinical Outcome Prediction