Outcome Prediction Models Research Articles

Development and Validation of a Risk Prediction Algorithm for Evaluating the Efficacy of Postoperative Adjuvant TACE Therapy for Hepatocellular Carcinoma.

There is a lack of a reliable outcome prediction model for patients evaluating the feasibility of postoperative adjuvant transarterial chemoembolization (PATACE) therapy. Our goal was to develop an easy-to-use tool specifically for these patients. From January 2013 to June 2017, patients with hepatocellular carcinoma from the Liver Center of the First Affiliated Hospital of Chongqing Medical University received postoperative adjuvant Transarterial chemoembolization (TACE) therapy after liver cancer resection. A Cox proportional hazards model was established for these patients, followed by internal validation (enhanced bootstrap resampling technique) to further evaluate the predictive performance and discriminanceevaluate the predictive performance and discriminance, and compare it with other predictive models. The prognostic factors considered included tumour number, maximum tumor diameter, Edmondson-Steiner (ES) grade, Microvascular invasion (MVI) grade, Ki67, age, sex, hepatitis B surface antigen, cirrhosis, Alpha-fetoprotein (AFP), Albumin-bilirubin (ALBI) grade, Childpugh grade, body mass index (BMI), Neutrophil-lymphocyte ratio (NLR), Platelet-to-lymphocyte ratio (PLR). The endpoint of the study was overall survival. The median overall survival was 36 (95%CI: 34.0-38.0) months, with 1-year, 2-year and 3-year survival rates being 96.3%, 84.0% and 75.3%, respectively. Tumour number, MVI grade, and BMI was incorporated into the model, which had good differentiation and accuracy. Internal validation (enhanced bootstrap) suggested that Harrell's C statistic is 0.72. The model consistently outperforms other currently available models. This model may be an easy-to-use tool for screening patients suitable for PA-TACE treatment and guiding the selection of clinical protocols. But further research and external validation are required.

Clinical prediction models for treatment outcomes in newly diagnosed epilepsy: A systematic review.

Up to 35% of individuals diagnosed with epilepsy continue to have seizures despite treatment, commonly referred to as drug-resistant epilepsy. Uncontrolled seizures can directly, or indirectly, negatively impact an individual's quality of life. To inform clinical management and life decisions, it is important to be able to predict the likelihood of seizure control. Those likely to achieve seizure control will be able to return sooner to their usual work and leisure activities and require less follow-up, whereas those with a poor prognosis will need more frequent clinical attendance and earlier consideration of epilepsy surgery. This is a systematic review aimed at identifying demographic, clinical, physiological (e.g., electroencephalographic), and imaging (e.g., magnetic resonance imaging) factors that may be predictive of treatment outcomes in patients with newly diagnosed epilepsy (NDE). MEDLINE and Embase were searched for prediction models of treatment outcomes in patients with NDE. Study characteristics were extracted and subjected to assessment of risk of bias (and applicability concerns) using the PROBAST (Prediction Model Risk of Bias Assessment Tool) tool. Baseline variables associated with treatment outcomes are reported as prognostic factors. After screening, 48 models were identified in 32 studies, which generally scored low for concerns of applicability, but universally scored high for susceptibility to bias. Outcomes reported fit broadly into four categories: drug resistance, short-term treatment response, seizure remission, and mortality. Prognostic factors were also heterogenous, but the predictors that were commonly significantly associated with outcomes were those related to seizure characteristics/types, epilepsy history, and age at onset. Antiseizure medication response was often included as a baseline variable, potentially obscuring other factor relationships at baseline. Currently, outcome prediction models for NDE demonstrate a high risk of bias. Model development could be improved with a stronger adherence to recommended TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) practices. Furthermore, we outline actionable changes to common practices that are intended to improve the overall quality of prediction model development in NDE.

Team Performance Indicators That Predict Match Outcome in Rugby Union

The aim of the study is to identify the most significant indicators of the national team's performance at the European Rugby Championships 15 and to design a model for predicting the outcomes of matches. Data was collected from teams’ performance at the European Rugby 15 Championships 2021, 2022 and 2023 for the analysis. The total number of matches was 41. All indicators presented in the official reports were taken: 22 for the home and away teams. The analysis of the team results was carried out according to all indicators: mean value, standard deviation, and test were used to compare the performance indicators of the winning and losing teams. Machine learning techniques were utilized to develop a predictive model for match outcomes. On one hand, 15 indicators (68.2%) are higher for teams that won (winning teams). On the other hand, 7 (31.8%) indicators are higher for teams that lost. The difference between the teams' means varies from -56.46% (the minus indicates that this indicator is higher for the teams that lost) to 273.68%. Based on the results, the Random Forest Classifier and Extra Trees Classifier algorithms have the best prediction accuracy (0.92). The most significant indicators of team performance that affect the final result of the match are tries (196.3% – the difference between the average values of winning and losing teams), conversions (176.7%), missed tackles (- 56.46%), offload (126.3%). Based on the data obtained, refining the team training process in Rugby 15 is possible.

The Effect of Situational Variables on Women’s Rink Hockey Match Outcomes

The main objective of the present study was to develop a concise predictive model to determine the likelihood of winning in female rink hockey based on various situational variables. Additionally, the study aimed to assess the individual impact of each predictor on match outcomes. The analysis encompassed a dataset of 840 matches during five consecutive seasons (from 2018–2019 to 2022–2023) in the Spanish first division (OkLiga). Employing the comprehensive method of all possible regressions, the most effective predictive logistic model for match outcomes was identified. This entire model featured five categorical predictor variables (match location, team level, opponent level, scoring first, and match status at halftime) and one binary outcome variable (match outcome). Subsequently, the final model, which exhibited a sensitivity and specificity surpassing 80% for a cut-off point of 0.439, emerged. This model was applied to predict winning a match in 18 frequent situations determined from a two-step cluster analysis. Within this predictive framework, match status at halftime emerged as the most influential predictor impacting the match outcome, followed by opponent level, team level, and match location. The implications of our findings extend to rink hockey coaches and practitioners. Recognizing the significant impact of situational variables on match outcomes empowers them to customize game plans and design more specific strategies, thereby enhancing game understanding and elevating the overall performance.

An Application of Advanced Sentiment Analysis on X (Twitter) Utilising Large Language Models for the Precise Prediction of Election Outcomes

An Application of Advanced Sentiment Analysis on X (Twitter) Utilising Large Language Models for the Precise Prediction of Election Outcomes

Pre-Operative Adiposity and Synovial Fluid Inflammatory Biomarkers Provide a Predictive Model for Post-Operative Outcomes Following Total Joint Replacement Surgery in Osteoarthritis Patients

A proportion of osteoarthritis (OA) patients are unsatisfied with post-operative outcomes following total joint replacement surgery (TJR), with insufficient pain relief or poor functional improvement. Predicting those who will have poor outcomes would be beneficial for patients and clinicians. The aim of this study was to determine the relationship between baseline anthropometric data and the concentration of pre-operative serum and peri-operative synovial fluid (SF) cytokines and 7-month post-operative outcomes in a cohort of knee and hip OA patients. 160 OA patients were recruited who were scheduled for TJR. The concentration of 24 cytokines was measured in blood and SF by multiplex assay. EQ5D index health status was assessed pre-operatively and at 7 months post-operatively. 13% of patients were identified as non-responders based on EQ5D index. Compared to responders, non-responders were of higher body mass index (BMI), had greater waist and hip circumference, and had higher levels of SF leptin but lower levels of SF resistin (p < 0.05). Linear regression analysis found a significant but weak relationship between pre-operative body weight and post-operative response (ΔEQ5D index; r = 0.222, p = 0.049). The combination of body weight with SF amphiregulin and SF IL-6 provided an improved predictive model of post-operative response (r = 0.470, p = 0.035).

Immune-related LncRNAs scores predicts chemotherapeutic responses and prognosis in cervical cancer patients

BackgroundLong non-coding RNAs (LncRNAs) regulating the immune microenvironment of cancer is a hot spot. But little is known about the influence of the immune-related lncRNA (IRlncRs) on the chemotherapeutic responses and prognosis of cervical cancer (CC) patients. The purpose of the study was to identify an immune-related lncRNAs (IRlncRs)-based model for the prospective prediction of clinical outcomes in CC patients.MethodsCC patients’ relevant data was acquired from The Cancer Genome Atlas (TCGA). Correlation analysis and Cox regression analyses were applied. A risk score formula was formulated. Prognostic factors were combined into a nomogram, while sensitivity for chemotherapy drugs was analyzed using the OncoPredict algorithm.ResultsEight optimal IRlncRs(ATP2A1-AS1, LINC01943, AL158166.1, LINC00963, AC009065.8, LIPE-AS1, AC105277.1, AC098613.1.) were incorporated in the IRlncRs model. The overall survival (OS) of the high-risk group of the model was inferior to those in the low-risk group. Further analysis demonstrated this eight-IRlncRs model as a useful prognostic marker. The Nomogram had a concordance index of survival prediction of 0.763(95% CI 0.746–0.780) and more robust predictive accuracy. Furthermore, patients in the low-risk group were found to be more sensitive to chemotherapy, including Paclitaxel, Rapamycin, Epirubicin, Vincristine, Docetaxel and Vinorelbine.ConclusionsAn eight-IRlncRs-based prediction model was identified that has the potential to be an important tool to predict chemotherapeutic responses and prognosis for CC patients.

Development and external validation of preoperative clinical prediction models for postoperative outcomes including preoperative aerobic fitness in patients approaching elective colorectal cancer surgery

IntroductionPreoperative aerobic fitness is associated with postoperative outcomes after elective colorectal cancer (CRC) surgery. This study aimed to develop and externally validate two clinical prediction models incorporating a practical test to assess preoperative aerobic fitness to distinguish between patients with and without an increased risk for 1) postoperative complications and 2) a prolonged time to in-hospital recovery of physical functioning after elective colorectal cancer (CRC) surgery. Materials and methodsModels were developed using prospective data from 256 patients and externally validated using prospective data of 291 patients. Postoperative complications were classified according to Clavien-Dindo. The modified Iowa level of assistance scale (mILAS) was used to determine time to postoperative in-hospital physical recovery. Aerobic fitness, age, sex, body mass index, American Society of Anesthesiologists (ASA) classification, neoadjuvant treatment, surgical approach, tumour location, and preoperative haemoglobin level were potential predictors. Areas under the curve (AUC), calibration plots, and Hosmer-Lemeshow tests evaluated predictive performance. ResultsAerobic fitness, sex, age, ASA, tumour location, and surgical approach were included in the final models. External validation of the model for complications and postoperative recovery presented moderate to fair discrimination (AUC 0.666 (0.598–0.733) and 0.722 (0.651–0.794), respectively) and good calibration. High sensitivity and high negative predictive values were observed in the lower predicted risk categories (<40 %). ConclusionBoth models identify patients with and without an increased risk of complications or a prolonged time to in-hospital physical recovery. They might be used for improving patient-tailored preoperative risk assessment and targeted and cost-effective application of prehabilitation interventions.

Addressing racial disparities in prostate cancer pathology prediction models: external validation and comparison of four models of pathological outcome prediction before radical prostatectomy in the multiethnic SEARCH cohort.

Certain widely used pathological outcome prediction models that were developed in tertiary centers tend to overpredict outcomes in the community setting; thus, the Michigan Urological-Surgery Improvement Collaborative (MUSIC) model was developed in general urology practice to address this issue. Additionally, the development of these models involved a relatively small proportion of Black men, potentially compromising the accuracy of predictions in this patient group. We tested the validity of the MUSIC and three widely used nomograms to compare their overall and race-stratified predictive performance. We extracted data from 4139 (1138 Black) men from the Shared Equal Access Regional Cancer Hospital (SEARCH) database of the Veterans Affairs health system. The predictive performance of the MUSIC model was compared to the Memorial-Sloan Kettering (MSK), Briganti-2012, and Partin-2017 models for predicting lymph-node invasion (LNI), extra-prostatic extension (EPE), and seminal vesicle invasion (SVI). The median PSA of Black men was higher than White men (7.8 vs. 6.8 ng/ml), although they were younger by a median of three years and presented at a lower-stage disease. MUSIC model showed comparable discriminatory capacity (AUC:77.0%) compared to MSK (79.2%), Partin-2017 (74.6%), and Briganti-2012 (76.3%), with better calibration for LNI. AUCs for EPE and SVI were 72.7% and 76.9%, respectively, all comparable to the MSK and Partin models. LNI AUCs for Black and White men were 69.6% and 79.6%, respectively, while EPE and SVI AUCs were comparable between races. EPE and LNI had worse calibration in Black men. Decision curve analysis showed MUSIC superiority over the MSK model in predicting LNI, especially among Black men. Although the discriminatory performance of all models was comparable for each outcome, the MUSIC model exhibited superior net benefit to the MSK model in predicting LNI outcomes among Black men in the SEARCH population.

Refining outcome prediction after traumatic brain injury with machine learning algorithms.

Outcome after traumatic brain injury (TBI) is typically assessed using the Glasgow outcome scale extended (GOSE) with levels from 1 (death) to 8 (upper good recovery). Outcome prediction has classically been dichotomized into either dead/alive or favorable/unfavorable outcome. Binary outcome prediction models limit the possibility of detecting subtle yet significant improvements. We set out to explore different machine learning methods with the purpose of mapping their predictions to the full 8 grade scale GOSE following TBI. The models were set up using the variables: age, GCS-motor score, pupillary reaction, and Marshall CT score. For model setup and internal validation, a total of 866 patients could be included. For external validation, a cohort of 369 patients were included from Leuven, Belgium, and a cohort of 573 patients from the US multi-center ProTECT III study. Our findings indicate that proportional odds logistic regression (POLR), random forest regression, and a neural network model achieved accuracy values of 0.3-0.35 when applied to internal data, compared to the random baseline which is 0.125 for eight categories. The models demonstrated satisfactory performance during external validation in the data from Leuven, however, their performance were not satisfactory when applied to the ProTECT III dataset.

Use of a commercial artificial intelligence-based mammography analysis software for improving breast ultrasound interpretations.

To evaluate the use of a commercial artificial intelligence (AI)-based mammography analysis software for improving the interpretations of breast ultrasound (US)-detected lesions. A retrospective analysis was performed on 1109 breasts that underwent both mammography and US-guided breast biopsy. The AI software processed mammograms and provided an AI score ranging from 0 to 100 for each breast, indicating the likelihood of malignancy. The performance of the AI score in differentiating mammograms with benign outcomes from those revealing cancers following US-guided breast biopsy was evaluated. In addition, prediction models for benign outcomes were constructed based on clinical and imaging characteristics with and without AI scores, using logistic regression analysis. The AI software had an area under the receiver operating characteristics curve (AUROC) of 0.79 (95% CI, 0.79-0.82) in differentiating between benign and cancer cases. The prediction models that did not include AI scores (non-AI model), only used AI scores (AI-only model), and included AI scores (integrated model) had AUROCs of 0.79 (95% CI, 0.75-0.83), 0.78 (95% CI, 0.74-0.82), and 0.85 (95% CI, 0.81-0.88) in the development cohort, and 0.75 (95% CI, 0.68-0.81), 0.82 (95% CI, 0.76-0.88), and 0.84 (95% CI, 0.79-0.90) in the validation cohort, respectively. The integrated model outperformed the non-AI model in the development and validation cohorts (p < 0.001 for both). The commercial AI-based mammography analysis software could be a valuable adjunct to clinical decision-making for managing US-detected breast lesions. The commercial AI-based mammography analysis software could potentially reduce unnecessary biopsies and improve patient outcomes. • Breast US has high rates of false-positive interpretations. • A commercial AI-based mammography analysis software could distinguish mammograms having benign outcomes from those revealing cancers after US-guided breast biopsy. • A commercial AI-based mammography analysis software may improve interpretations for breast US-detected lesions.

Development, Validation and Clinical Utility of a Risk Prediction Model for Maternal and Neonatal Adverse Outcomes in Pregnant Women with Hypothyroidism.

This study aimed to create, verify and assess the clinical utility of a prediction model for maternal and neonatal adverse outcomes in pregnant women with hypothyroidism. A prediction model was developed, and its accuracy was tested using data from a retrospective cohort. The study focused exclusively on female patients diagnosed with hypothyroidism who were admitted to a tertiary hospital. The development and validation cohort comprised individuals who gave birth between 1 October 2020 and 31 December 2022. The primary outcome was a combination of crucial maternal and newborn problems (eg premature births, abortions and neonatal asphyxia). The prediction model was developed using logistic regression. Evaluation of the model's performance was conducted based on its ability to discriminate, calibrate and provide clinical value. In total, nine variables were chosen to develop the predictive model for adverse maternal and neonatal outcomes during pregnancy with hypothyroidism. The area under the curve of the model for predicting maternal adverse outcomes was 0.845, and that for predicting neonatal adverse outcomes was 0.685. The calibration plots showed good agreement between the nomogram predictions and the actual observations in both the training and validation cohorts. Furthermore, decision curve analysis suggested that the nomograms were clinically useful and had good discriminative power to identify high-risk mother-infant cases. Two models to predict the risk probability of maternal and neonatal adverse outcomes in pregnant women with hypothyroidism were developed and verified to assist physicians in evaluating maternal and neonatal adverse outcomes throughout pregnancy with hypothyroidism and to facilitate decision-making regarding therapy.

Machine learning prediction model of major adverse outcomes after pediatric congenital heart surgery: a retrospective cohort study.

Major adverse postoperative outcomes (APOs) can greatly affect mortality, hospital stay, care management and planning, and quality of life. This study aimed to evaluate the performance of five machine learning (ML) algorithms for predicting four major APOs after pediatric congenital heart surgery and their clinically meaningful model interpretations. Between August 2014 and December 2021, 23000 consecutive pediatric patients receiving congenital heart surgery were enrolled. Based on the split date of 1 January 2019, the authors selected 13927 participants for the training cohort, and 9073 participants for the testing cohort. Four predefined major APOs including low cardiac output syndrome (LCOS), pneumonia, renal failure, and deep venous thrombosis (DVT) were investigated. Thirty-nine clinical and laboratory features were inputted in five ML models: light gradient boosting machine (LightGBM), logistic regression (LR), support vector machine, random forest, and CatBoost. The performance and interpretations of ML models were evaluated using the area under the receiver operating characteristic curve (AUC) and Shapley Additive Explanations (SHAP). In the training cohort, CatBoost algorithms outperformed others with the mean AUCs of 0.908 for LCOS and 0.957 for renal failure, while LightGBM and LR achieved the best mean AUCs of 0.886 for pneumonia and 0.942 for DVT, respectively. In the testing cohort, the best-performing ML model for each major APOs with the following mean AUCs: LCOS (LightGBM), 0.893 (95% CI: 0.884-0.895); pneumonia (LR), 0.929 (95% CI: 0.926-0.931); renal failure (LightGBM), 0.963 (95% CI: 0.947-0.979), and DVT (LightGBM), 0.970 (95% CI: 0.953-0.982). The performance of ML models using only clinical variables was slightly lower than those using combined data, with the mean AUCs of 0.873 for LCOS, 0.894 for pneumonia, 0.953 for renal failure, and 0.933 for DVT. The SHAP showed that mechanical ventilation time was the most important contributor of four major APOs. In pediatric congenital heart surgery, the established ML model can accurately predict the risk of four major APOs, providing reliable interpretations for high-risk contributor identification and informed clinical decisions-making.

318 Discovering Subgroups with Supervised Machine Learning Models for Heterogeneity of Treatment Effect Analysis

OBJECTIVES/GOALS: The goal of the study is to provide insights into the use of machine learning methods as a means to predict heterogeneity of treatment effect (HTE) in participants of randomized clinical trials. METHODS/STUDY POPULATION: Using data from 2,441 participants enrolled in the ASPirin in Reducing Events in the Elderly (ASPREE) randomized controlled trial of daily low-dose aspirin vs placebo in the United States, we developed multivariable risk prediction models for the composite outcome of dementia, disability, or death. We used two machine learning techniques, decision trees and random forests, to develop novel non-parametric outcomes classifiers and generate risk-based subgroups. The comparator method was an extant semi-parametric proportional hazards predictive risk model. We then assessed HTE by examining the 5-year absolute risk reduction (ARR) of aspirin vs placebo in each risk subgroup. RESULTS/ANTICIPATED RESULTS: In the random forest classifier, the ARR at 5 years in the highest risk quintile was 13.7% (95% CI 3.1% to 24.4%). For the semi-parametric proportional hazards model, the ARR in the highest risk quintile was 15.1% (95% CI 4.0% to 26.3%). These results were comparable and provide evidence of the viability of internally developed parsimonious non-parametric machine learning models for HTE analysis. The decision tree model results (5-year ARR = 17.0%, 95% CI= -5.4% to 39.4% in the highest risk subgroup) exhibited more uncertainty in the results. DISCUSSION/SIGNIFICANCE: None of the models detected significant HTE on the relative scale; there was substantial HTE on the absolute scale in three of the models. Treatment benefit on the absolute scale may be regarded as bearing greater clinical importance and may be present even in the absence of benefit on the relative scale.

Mental Health Prediction Using Catboost Algorithm

Abstract: This study investigates the application of the CatBoost algorithm in predicting mental health outcomes using Python programming language. Mental health prediction is a critical area of research due to its significant impact on individuals and society. Traditional predictive modeling techniques often encounter challenges in handling complex and highdimensional data inherent in mental health datasets. CatBoost , a state- of-the-art gradient boosting algorithm, has shown promise in effectively addressing these challenges by handling categorical variables seamlessly and exhibiting robust performance in various domains. Leveraging its powerful capabilities, this study aims to develop predictive models for mental health outcomes utilizing a comprehensive dataset encompassing diverse socio- demographic, behavioural , and clinical factors. The predictive performance of the CatBoost algorithm will be evaluated and compared against other commonly used machine learning algorithms, demonstrating its effectiveness in accurately predicting mental health outcomes. This research contributes to the advancement of predictive modeling in mental health research and holds potential implications for personalized interventions and resource allocation in mental healthcare systems

Prognostic Models for Mortality and Morbidity in Heart Failure With Preserved Ejection Fraction

Accurate risk prediction of morbidity and mortality in patients with heart failure with preserved ejection fraction (HFpEF) may help clinicians risk stratify and inform care decisions. To develop and validate a novel prediction model for clinical outcomes in patients with HFpEF using routinely collected variables and to compare it with a biomarker-driven approach. Data were used from the Dapagliflozin Evaluation to Improve the Lives of Patients With Preserved Ejection Fraction Heart Failure (DELIVER) trial to derive the prediction model, and data from the Angiotensin Receptor Neprilysin Inhibition in Heart Failure With Preserved Ejection Fraction (PARAGON-HF) and the Irbesartan in Heart Failure With Preserved Ejection Fraction Study (I-PRESERVE) trials were used to validate it. The outcomes were the composite of HF hospitalization (HFH) or cardiovascular death, cardiovascular death, and all-cause death. A total of 30 baseline candidate variables were selected in a stepwise fashion using multivariable analyses to create the models. Data were analyzed from January 2023 to June 2023. Models to estimate the 1-year and 2-year risk of cardiovascular death or hospitalization for heart failure, cardiovascular death, and all-cause death. Data from 6263 individuals in the DELIVER trial were used to derive the prediction model and data from 4796 individuals in the PARAGON-HF trial and 4128 individuals in the I-PRESERVE trial were used to validate it. The final prediction model for the composite outcome included 11 variables: N-terminal pro-brain natriuretic peptide (NT-proBNP) level, HFH within the past 6 months, creatinine level, diabetes, geographic region, HF duration, treatment with a sodium-glucose cotransporter 2 inhibitor, chronic obstructive pulmonary disease, transient ischemic attack/stroke, any previous HFH, and heart rate. This model showed good discrimination (C statistic at 1 year, 0.73; 95% CI, 0.71-0.75) in both validation cohorts (C statistic at 1 year, 0.71; 95% CI, 0.69-0.74 in PARAGON-HF and 0.75; 95% CI, 0.73-0.78 in I-PRESERVE) and calibration. The model showed similar discrimination to a biomarker-driven model including high-sensitivity cardiac troponin T and significantly better discrimination than the Meta-Analysis Global Group in Chronic (MAGGIC) risk score (C statistic at 1 year, 0.60; 95% CI, 0.58-0.63; delta C statistic, 0.13; 95% CI, 0.10-0.15; P < .001) and NT-proBNP level alone (C statistic at 1 year, 0.66; 95% CI, 0.64-0.68; delta C statistic, 0.07; 95% CI, 0.05-0.08; P < .001). Models derived for the prediction of all-cause and cardiovascular death also performed well. An online calculator was created to allow calculation of an individual's risk. In this prognostic study, a robust prediction model for clinical outcomes in HFpEF was developed and validated using routinely collected variables. The model performed better than NT-proBNP level alone. The model may help clinicians to identify high-risk patients and guide treatment decisions in HFpEF.

Functional Outcomes and Symptomatic Intracranial Hemorrhage After Endovascular Treatment in Acute Vertebrobasilar Artery Occlusions: External Validation of Prediction Models

Background Vertebrobasilar artery occlusion (VBAO) is a severe type of stroke. Multiple prediction models for outcome and symptomatic intracranial hemorrhage (sICH) of patients with acute ischemic stroke treated with endovascular treatment have been developed to improve patient management, but few are based on VBAO. This study aimed to provide an overview of published models to predict functional outcome and sICH as well as to validate their ability in patients with acute VBAO treated with endovascular treatment. Methods We performed a systematic search to identify models either developed or validated to predict functional outcomes or sICH after endovascular treatment. Models were externally validated in the Posterior Circulation Ischemic Stroke Registry (PERSIST) study (n = 2422). Outcome measures included the modified Rankin Scale (mRS) score at 90 days and sICH. Model performance was evaluated with discrimination (c‐statistic) and calibration (slope and intercept). Results A total of 65 models were included in overview. The most frequently used predictors were baseline National Institutes of Health Stroke Scale score (n = 57), age (n = 45), and glucose (n = 32). In the external validation cohort, 777 of 2353 patients (33.0%) achieved mRS score 0–2 at 90 days, 1061 of 2353 patients (45.1%) patients achieved mRS score 0–3 at 90 days, and sICH occurred in 170 of 2422 patients (7.0%). Finally, 27 models were included in external validation. For functional outcome models focusing on mRS score 0–2/3–6, discrimination ranged from 0.63 to 0.66 and best calibrated model was SC (Stroke Checkerboard) (intercept, −0.13 [95% CI, −0.27 to 0.01]; slope, 0.92 [95% CI, 0.67–1.17]). For functional outcome models focusing on mRS score 0–3/4–6, discrimination ranged from 0.64 to 0.74 and best calibrated model was modified Houston Intra‐Arterial Therapy 2 (mHIAT2) (intercept, −0.12 [95% CI, −0.31 to 0.07]; slope, 0.85 [95% CI, 0.65–1.04]). For sICH models, discrimination ranged from 0.53 to 0.83 and best calibrated model was Thrombolysis in Cerebral Infarction score, Alberta Stroke Program Early CT Score, and glucose (TAG) (intercept, 0.13 [95% CI, −0.25 to 0.51]; slope, 0.93 [95% CI, 0.63–1.23]). Conclusions The currently published models are inadequate for predicting functional outcomes and sICH in patients with acute VBAO undergoing endovascular treatment and, therefore, there is a need for more effective models specifically developed for VBAO conditions.

Abstract 4868: Avoiding the pitfalls in developing and validating prediction models for rare outcomes in nested case-control

Abstract Background: Molecular-biomarker-based prediction models, whether based on statistics or artificial intelligence, are becoming increasingly popular in the medical community as an alternative or complement to more traditional prediction tools, such as staging systems. Population-based cohorts are the preferred study design for building and validating such models, but they are generally expensive and frequently unfeasible, particularly, when the outcome of interest is rare, and when the collected data include expensive or difficult-to-obtain molecular biomarkers. As such, studies often develop or validate risk models in cohorts obtained based on data availability, rather than on the representativeness of a target population. Nested case-control (NCC) is an efficient study design for building and validating such models in a rare outcome setting, but it requires a fully enumerated source population, and appropriate methodologies to accommodate the under-sampling of the controls and any matching. Unfortunately, these methodologies are not systematically described in a form easily accessible to clinicians. Methods: We systematically characterized how to correctly develop and evaluate the performance of prediction models in NCC cohorts, by weighing the subjects based on sampling probabilities to account for the sub-cohort sampling. We illustrated the use of weighted metrics for the NCC, with a validation of a model that predicts breast cancer development - BOADICEA version 5 - in the population-based Rotterdam study. We used the C-index, threshold-based metrics, observed-to-expected events ratio (O/E ratio), calibration slope, and decision curve analysis as performance metrics. We compared the metrics obtained in the full cohort with those obtained in NCC cohorts sampled from the Rotterdam study, with and without a matched design. Results: NCC cohorts of 326 women were derived from the full cohort of 4,377 women from the Rotterdam Study. Performance metrics without weight adjustment were biased on the NCC cohorts: the unweighted C-index of the BOADICEA model was 0.61 (0.58-0.63) for the unmatched design. However, with weight adjustment, the C-index in the NCC cohorts (0.65 (0.61-0.69)) corresponds to the C-index in the full cohort (0.65 (0.62-0.69)), despite the NCC cohort being much smaller in size. Similarly, weighted adjustments of O/E ratio, threshold-based metrics, and net benefit for decision curves were unbiased estimates of the corresponding metrics in the full cohort, while the corresponding unweighted metrics were biased. Conclusions: We provided a practical guide for clinicians on how to develop and validate prediction models in NCC cohorts. We showed that this study design is an efficient solution, in case of expensive or difficult-to-obtain biomarkers, and when the outcome is rare; but the performance metrics must be appropriately adjusted to the sampling procedure. Citation Format: Barbara Rentroia-Pacheco, Domenico Bellomo, Inge MM Lakeman, Marlies Wakkee, Loes M. Hollestein, David van Klaveren. Avoiding the pitfalls in developing and validating prediction models for rare outcomes in nested case-control [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 4868.

EIF2Ss, a Novel c-Myc-Correlated Gene Family, is Associated with Poor Prognosis and Immune Infiltration in Pancreatic Adenocarcinoma.

Pancreatic adenocarcinoma (PAAD) is a highly malignant tumor in urgent need of novel diagnostics, prognostic markers, and treatments. Eukaryotic translation initiation factor 2 subunits (EIF2Ss), comprising Eukaryotic translation initiation factor 2 subunit alpha (EIF2S1), Eukaryotic translation initiation factor 2 subunit beta (EIF2S2), and Eukaryotic translation initiation factor 2 subunit gamma (EIF2S3), is a family of eukaryotic initiation factors that participate in early protein synthesis and are crucial for tumor initiation and progression. However, the role of EIF2Ss in PAAD has yet to be reported. The aim of this study was therefore to analyze EIF2Ss in relation to the diagnosis, prognosis, and treatment of PAAD. The cancer genome atlas (TCGA) database was used to investigate gene expression and patient survival. Gene alterations, immune cell infiltration, and immune checkpoints in PAAD were also evaluated. Univariate and multivariate analysis, nomograms, calibration curves, and Decision Curve Analysis (DCA) diagrams were used to develop and evaluate a prediction model for patient outcome. Single-cell RNA-seq (scRNA) analysis, functional enrichment, co-IP assay, mass spectrometry, and western blot were used to study the relationship between EIF2Ss and c-myc in PAAD. EIF2Ss are over-expressed in PAAD tissue and are associated with poor prognosis. The frequency of EIF2S1, EIF2S2, and EIF2S3 gene alteration in PAAD was 0.2%, 0.4%, and 0.2%, respectively. High EIF2Ss expression was associated with Th2 cell infiltration, whereas low expression was associated with pDC infiltration. Moreover, EIF2Ss expression was positively correlated with the expression of the NT5E, ULBP1, PVR, CD44, IL10RB, and CD276 checkpoints. A prediction model developed using EIF2Ss and important clinicopathologic features showed good predictive value for the overall survival of PAAD patients. ScRNA-Seq data showed that EIF2Ss was associated with enrichment for endothelial cells, fibroblasts, malignant cells, and ductal cells. EIF2Ss expression was also correlated with adipogenesis, interferon-alpha response, epithelial-mesenchymal transition, myc targets, G2M checkpoint, oxidative phosphorylation, and hypoxia. Functional enrichment analysis of EIF2Ss showed a close correlation with the myc pathway, and interactions between EIF2Ss and c-myc were confirmed by co-IP assay and mass spectrometry. Importantly, knockdown of c-myc decreased the expression of EIF2S1, EIF2S2, and EIF2S3 in PAAD cells. EIF2Ss were found to have significant clinical implications for the prognosis and treatment of PAAD. Inhibition of c-myc caused the downregulation of EIF2S1, EIF2S2, and EIF2S3 expression.

Point of View on Outcome Prediction Models in Post-Stroke Motor Recovery

Stroke is a leading cause of disability worldwide which can cause significant and persistent upper limb (UL) impairment. It is difficult to predict UL motor recovery after stroke and to forecast the expected outcomes of rehabilitation interventions during the acute and subacute phases when using clinical data alone. Accurate prediction of response to treatment could allow for more timely and targeted interventions, thereby improving recovery, resource allocation, and reducing the economic impact of post-stroke disability. Initial motor impairment is currently the strongest predictor of post-stroke motor recovery. Despite significant progress, current prediction models could be refined with additional predictors, and an emphasis on the time dependency of patient-specific predictions of UL recovery profiles. In the current paper a panel of experts provide their opinion on additional predictors and aspects of the literature that can help advance stroke outcome prediction models. Potential strategies include close attention to post-stroke data collection timeframes and adoption of individual-computerized modeling methods connected to a patient’s health record. These models should account for the non-linear and the variable recovery pattern of spontaneous neurological recovery. Additionally, input data should be extended to include cognitive, genomic, sensory, neural injury, and function measures as additional predictors of recovery. The accuracy of prediction models may be further improved by including standardized measures of outcome. Finally, we consider the potential impact of refined prediction models on healthcare costs.