Model For Prediction Research Articles

Predicting Hass Avocado Maturity with NIR Spectroscopy for Non-invasive Dry Matter Estimation in Hass Avocados

Aims: To develop a rapid, non-invasive method for predicting Hass avocado maturity using near- infrared diffuse reflectance spectroscopy (NIR-DRS) combined with machine learning algorithms, and to identify the optimal NIR wavelength range for accurate dry matter content prediction. Study Design: An experimental design involving spectral data collection from Hass avocados and the development of machine learning models for dry matter prediction. Methodology: Spectral data from 200 Hass avocados were collected using near-infrared diffuse reflectance spectroscopy (900-2500 nm). To improve the quality of the spectral data and reduce noise, standard normal variate was used to correct for scattering and remove unwanted variability in the spectral data. PCA was then performed to reduce the dimension of the spectral data while retaining the most significant variance. Following preprocessing, machine learning models, including Convolutional Neural Networks (CNN), were trained to predict dry matter content, and the optimal wavelength range was determined for accurate prediction. Results: The CNN model demonstrated superior performance for dry matter prediction with R² of 0.91 in the testing set. The wavelength range of 1000-1500 nm was identified as optimal, offering accurate predictions while reducing computational complexity. This range shows potential for developing cost-effective NIR devices for real-time maturity assessment. Conclusion: NIR spectroscopy combined with machine learning offers a non-invasive, accurate method for predicting avocado dry matter, with potential applications for quality control in the avocado industry. The findings demonstrate that focusing on specific wavelength ranges can lead to more affordable and efficient NIR solutions.

Generalization Investigation in Heart Disease Prediction: Comparative Analysis of MLP, Random Forest, and SVM with Cross-Domain Datasets

The traditional way of predicting heart disease is usually through the subjective judgment of doctors, which is high subjectivity. Current machine learning methods do not pay enough attention to its universality. The two datasets used in this study are from Kaggle and preprocessed in advance, including standardization by using Z-Score and dimensionality reduction by KPCA. The two datasets were designated as the source domain and the target domain. The source domain dataset was further divided into two subsets with an 80:20 split, where 80% was used for training the model and the remaining 20% served as the test set. The trained model was then applied to the target domain to compare the differences in prediction results. Through the exploration of Multilayer Perceptron (MLP), Random Forest and Support Vector Machine (SVM), for the data set used in this study, when MLP uses a simple structure, the difference in prediction accuracy dropped from 13.8% to 5.07. For Random Forest, by increasing the number of decision trees and decreasing the minimum number of samples required for splitting, the difference is reduced from 15.6% to 9.32%. By modifying the penalty parameter C value of SVM, the difference on different datasets is reduced from 13.39% to 4.46%. This study is one of the few to explore the generalizations of heart disease prediction. The results demonstrate that the generalization performance of the model for heart disease prediction can be significantly enhanced through appropriate modifications to its structure and hyperparameters.

Predictive modeling of bronchopulmonary dysplasia in premature infants: the impact of new diagnostic standards

AimTo provide a risk prediction for bronchopulmonary dysplasia (BPD) in premature infants under the new diagnostic criteria and establish a prediction model.MethodsIn this study, we retrospectively collected case data on preterm infants admitted to the NICU from August 2015 to August 2018. A lasso analysis was performed to identify the risk factors associated with the development of BPD. A nomogram predictive model was constructed in accordance with the new diagnostic criteria for BPD.ResultA total of 276 preterm infants were included in the study.The incidence of BPD under the 2018 diagnostic criteria was 11.2%. Mortality was significantly higher in the BPD group than the non-BPD group under the 2018 diagnostic criteria (P < 0.05). Fourteen possible variables were selected by the Lasso method, with a penalty coefficient λ=0.0154. The factors that eventually entered the logistic regression model included birth weight [BW, OR = 0.9945, 95% CI: 0.9904–0.9979], resuscitation way (OR = 4.8249, 95% CI: 1.3990–19.4752), intrauterine distress (OR = 8.0586, 95% CI: 1.7810–39.5696), score for SNAPPE-II (OR = 1.0880, 95% CI: 1.0210–1.1639), hematocrit (OR = 1.1554, 95% CI: 1.0469–1.2751) and apnea (OR = 7.6916, 95% CI: 1.4180–52.1236). The C-index after adjusting for fitting deviation was 0.894.ConclusionThis study made a preliminary exploration of the risk model for early prediction of BPD and indicated good discrimination and calibration in premature infants.

Transcriptional response of Methanosarcina acetivorans to repression of the energy-conserving methanophenazine: CoM-CoB heterodisulfide reductase enzyme HdrED

ABSTRACT Methane-producing archaea are key organisms in the anaerobic carbon cycle. These organisms, also called methanogens, grow by converting substrate to methane gas in a process called methanogenesis. Previous research showed that the reduction of the terminal electron acceptor is the rate-limiting step in methanogenesis by Methanosarcina acetivorans . In order to gain insight into how the cells sense and respond to the availability of the terminal electron acceptor, we designed an experiment to deplete cells of the essential terminal oxidase enzyme, HdrED. We found that the depletion of HdrED in vivo results in a higher abundance of transcripts for methyltransferases ( mtaC2, mtaB3, mtaC3 ), coenzyme B biosynthesis, C1 metabolism, and pyrimidine compounds. In most cases, these changes were distinct from transcript abundance changes observed during the transition from exponential growth to stationary phase cultures. These data implicate the methylotrophic methanogenesis regulator MsrC (MA4383) in CoM-S-S-CoB heterodisulfide sensing and indicate cells have a specific mechanism to sense intracellular ratio of CoM-S-S-CoB, coenzyme M, and coenzyme B thiols and further suggest transcripts encoding translation and methanogenesis functions are controlled by feed-forward regulation depending on substrate availability. IMPORTANCE Methanosarcina is an emerging model archaeon and synthetic biology platform for the production of renewable energy and sustainable chemicals to reduce dependence on petroleum. Research into metabolic networks and gene regulation in this organism and other methanogens will inform genome-scale metabolic modeling and microbial function prediction in uncultured or non-model anaerobes and archaea. This study suggests methanogens use unknown mechanisms to efficiently couple methanogenesis to gene regulation via CoM-S-S-CoB and ATP availability.

AGCNAF: predicting disease-gene associations using GCN and multi-head attention to fuse the similarity features

Abstract Association prediction between diseases and genes is a critical step in revealing the molecular mechanisms of diseases and developing drug treatment strategies. With the explosive growth of data in the biomedical field, how to effectively utilize these data for accurate prediction has become a hotspot and challenge in current research. To overcome the limitations of current prediction methods in dealing with complex biological network structures and feature extraction, this study proposes AGCNAF, a method that combines an unsupervised Graph Convolutional Network (GCN) and a multi-head attention mechanism. The metagraph-guided random walk strategy enables AGCNAF to capture local and high-order topological structures in the graph, while GCN is responsible for realizing deep feature extraction of these structures. By incorporating similarity features through the multi-attention mechanism, AGCNAF achieves effective integration of global and local features, which significantly improves the prediction performance. By incorporating similarity features through the multi-attention mechanism, AGCNAF achieves effective integration of global and local features, which significantly improves the prediction performance. By utilizing the machine learning binary classification model for prediction, the experimental results through five-fold cross-validation show that AGCNAF demonstrates significant advantages in prediction performance compared to existing methods, with its AUC and AUPR reaching 0.9686 and 0.9709, respectively, and the AUC up to 0.9812 under specific conditions. 
To verify the practical application value of AGCNAF, this study also conduct case studies on Alzheimer's disease, lung cancer, and breast cancer. The results further confirm the excellent performance of AGCNAF in identifying potential disease-gene associations, which opens up new possibilities for future disease-gene research.

Comprehensive machine learning models for prediction of heart failure in 476,393 women and men from the UK Biobank reveal sex differences and underutilized risk factors

Abstract Background Machine learning (ML) models provide potential advantage over ‘traditional’ regression models in heart failure (HF) prediction. Objective To compare performances of Cox PH models and ML survival models for incident HF in men and women without prevalent ischemic heart disease (IHD). We also aimed to identify potential high-risk precursors otherwise ignored by conventional survival models, and to investigate differences between sex-specific models. Methods We included 476,393 participants (55.6% women) from the UK Biobank, after excluding participants with a history of HF or IHD, and defined sex-specific datasets. We predicted incident HF events using over 400 baseline characteristics. We constructed multivariable Cox PH models, which included all predictor variables and subsequently only those remaining after LASSO stability selection. We also developed two supervised ML models (Random Survival Forest (RSF), eXtreme Gradient Survival Boosting (XGBoost)). We identified the 15 most important sex-specific predictors in each model and performances were compared using the C-index. Models were validated using hold-out sets. Results During 12.3 ± 1.9 years of follow-up, 4680 (1.76%) women and 6631 (3.14%) men developed incident HF. XGBoost showed the best performance during model training (C-index, training: 0.89 in men, 0.97 in women; validation 0.77 in men, 0.80 in women). The multivariable Cox model performed second-best (C-index, training: 0.78 in men, 0.82 in women; validation: 0.76 in men, 0.78 in women). RSF performed slightly worse (C-index, training: 0.75 in men, 0.79 in women; validation: 0.75 in men, 0.79 in women) but did not show performance drop during validation. LASSO stability selection performed similar to RSF. Age, self-reported lifetime treatments and medications, cystatin-C, waist circumference and FEV1-scores were identified as strong risk factors in all models for both sexes. Reduced albumin levels and elevated HbA1c were more strongly associated with high risk in men, while elevated systolic BP showed higher importance in women. Traditional Cox models observed CRP as important only in men, while the ML models identified CRP as important for both sexes. Neutrophil count was considered a strong risk factor in both sexes in the traditional Cox models, yet it was not among the most important predictors in both ML models. Presence of other heart disease (which included a.o. pericardial disease, valve disorders and arrhythmias) was an important predictor variable only in the ML models. Conclusion ML models showed similar performance to Cox PH models for HF prediction. Despite this, differences in predictor importance were identified between models. Sex-specific risk predictors were found, and FEV1 score, which is not commonly included in existing models, was identified as an important risk factor. These results suggest that ML models may reveal additional insights that would otherwise remain unnoticed.

A novel electrocardiogram-based model for prediction of dementia: the Atherosclerosis Risk in Communities (ARIC) study

Abstract Background Current dementia prediction models, such as the UK Biobank Dementia Risk Score (UKBDRS, 12 variables), are limited by having variables that are difficult to obtain, subjectively measured, and self-reported. Although individual ECG parameters are linked to higher dementia risk, no study has combined multiple ECG parameters to create an ECG model for dementia prediction. Purpose Using data from the ARIC study, a prospective community-based cohort of Black and White participants in the USA that started in 1987-89, we created an ECG model (based on 7 pre-defined ECG parameters linked to dementia) and compared its model discrimination with the UKBDRS. Methods Participants without prevalent dementia from 2 baselines were included: Visit 4 (V4, 1996-98) and V5 (2011-13). Seven ECG parameters were examined: heart rate, abnormal P-wave terminal force in lead V1 (PTFV1), prolonged P wave duration, abnormal P wave axis, advanced interatrial block, left ventricular hypertrophy (LVH, Cornell voltage criteria), and corrected QT interval (QTc). Dementia cases were identified through 2019 using in-person and phone cognitive assessments, hospitalisation codes, and death certificates. Using a backward selection method (keeping variables with P<0.10 and age, since it is an important dementia risk factor), parsimonious Cox regression models were constructed for prediction of dementia over the longest available follow-up (V4: 23 years, V5: 8 years). Model discrimination of the ECG models and UKBDRS variables was evaluated by the Harrell’s C-statistic (95% CI). Results At V4, 11,520 participants (mean age, 62.8 ± 5.7 years; 56% women; 22% Black) were included; 2,427 developed dementia. At V5, 5,667 participants (mean age, 75.4 ± 5.1 years; 59% women; 22% Black) were included; 830 developed dementia. Figures 1 and 2 show the C-statistic (95%) and area under the ROC curve for the ECG model and UKBDRS variables at V4 and V5, respectively. The C-statistic (95% CI) for both ECG models were >0.70 and comparable with the corresponding UKBDRS models. Conclusion For middle-aged and older adults, a dementia prediction model comprising age and 4 ECG variables has good discrimination for dementia that is comparable with the 12-variable UKBDRS model. Since ECG variables are easily obtainable and objectively measured, these ECG models will be easier to adopt than the UKBDRS model. Further work to validate this model in an external cohort and develop a risk score will facilitate clinical translation.

A practical risk score for prediction of 6-month mortality in patients with acute heart failure: CHAM2P2

Abstract Background Admission with acute heart failure carries a poor prognosis. Multiple predictive risk scores exist in heart failure presentation, however the utility of these scores in acute heart failure setting among older adults is limited. Aim The purpose of this study was to develop a practical risk score to predict 6-month mortality in patients presenting with acute heart failure. Methods Between January-August 2020, 474 consecutive patients (mean age 80 ± 11 years) admitted with acute heart failure to a large tertiary hospital in the South-West of England were identified. Data was retrospectively collected on admission bloods, comorbidities, imaging, echocardiography and post discharge mortality. Prediction modelling with multivariate Cox regression was then used to identify clinically relevant predictors of six-month mortality. Continuous variables were dichotomised around the mean/median for simplification. Variables were assigned points proportional to hazard ratios on Cox regression. Results Mortality at six months was 34% (161/474). Cox regression identified five significant (p<0.05) predictors of mortality: C: Clinical Frailty Score>5, H: hospital stay >10 days, A: age >80yrs, M: malignancy (current or previous) history and P: NT pro-BNP>4500pg/ml. In our final model, malignancy and NT pro-BNP were assigned two points with the remainder being assigned one. In total, 304 patients had data available to generate CHAM2P2 scores. Mortality increased sequentially from 16.0% in patients with CHAM2P2 0-2, 44.4% in patients with CHAM2P2 scores of 3-5 and 94.1% in patients with CHAM2P2 scores of 6 or 7 (log rank <0.01). AUC demonstrated good discriminatory value of the CHAM2P2 score in predicting six-month mortality (AUC 0.74). Conclusion The CHAM2P2 score is a readily available and simple tool to predict 6-month mortality in patients admitted with acute heart failure.

Digitisation of electrocardiographic images enable artificial intelligence-based mortality prediction

Abstract Background Artificial intelligence-based electrocardiogram (AI-ECG) models have shown excellent performance in a vast number of diagnostic and prognostic tasks. One of such tasks, which is critical for patient risk stratification and clinical intervention, is the prediction of adverse events, including death. Although a plethora of existing AI-ECG models have shown their accuracy in mortality prediction, their development and utilization currently rely on digital ECG signals for training and application. Purpose Many hospitals around the world have a significant number of ECGs stored in image formats or printed as paper ECGs. To address this problem, we developed and tested an ECG image digitisation pipeline to extract digital traces from ECG images and compare the performance of AI-ECG models for mortality prediction, under both natively digital and digitised signal formats. Methods A secondary care dataset of over 1 million ECGs (1,163,401 ECGs; 189,539 patients) was used for this analysis, in which both a natively digital signal and a PDF image was available. The ECG digitisation pipeline comprised of coloured PDF image inputs (2200 x 1700), initially processed to extract the ECG traces and grid features. ECG signals were extracted from the traces after cropping/restoration of the images and scaled appropriately based on the detected grid (2.5-sec strips per lead, 0.5 mV per grid block). Subsequently, both digital and digitised signals were identically pre-processed in order to acquire the ECG data fed into the AI model. In particular, a previously successful convolutional neural network architecture with a discrete-time survival loss function was used for mortality prediction. The model was trained under multiple configurations, using either the digital or the digitised ECGs as training/testing datasets (Figure 1). ECGs were split by patient ID to ensure cross-subject generalization with a training/validation/testing split of 50%/10%/40% ratio, respectively. Finally, the models were evaluated by the c-index. Results A comparison of the digital and digitised ECGs on the respective 2.5-sec strips available in both formats revealed a mean absolute error of 0.014 (CI: 0.07 – 0.021) and a Pearson’s R of 0.98 (CI: 0.96 – 1.00) between the signals, demonstrating an accurate ECG digitisation (Figure 2 shows overlapping natively digital (black) and digitised (red) traces). The performance of the digitisation was also evident in the performance of the AI-ECG mortality prediction task, with the model achieving c-indices of 0.775 (CI: 0.774 – 0.776) and 0.772 (CI: 0.771 – 0.774) when trained and tested on digital and digitised signals, respectively. Conclusion We described the first digitisation-based AI-ECG model for mortality prediction. Our results demonstrate the potential for clinical settings that lack the digital ECG infrastructure, to develop and deploy AI models using ECG image formats.Figure 1Figure 2

A simplified bedside model for acute kidney injury prediction after percutaneous coronary intervention

Abstract Background Concern for acute kidney injury (AKI) after percutaneous coronary intervention (PCI) is a major determinant in the decision to pursue cardiac catheterization if otherwise indicated. A bedside AKI risk prediction model derived from the National Cardiovascular Data Registry (NCDR), most recently updated in 2023, employs relatively complex parameters for AKI prediction. Purpose To test the efficacy of a simplified pre-procedural 5-variable bedside score for post-PCI AKI prediction based on the NCDR model. Methods We analyzed our institution’s NCDR-reported catheterization laboratory registry between Q2 2022 and Q1 2023. Inclusion criteria, variable selection, and definitions were based on the NCDR model. A bedside prediction model was validated, including 5 pre-procedural variables: anemia (hemoglobin <10 mg/dl), history of heart failure, stage 4 or 5 chronic kidney disease (glomerular filtration rate <30 mg/dl), cardiac arrest, and shock status. Results We included 840 patients, of which 593 (70.6%) were men. Mean age was 69 (±12.5). Most procedures were performed on an urgent basis (444, 52.9%). AKI developed in 107 (12.7%) patients. Compared to the non-AKI group, patients who developed AKI were older (age 74±11.5 vs 68.3 ±12.5), and more frequently had anemia (23.6% vs 7.4%), diabetes (51.4% vs 35.2%), severe frailty (45.8% vs 23.7%), heart failure (62.6% vs 29.7%), cardiac arrest (13.1% vs 2.9%), CKD (30.8% vs 6%), shock (12.1% vs 1.5%), and mechanical circulatory support (7.5% vs 2.9%), p<0.001. Utilizing integer scores, the simplified bedside model employing 5 pre-procedural variables demonstrated discrimination similar to the bedside NCDR model, with a C-statistic of 0.77 (0.60-0.88) and excellent calibration (intercept of 0.31 and slope of 1.03) (Figure), indicating the robust performance of the model (compared to bedside NCDR model with a C-statistic 0.793, intercept of -0.197, and slope of 0.957). Conclusion A simplified NCDR-based bedside score using 5 pre-procedural variables is an effective tool for predicting AKI as a complication of PCI, with comparable performance to the 2023 updated bedside NCDR AKI risk model.

Large language models to facilitate pregnancy prediction after invitro fertilization.

We evaluated the efficacy of large language models (LLMs), specifically, generative pre-trained transformer-4 (GPT-4), in predicting pregnancy following invitro fertilization (IVF) treatment and compared its accuracy with results from an original published study. Our findings revealed that GPT-4 can autonomously develop and refine advanced machine learning models for pregnancy prediction with minimal human intervention. The prediction accuracy was 0.79, and the area under the receiver operating characteristic curve (AUROC) was 0.89, exceeding or being at least equivalent to the metrics reported in the original study, that is, 0.78 for accuracy and 0.87 for AUROC. The results suggest that LLMs can facilitate data processing, optimize machine learning models in predicting IVF success rates, and provide data interpretation methods. This capacity can help bridge the knowledge gap between data scientists and medical personnel to solve the most pressing clinical challenges. However, more experiments on diverse and larger datasets are needed to validate and promote broader applications of LLMs in assisted reproduction.

The CNN-LSTM-attention Model for short term prediction of the Polar Motion

Abstract The accuracy of polar motion prediction significantly impacts the fields of coordinate frame transformation, satellite orbit determination, and deep space exploration. The present study develops two short term forecasting models based on the EOP 14C04 series. One hybrid approach incorporates convolutional neural networks (CNN) and long short-term memory networks (LSTM), augmented with an attention mechanism; whereas another baseline model comprises CNN and LSTM. The first model, in contrast to the second model, incorporates an attention mechanism module for a more comprehensive integration of temporal information at each time step. In the initial short-term forecasting experiment, we conducted 360 repeated predictions, and the findings revealed that the parameters suitable for PMX forecasting may not necessarily be applicable to PMY forecasting. In the second experiment, the two models generated a total of 500 forecasts, each encompassing short-term predictions ranging from 1 to 30 days. The experimental results demonstrate that the first model exhibits mean absolute error (MAE) range of 0 ~ 7.72 mas for PMX and 0 ~ 4.73 mas for PMY, while the second model shows MAE range of 0 ~ 7.88 mas for PMX and 0 ~ 4.78 mas for PMY. After two exploratory experiments, we discovered the following results: the first model exhibits marginally superior predictive accuracy compared to the second model. Furthermore, this study substantiates the robustness of both models in short-term prediction and affirms the significance of assigning distinct weights to past temporal intervals in forecasting, thereby offering a novel perspective for polar motion prediction research.

Accurate detection of dilated cardiomyopathy onset through machine learning predictions from ECG data

Abstract Background Dilated cardiomyopathy (DCM) is a primary heart muscle disease characterised by left-ventricular dilatation and reduced contractility, often associated with arrhythmia and conduction disease. DCM often has a genetic component, putting family members of diagnosed patients at risk for the disease. Stratification tools to provide personalized screening advice are currently lacking. Artificial intelligence (AI) based analysis of the electrocardiogram (ECG) offers a potential solution to this challenge. Purpose Our study aims to develop and validate a novel AI-ECG model capable of prediction and early detection of DCM. Methods An AI-ECG model was developed on a United States (US) based secondary care centre dataset including 337,463 ECGs from 50,932 patients within 4 months of diagnostic echocardiography and 265,576 ECGs from 51,086 patients prior to the first diagnostic echo. The data was split 50/10/40% for training, validation, and testing, with each subset containing 9-12% DCM positive labels. The model architecture was based on a convolutional neural network with residual blocks with a modified final layer for a discrete-time survival model. The AI-ECG score was linked to changes in the median ECG within the test set to provide explainability. External validation was performed on 68,885 diagnostic ECGs (0.08% DCM positive) from the UK biobank (UKB) and a Chinese general hospital dataset with 317,854 diagnostic ECGs (2.2% DCM positive) and 76,936 predictive ECGs (1.7% DCM positive). We performed common and rare variant association studies for the AI ECG scores on quality-controlled UKB data. Results The AI-ECG score for predicted future DCM was associated with diverse ECG features (Fig. 1). In the US cohort test set, the model achieved high diagnostic accuracy with an AUC of 0.9 (95% CI 0.89-0.90) overall, and 0.89 (0.85-0.92) in a subset excluding ischemic or valvular heart disease. External datasets showed AUCs of 0.86 (0.81-0.91) in the UKB and 0.97 (96-0.97) in the Chinese dataset. The predictive performance of the AI-ECG score, adjusted for age and sex, closely approximated echocardiographic traits (LVEF and LVEDD, Fig. 2) with a concordance index (CI) of 0.79 (0.76-0.83) and 0.81 (0.77-0.85), respectively, at 10 years of follow up (FU). Combining AI-ECG scores with echocardiographic traits enhanced the CI to 0.84 (0.8-0.87). Similar findings were observed in the Chinese dataset, with a CI of 0.87 (0.85-0.89) for the AI-ECG score and 0.9 (0.88-0.92) for echocardiographic traits. Genetic analyses in UKB identified associations between the AI-ECG score and rare TTN variants as well as 25 common variant loci. Conclusion Our explainable AI ECG model for future DCM prediction demonstrates robust performance across datasets, providing comparable predictive accuracy to echocardiographic traits up to 10 years pre-diagnosis. Genetic associations support the model's validity and provide new tools for DCM genetic discovery.

Prediction model for survival of younger patients with breast cancer using the breast cancer public staging database

Breast cancer (BC) is a major contributor to female mortality worldwide, particularly in young women with aggressive tumors. Despite the need for accurate prognosis in this demographic, existing studies primarily focus on broader age groups, often using the SEER database, which has limitations in variable selection. This study aimed to develop an ML-based model to predict survival outcomes in young BC patients using the BC public staging database. A total of 3,401 patients with BC were included in the study. Patients were categorized as younger (n = 1574) and older (n = 1827). We applied several survival models—Random Survival Forest, Gradient Boosting Survival, Extra Survival Trees (EST), and penalized Cox models (Lasso and ElasticNet)—to compare mortality characteristics. The EST model outperformed others in predicting mortality for both age groups. Older patients exhibited a higher prevalence of comorbidities compared to younger patients. Tumor stage was the primary variable used to train the model for mortality prediction in both groups. COPD was a significant variable only in younger patients with BC. Other variables exhibited varying degrees of consistency in each group. These findings can help identify high-risk young female patients with BC who require aggressive treatment by predicting the risk of mortality.

Prediction of microvascular invasion in hepatocellular carcinoma with conventional ultrasound, Sonazoid-enhanced ultrasound, and biochemical indicator: a multicenter study

PurposeTo develop and validate a preoperative prediction model based on multimodal ultrasound and biochemical indicator for identifying microvascular invasion (MVI) in patients with a single hepatocellular carcinoma (HCC) ≤ 5 cm.MethodsFrom May 2022 to November 2023, a total of 318 patients with pathologically confirmed single HCC ≤ 5 cm from three institutions were enrolled. All of them underwent preoperative biochemical, conventional ultrasound (US), and contrast-enhanced ultrasound (CEUS) (Sonazoid, 0.6 mL, bolus injection) examinations. Univariate and multivariate logistic regression analyses on clinical information, biochemical indicator, and US imaging features were performed in the training set to seek independent predictors for MVI-positive. The models were constructed and evaluated using the area under the receiver operating characteristic curve (AUC), calibration curve, and decision curve analysis in both validation and test sets. Subgroup analyses in patients with different liver background and tumor sizes were conducted to further investigate the model’s performance.ResultsLogistic regression analyses showed that obscure tumor boundary in B-mode US, intra-tumoral artery in pulsed-wave Doppler US, complete Kupffer-phase agent clearance in Sonazoid-CEUS, and biomedical indicator PIVKA-II were independently correlated with MVI-positive. The combined model comprising all predictors showed the highest AUC, which were 0.937 and 0.893 in the validation and test sets. Good calibration and prominent net benefit were achieved in both sets. No significant difference was found in subgroup analyses.ConclusionsThe combination of biochemical indicator, conventional US, and Sonazoid-CEUS features could help preoperative MVI prediction in patients with a single HCC ≤ 5 cm.Critical relevance statementInvestigation of imaging features in conventional US, Sonazoid-CEUS, and biochemical indicators showed a significant relation with MVI-positivity in patients with a single HCC ≤ 5 cm, allowing the construction of a model for preoperative prediction of MVI status to help treatment decision making.Key PointsMVI status is important for patients with a single HCC ≤ 5 cm.The model based on conventional US, Sonazoid-CEUS and PIVKA-II performs best for MVI prediction.The combined model has potential for preoperative prediction of MVI status.Graphical

Incidence, predictors and outcomes of non-home discharge following transcatheter aortic valve implantation: a multicentre Australian experience. The NHD TAVI study

Abstract Background Patients undergoing transcatheter aortic valve implantation (TAVI) commonly experience non-home discharge (NHD), a phenomenon associated with increased health-care expenditure and possibly poorer outcomes. Despite its relevance to patients, post-TAVI NHD remains poorly explored in the literature. Purpose To characterise the incidence and predictors of NHD following TAVI, understand its impact on quality of life, and identify the proportion of patients that require long-term residential care following initial NHD. Methods Retrospective analysis of prospectively collected data of 2229 patients undergoing TAVI from 2010–2023 included in an Australian TAVI Registry. Results Median age was 82 (IQR: 78-86) years and 41% were female. 257 (11.5%) patients were not discharged home following TAVI, with incidence falling over time (R2=0.636, p<0.001). Multivariable logistic regression modelling for NHD prediction was developed with excellent calibration and discrimination (C-Statistic 0.835). Independent predictors of NHD were post-procedural stroke (adjusted odds ratio [aOR] 11.1), procedure at a private hospital (aOR 3.0), living alone (aOR 2.4), vascular access site complications (aOR 2.1), frailty (aOR 1.9), age>80 (aOR 1.8), hypoalbuminemia (aOR 1.8), NYHA III-IV (aOR 1.7) and hospital length of stay (aOR 1.1) (all p<0.05). NHD was not associated with mortality at 30-days and less than one percent of all patients required longer-term residential care. Conclusion While common following TAVI, NHD does not predict short-term mortality and most patients successfully return home within 30-days. When used appropriately, NHD may serve as a brief and effective method of optimising functional status without compromising long-term independence.Predictors of Non Home Discharge

Time-varying cox model for heart failure prediction in phospholamban p.(Arg14del)-positive individuals

Abstract Background Phospholamban (PLN) p.(Arg14del)-positive individuals are at risk for developing severe heart failure (HF), yet a comprehensive risk model for this outcome is missing. Our PLN registry contains longitudinal data which enables us to capture dynamic changes in covariables over time, while prior prediction models rely on baseline clinical information which may change over time. Using this longitudinal clinical information will enhance the accuracy of HF prediction and may aid patients selection for future (gene)therapy. Purpose The aim of this study is to develop a dynamic HF risk prediction model for PLN p.(Arg14del)-positive individuals. Method Data were collected of 594 PLN p.(Arg14del)-positive individuals who had no documented history of HF or myocardial infarction at time of the first echocardiography. We incorporated two time-dependent covariates in our time-varying Cox regression model: longitudinal individual predicted left ventricular ejection fraction (LVEF) and longitudinal individual predicted relative risk for major ventricular arrhythmia (VA). Predicted LVEF values were derived using observed LVEF values in a linear mixed-effect model, while the relative risk for major VA was predicted using a Cox regression model with major VA as the outcome. These time-dependent covariates were included in the time-varying Cox regression along with baseline covariables age at inclusion and previous major VA. The results were validated using 5-fold cross-validation. Results Over a median follow-up period of 3.6 years (interquartile range 1.4-6.8), 77 (13%) individuals developed HF, defined as a composite endpoint of HF hospitalisation, implantation of left ventricular assist device, heart transplantation and HF-related mortality. Our time-varying Cox regression model demonstrated a robust 5-year mean C-statistic of 0.904 (95% CI 0.901-0.908). The hazard ratio for time-dependent LVEF predictions was 0.89 per % LVEF increase (95% CI 0.87-0.91, p <0.001) and for time-dependent relative major VA risk predictions 3.59 (95% CI 1.49-8.69, p = 0.005). Regarding baseline covariates, age at inclusion had a hazard ratio of 1.01 (95% CI 0.99-1.03, p = 0.47) and previous major VA 0.48 (95% CI 0.21-1.1, p = 0.08). Combining the hazard ratios of time-dependent predicted relative risk for major VA and major VA in history results in a hazard ratio of 1.73. Conclusion Our study introduces a novel time-varying Cox regression model for the individual prediction of heart failure events in PLN p.(Arg14del)-positive individuals which may aid patient selection for future (gene)therapy. Results demonstrate robust predictive performance (C-statistic of 0.904, 95% CI 0.901-0.908), highlighting the significance of incorporating longitudinal data for enhanced prediction accuracy.

Predicting 1-year atrial fibrillation risk from a patch-based ambulatory ECG monitoring without atrial fibrillation

Abstract Introduction Early identification of atrial fibrillation (AF) can enable interventions that reduce cardiovascular events. However, AF remains undiagnosed in about 10-20% of affected individuals, and paroxysmal AF often remains undetected despite the use of in-clinic ECGs, consumer smartwatch ECGs, or ambulatory ECG monitoring. We therefore investigated whether ambulatory ECG patch-based monitoring without AF can predict AF in the subsequent 1-year post monitoring. Purpose This study aims to validate a machine-learning model for AF prediction within 1-year post-wear, based on the observation of non-AF single-lead ECG recordings. Methods We considered a clinically monitored cohort of 67,296 individuals wearing multiple single-lead ECG multi-day ambulatory monitoring patches as part of real-world care. For all selected individuals, the first patch recorded no occurrences of AF and was worn for a minimum of 6.5 days. We assess the risk of future AF based on the raw ECG, heartrate variability, frequency of ectopic beats, age and gender (ECG model), which is then compared with the risk assessment by a model based on age and gender (Baseline model). We then observe AF incidence on subsequent patches worn within one year from the first patch. We evaluated the model's accuracy by observing the number of AF cases identified (or missed) among the individuals with highest risk according to the model. Results In the clinically monitored cohort, a total of 3463 individuals experienced AF within the next year (incidence 5.1%) after a first AF-free monitoring period of a minimum 6.5 days. The receiver operating characteristic (ROC) curve shows an area under the curve (AUC) for the ECG model of 0.75 (95% CI: [0.74, 0.76]), and for the Baseline model of 0.66 (CI: [0.65, 0.67]). When selecting the top 20% (13,459) participants with the highest risk and verifying who developed AF in the following 1-year, we obtain positive prediction value (PPV) or precision of 12% and sensitivity (recall) of 46% for the ECG model (which correctly identifies 1,594 individuals with incident AF), and of 8% and 29%, respectively, for the Baseline model (p-value<0.001). (Figure) Conclusion A machine learning model using as input an AF-free single-lead ECG can quantify the risk of observing AF in a population including individuals with low burdens of paroxysmal or new onset of AF within 1 year. Further research and validation are needed to enhance the interpretability and clinical applicability of the model, while prediction may be further improved with clinical AF diagnoses ascertained from other data sources. The findings of this study have important implications for the early detection and management of AF, potentially enabling earlier intervention and improved outcomes in individuals with undiagnosed AF.

Clinical validated risk prediction model for development of heart failure after contemporary breast cancer treatment

Abstract Background Patients receiving breast cancer (BC) treatment are at risk for heart failure and cardiomyopathy (HF/CM). European Society of Cardiology (ESC) cardio-oncology guidelines recommend baseline cardiovascular (CV) assessment, however, there is limited data about cardiotoxicity risk specific to patients with BC. Our aim was to develop and validate model for prediction of HF/CM in women receiving contemporary treatment for invasive early-stage BC. Methods We assembled a cohort women diagnosed with Stage I-III BC from 2008-2020 and followed through 2021 in a large health care plan. Women age >80, with history of HF/CM, and women who received both anthracyclines and trastuzumab were excluded, leaving N=26,044 for analysis. Study cohort was randomly split into derivation (60%) and validation (40%) datasets. Elastic-net penalized Cox proportional hazard model was created using the derivation dataset to select risk factors and to calculate the panelized coefficient which defined the HF/CM risk score. Risk score system was assessed by Hosmer-Lemeshow goodness-of-fit test and area under the ROC curve (AUC). Event rates of <5% constituted low risk; 5-14% moderate; and >15% high risk category. Incident HF/CM events were examined using Kaplan-Meier approach in derivation and validation datasets. Results HF/CM risk score had high discrimination ability and sufficiently good model of fit regarding HF/CM events (P value=0.163 and 0.667 from Hosmer–Lemeshow test;AUC 0.751 and 0.764, in derivation and validation sets, respectively). There were 10% of women in the high risk; 30% in moderate; and 60% in low-risk group. Variables associated with the higher HF/CM risk score included age, receipt of anthracycline-based chemotherapy or anti-HER2 therapy, presence of hypertension, diabetes, or history of other CV conditions at baseline. The score pattern showed major effect of age, with anthracycline and anti-HER2 therapy increasing the risk across all age groups (Figure 1). Kaplan-Meier analysis of incident HF/CM over 14 years after BC diagnosis showed significantly lower number of HF/CM events in low risk compared to moderate and high-risk score groups (p<0.0001, Figure 2). Conclusion Our robust risk prediction model for HF/CM among women with early-stage BC provides a new tool to identify women at higher CV risk and inform cardiac function monitoring during and after BC treatment. Our findings may guide further validation studies and build evidence for the cardiotoxicity risk stratification in patients with BC.Risk Score for HF/CM by Age/ChemoTxK-M curves of freedom from HF/CM

Predicting hospital readmission in multimorbid patients with the use of AI: A systematic review

Abstract Background Multimorbid patients are at higher risk of hospital readmission due to the complex nature of their conditions. Identifying those who may be at particularly high risk would allow us to intervene early and potentially delay or prevent such readmissions occurring, thus reducing healthcare costs. We conducted a systematic review investigating the use of machine learning models in predicting 30-days unplanned hospital readmission of multimorbid patients. Methods This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Guidelines, and was registered with PROSPERO (CRD42022373937). We searched MEDLINE, Embase, Web of Science and Cumulative Index to Nursing and Allied Health Literature (CINAHL). Included studies developed an AI model for prediction of 30-days hospital readmission for adult patients with two or more health conditions. The CHARM and PROBAST checklists for data extraction and bias assessment were used. The quality of included studies was assessed with the CASP checklist. Results Eighteen papers were eligible for inclusion. A total of 669 predictors were reported with an average of 37 used per model. Predictors were classified as modifiable and non-modifiable with the most common modifiable predictors being hospital length of stay, hypertension, anaemia and obesity. Average sensitivity and specificity of the models was reported to be 72% in 13 studies and 79.2 in 11 studies, respectively. Area under the curve (AUC) was reported in 13 studies, five of which were considered to have good discrimination power (AUC>0.8). Conclusions Machine learning models are capable of accurately predicting 30-days hospital readmission of multimorbid patients. Identifying modifiable predictors with highest weight allows for better planning and resource allocation to potentially reduce the risk of 30-days readmissions. An important area for future work would be the implementation of these high performance models in practice. Key messages • Hospital readmissions are preventable and identifying those at higher risk allows us to intervene early. • Machine learning models are capable of predicting 30-days readmission of multimorbid patients.