Proxy endpoints - bridging clinical trials and real world data.

OBJECTIVE: Disease severity scores, or endpoints, are routinely measured during Randomized Controlled Trials (RCTs) to closely monitor the effect of treatment. In real-world clinical practice, although a larger set of patients is observed, the specific RCT endpoints are often not captured, which makes it hard to utilize real-world data (RWD) to evaluate drug efficacy in larger populations. METHODS: To overcome this challenge, we developed an ensemble technique which learns proxy models of disease endpoints in RWD. Using a multi-stage learning framework applied to RCT data, we first identify features considered significant drivers of disease available within RWD. To create endpoint proxy models, we use Explainable Boosting Machines (EBMs) which allow for both end-user interpretability and modeling of non-linear relationships. RESULTS: We demonstrate our approach on two diseases, rheumatoid arthritis (RA) and atopic dermatitis (AD). As we show, our combined feature selection and prediction method achieves good results for both disease areas, improving upon prior methods proposed for predictive disease severity scoring. CONCLUSION: Having disease severity over time for a patient is important to further disease understanding and management. Our results open the door to more use cases in the space of RA and AD such as treatment effect estimates or prognostic scoring on RWD. Our framework may be extended beyond RA and AD to other diseases where the severity score is not well measured in electronic health records.

Predictive value of changes in basal myocardial 18F-fluorodeoxyglucose uptake for cardiotoxicity in locally advanced esophageal cancer patients receiving definitive radiotherapy.

PURPOSE: To investigate the predictive value of changes in segmental myocardial 18F- fluorodeoxyglucose (FDG) uptake for major adverse cardiac events (MACEs) in patients with locally advanced esophageal cancer undergoing definitive radiation therapy (RT). MATERIALS AND METHODS: Between August 2012 and January 2019, 482 patients with stage II-III esophageal cancer from two institutions were enrolled and divided into the training (n = 285) and external validation (n = 197) cohorts. All patients underwent 18F-FDG PET within 1 week before treatment and within 3 months of treatment. Myocardial delineation was performed using the Carimas software based on the AHA 17-segment model and was automatically divided into basal, middle, and apical regions. The main endpoint was the occurrence of MACEs, including unstable angina, myocardial infarction, coronary revascularization, hospitalization for heart failure or urgent visits, and cardiac death. Analyses included competing risk and Cox regression. Model performance was assessed using the area under the receiver operating characteristic curve (AUC) and Brier score. RESULTS: Thirty-four patients (11.9%) developed MACEs at a median follow-up of 78 months. The basal region (median: 19.44 Gy) of the myocardium received the highest radiation dose, followed by the middle (median: 13.02 Gy) and apical regions (median: 9.32 Gy). Multivariate analysis showed that the change ratio in pre- and post-treatment basal myocardial SUVmean (basal ∆SUVRmean) remained significant after adjusting for age, pre-existing cardiac disease, and dosimetric parameters. The AUCs and Brier scores demonstrated favorable predictive accuracies of models integrating variables with significant differences in the multivariate analysis when predicting MACEs in the training and validation cohorts. CONCLUSION: Basal ∆SUVRmean was an independent predictor of MACEs in patients with locally advanced esophageal cancer receiving definitive RT. Changes in basal myocardial FDG uptake are promising biomarkers for predicting radiation-induced cardiotoxicity.

Enhancing Patient Selection in Sepsis Clinical Trials Design Through an AI Enrichment Strategy: Algorithm Development and Validation.

BACKGROUND: Sepsis is a heterogeneous syndrome, and enrollment of more homogeneous patients is essential to improve the efficiency of clinical trials. Artificial intelligence (AI) has facilitated the identification of homogeneous subgroups, but how to estimate the uncertainty of the model outputs when applying AI to clinical decision-making remains unknown. OBJECTIVE: We aimed to design an AI-based model for purposeful patient enrollment, ensuring that a patient with sepsis recruited into a trial would still be persistently ill by the time the proposed therapy could impact patient outcome. We also expected that the model could provide interpretable factors and estimate the uncertainty of the model outputs at a customized confidence level. METHODS: In this retrospective study, 9135 patients with sepsis requiring vasopressor treatment within 24 hours after sepsis onset were enrolled from Beth Israel Deaconess Medical Center. This cohort was used for model development, and 10-fold cross-validation with 50 repeats was used for internal validation. In total, 3743 patients with sepsis from the eICU Collaborative Research Database were used as the external validation cohort. All included patients with sepsis were stratified based on disease progression trajectories: rapid death, recovery, and persistent ill. A total of 148 variables were selected for predicting the 3 trajectories. Four machine learning algorithms with 3 different setups were used. We estimated the uncertainty of the model outputs using conformal prediction (CP). The Shapley Additive Explanations method was used to explain the model. RESULTS: The multiclass gradient boosting machine was identified as the best-performing model with good discrimination and calibration performance in both validation cohorts. The mean area under the receiver operating characteristic curve with SD was 0.906 (0.018) for rapid death, 0.843 (0.008) for recovery, and 0.807 (0.010) for persistent ill in the internal validation cohort. In the external validation cohort, the mean area under the receiver operating characteristic curve (SD) was 0.878 (0.003) for rapid death, 0.764 (0.008) for recovery, and 0.696 (0.007) for persistent ill. The maximum norepinephrine equivalence, total urine output, Acute Physiology Score III, mean systolic blood pressure, and the coefficient of variation of oxygen saturation contributed the most. Compared to the model without CP, using the model with CP at a mixed confidence approach reduced overall prediction errors by 27.6% (n=62) and 30.7% (n=412) in the internal and external validation cohorts, respectively, as well as enabled the identification of more potentially persistent ill patients. CONCLUSIONS: The implementation of our model has the potential to reduce heterogeneity and enroll more homogeneous patients in sepsis clinical trials. The use of CP for estimating the uncertainty of the model outputs allows for a more comprehensive understanding of the model's reliability and assists in making informed decisions based on the predicted outcomes.

Development and application of a nomogram model for predicting the risk of central precocious puberty in obese girls.

Objective: The purpose of this study is to develop and assess a nomogram risk prediction model for central precocious puberty (CPP) in obese girls. Methods: We selected 154 cases of obese girls and 765 cases of non-obese girls with precocious puberty (PP) who underwent the gonadotropin-releasing hormone stimulation test at the Jiangxi Provincial Children's Hospital. Univariate analysis and multivariate analysis were conducted to identify predictors of progression to CPP in girls with PP. A predictive model was developed and its predictive ability was preliminarily evaluated. The nomogram was used to represent the risk prediction model for CPP in girls with obesity. The model was validated internally using the Bootstrap method, and its efficacy was assessed using calibration curves and clinical decision analysis curves. Results: In obese girls with PP, basal luteinizing hormone (LH) and follicular stimulating hormone (FSH) levels, as well as uterine volume, were identified as independent risk factors for progression to CPP. In non-obese girls, the basal LH level, bone age, and uterine volume were identified as independent risk factors for progression to CPP. With an AUC of 0.896, the risk prediction model for obese girls, was found to be superior to that for non-obese girls, which had an AUC of 0.810. The model displayed strong predictive accuracy. Additionally, a nomogram was used to illustrate the CPP risk prediction model for obese girls. This model performs well in internal validation and is well calibrated, providing a substantial net benefit for clinical use. Conclusion: A medical nomogram model of CPP risk in obese girls comprised of basal LH value, basal FSH value, and uterine volume, which can be used to identify those at high risk for progression of CPP in obese girls and develop individualized prevention programs.

Connectome-based predictive modeling of brain pathology and cognition in Autosomal Dominant Alzheimer's Disease.

INTRODUCTION: Autosomal Dominant Alzheimer's Disease (ADAD) through genetic mutations can result in near complete expression of the disease. Tracking AD pathology development in an ADAD cohort of Presenilin-1 (PSEN1) E280A carriers' mutation has allowed us to observe incipient tau tangles accumulation as early as 6 years prior to symptom onset. METHODS: Resting-state functional Magnetic Resonance Imaging (fMRI) and Positron-Emission Tomography (PET) scans were acquired in a group of PSEN1 carriers (n=32) and non-carrier family members (n=35). We applied Connectome-based Predictive Modeling (CPM) to examine the relationship between the participant's functional connectome and their respective tau/amyloid-ß levels and cognitive scores (word list recall). RESULTS: CPM models strongly predicted tau concentrations and cognitive scores within the carrier group. The connectivity patterns between the temporal cortex, default mode network, and other memory networks were the most informative of tau burden. DISCUSSION: These results indicate that resting-state fMRI methods can complement PET methods in early detection and monitoring of disease progression in ADAD.

Association of Cardiac Troponin T With Coronary Atherosclerosis in Asymptomatic Primary Prevention People With HIV.

Background: Coronary plaque is common among people with HIV (PWH) with low-to-moderate traditional atherosclerotic cardiovascular disease (ASCVD) risk. Objectives: The purpose of this study was to determine the association of high-sensitivity cardiac troponin T (hs-cTnT) levels with coronary plaque characteristics and evaluate if hs-cTnT improves identification of these features beyond traditional ASCVD risk factors among PWH. Methods: Among PWH receiving stable antiretroviral therapy with low-to-moderate ASCVD risk and no known history of ASCVD, hs-cTnT levels and measures of plaque by coronary computed tomography angiography were assessed. Primary outcomes included the association of hs-cTnT level with the presence of any plaque, vulnerable plaque, coronary artery calcium (CAC) score, and Leaman score. Assessment of model discrimination of hs-cTnT for plaque characteristics was also performed. Results: The cohort included 708 U.S. participants with a mean age of 51 ± 6 years, 119 (17%) females, a median ASCVD risk score of 4.4% (Q1-Q3: 2.5%-6.6%), and a median hs-cTnT level of 6.7 ng/L (detectable level ≥6 ng/L in 61%). Any plaque was present in 341 (48%), vulnerable plaque in 155 (22%), CAC>100 in 68 (10%), and a Leaman score >5 in 105 (15%). After adjustment for ASCVD risk score, participants with hs-cTnT >9.6 ng/L (highest category) versus an undetectable level (<6 ng/L) had a greater relative risk for any plaque (1.37, 95% CI: 1.12-1.67), vulnerable plaque (1.47, 95% CI: 1.16-1.87), CAC>100 (2.58, 95% CI: 1.37-4.83), and Leaman score >5 (2.13, 95% CI: 1.32-3.46). The addition of hs-cTnT level modestly improved the discrimination of ASCVD risk score to identify critical plaque features. Conclusions: In PWH without known ASCVD, hs-cTnT levels were strongly associated with and improved prediction of subclinical coronary plaque. (Evaluating the Use of Pitavastatin to Reduce the Risk of Cardiovascular Disease in HIV-Infected Adults [REPRIEVE]; NCT02344290).

Data-driven approaches to study the spectral properties of chemical structures.

The molecular energy, which is the sum of all eigenvalues, is crucial in determining the total π-electron energy of conjugated hydrocarbon molecules. We used machine learning techniques to calculate the energy, inertia, nullity, signature, and Estrada index of molecular graphs for bismuth tri-iodide and benzene rings embedded in P-type surfaces within 2D networks. We applied MATLAB to extract the actual eigenvalues from the data and developed general equations for these molecular properties. We then used these equations to estimate the values and compared them to the actual values through graphical analysis. Our results demonstrate the potential of data-driven techniques in predicting molecular properties and enhancing our understanding of spectral theory.

Spatiotemporal Forecasting of Opioid-Related Fatal Overdoses: Towards Best Practices for Modeling and Evaluation.

To inform public health interventions, researchers have developed models to forecast opioid-related overdose mortality. These efforts often have limited overlap in the models and datasets employed, presenting challenges to assessing progress in this field. Furthermore, common error-based performance metrics, such as root mean squared error (RMSE), cannot directly assess a key modeling purpose: the identification of priority areas for interventions. We recommend a new intervention-aware performance metric, Percentage of Best Possible Reach (%BPR). We compare metrics for many published models across two distinct geographic settings, Cook County, Illinois and Massachusetts, assuming the budget to intervene in 100 census tracts out of 1000s in each setting. The top-performing models based on RMSE recommend areas that do not always reach the most possible overdose events. In Massachusetts, the top models preferred by %BPR could have reached 18 additional fatal overdoses per year in 2020-2021 compared to models favored by RMSE. In Cook County, the different metrics select similar top-performing models, yet other models with similar RMSE can have significant variation in %BPR. We further find that simple models often perform as well as recently published ones. We release open code and data for others to build upon.

A triple hormone receptor ER, AR, and VDR signature is a robust prognosis predictor in breast cancer.

BACKGROUND: Despite evidence indicating the dominance of cell-of-origin signatures in molecular tumor patterns, translating these genome-wide patterns into actionable insights has been challenging. This study introduces breast cancer cell-of-origin signatures that offer significant prognostic value across all breast cancer subtypes and various clinical cohorts, compared to previously developed genomic signatures. METHODS: We previously reported that triple hormone receptor (THR) co-expression patterns of androgen (AR), estrogen (ER), and vitamin D (VDR) receptors are maintained at the protein level in human breast cancers. Here, we developed corresponding mRNA signatures (THR-50 and THR-70) based on these patterns to categorize breast tumors by their THR expression levels. The THR mRNA signatures were evaluated across 56 breast cancer datasets (5040 patients) using Kaplan-Meier survival analysis, Cox proportional hazard regression, and unsupervised clustering. RESULTS: The THR signatures effectively predict both overall and progression-free survival across all evaluated datasets, independent of subtype, grade, or treatment status, suggesting improvement over existing prognostic signatures. Furthermore, they delineate three distinct ER-positive breast cancer subtypes with significant survival in differences-expanding on the conventional two subtypes. Additionally, coupling THR-70 with an immune signature identifies a predominantly ER-negative breast cancer subgroup with a highly favorable prognosis, comparable to ER-positive cases, as well as an ER-negative subgroup with notably poor outcome, characterized by a 15-fold shorter survival. CONCLUSIONS: The THR cell-of-origin signature introduces a novel dimension to breast cancer biology, potentially serving as a robust foundation for integrating additional prognostic biomarkers. These signatures offer utility as a prognostic index for stratifying existing breast cancer subtypes and for de novo classification of breast cancer cases. Moreover, THR signatures may also hold promise in predicting hormone treatment responses targeting AR and/or VDR.

Refinement of machine learning arterial waveform models for predicting blood loss in canines.

Introduction: Hemorrhage remains a leading cause of death in civilian and military trauma. Hemorrhages also extend to military working dogs, who can experience injuries similar to those of the humans they work alongside. Unfortunately, current physiological monitoring is often inadequate for early detection of hemorrhage. Here, we evaluate if features extracted from the arterial waveform can allow for early hemorrhage prediction and improved intervention in canines. Methods: In this effort, we extracted more than 1,900 features from an arterial waveform in canine hemorrhage datasets prior to hemorrhage, during hemorrhage, and during a shock hold period. Different features were used as input to decision tree machine learning (ML) model architectures to track three model predictors-total blood loss volume, estimated percent blood loss, and area under the time versus hemorrhaged blood volume curve. Results: ML models were successfully developed for total and estimated percent blood loss, with the total blood loss having a higher correlation coefficient. The area predictors were unsuccessful at being directly predicted by decision tree ML models but could be calculated indirectly from the ML prediction models for blood loss. Overall, the area under the hemorrhage curve had the highest sensitivity for detecting hemorrhage at approximately 4 min after hemorrhage onset, compared to more than 45 min before detection based on mean arterial pressure. Conclusion: ML methods successfully tracked hemorrhage and provided earlier prediction in canines, potentially improving hemorrhage detection and objectifying triage for veterinary medicine. Further, its use can potentially be extended to human use with proper training datasets.

Predicting Quality of Life in Parkinson's Disease: A Machine Learning Approach Employing Common Clinical Variables.

Background: Parkinson's Disease significantly impacts health-related quality of life, with the Parkinson's Disease Questionnaire-39 extensively used for its assessment. However, predicting such outcomes remains a challenge due to the subjective nature and variability in patient experiences. This study develops a machine learning model using accessible clinical data to enable predictions of life-quality outcomes in Parkinson's Disease and utilizes explainable machine learning techniques to identify key influencing factors, offering actionable insights for clinicians. Methods: Data from the Parkinson's Real-world Impact Assessment study (PRISM), involving 861 patients across six European countries, were analyzed. After excluding incomplete data, 627 complete observations were used for the analysis. An ensemble machine learning model was developed with a 90% training and 10% validation split. Results: The model demonstrated a Mean Absolute Error of 4.82, a Root Mean Squared Error of 8.09, and an R2 of 0.75 in the training set, indicating a strong model fit. In the validation set, the model achieved a Mean Absolute Error of 11.22, a Root Mean Squared Error of 13.99, and an R2 of 0.36, showcasing moderate variation. Key predictors such as age at diagnosis, patient's country, dementia, and patient's age were identified, providing insights into the model's decision-making process. Conclusions: This study presents a robust model capable of predicting the impact of Parkinson's Disease on patients' quality of life using common clinical variables. These results demonstrate the potential of machine learning to enhance clinical decision-making and patient care, suggesting directions for future research to improve model generalizability and applicability.

Connectome-based prediction modeling of cognitive control using functional and structural connectivity.

BACKGROUND: Cognitive control involves flexibly configuring mental resources and adjusting behavior to achieve goal-directed actions. It is associated with the coordinated activity of brain networks, although it remains unclear how both structural and functional brain networks can predict cognitive control. Connectome-based predictive modeling (CPM) is a powerful tool for predicting cognitive control based on brain networks. METHODS: The study used CPM to predict cognitive control in 102 healthy adults from the UCLA Consortium for Neuropsychiatric Phenomics dataset and further compared structural and functional connectome characteristics that support cognitive control. RESULTS: Our results showed that both structural (r values 0.263-0.375) and functional (r values 0.336-0.503) connectomes can significantly predict individuals' cognitive control subcomponents. There is overlap between the functional and structural networks of all three cognitive control subcomponents, particularly in the frontoparietal (FP) and motor (Mot) networks, while each subcomponent also has its own unique weight prediction network. Overall, the functional and structural connectivity that supports different cognitive control subcomponents manifests overlapping and distinct spatial patterns. CONCLUSIONS: The structural and functional connectomes provide complementary information for predicting cognitive control ability. Integrating information from both connectomes offers a more comprehensive understanding of the neural underpinnings of cognitive control.

Stem cell-based approaches for developmental neurotoxicity testing.

Neurotoxicants are substances that can lead to adverse structural or functional effects on the nervous system. These can be chemical, biological, or physical agents that can cross the blood brain barrier to damage neurons or interfere with complex interactions between the nervous system and other organs. With concerns regarding social policy, public health, and medicine, there is a need to ensure rigorous testing for neurotoxicity. While the most common neurotoxicity tests involve using animal models, a shift towards stem cell-based platforms can potentially provide a more biologically accurate alternative in both clinical and pharmaceutical research. With this in mind, the objective of this article is to review both current technologies and recent advancements in evaluating neurotoxicants using stem cell-based approaches, with an emphasis on developmental neurotoxicants (DNTs) as these have the most potential to lead to irreversible critical damage on brain function. In the next section, attempts to develop novel predictive model approaches for the study of both neural cell fate and developmental neurotoxicity are discussed. Finally, this article concludes with a discussion of the future use of in silico methods within developmental neurotoxicity testing, and the role of regulatory bodies in promoting advancements within the space.

Influencing factors on the quality of recovery after total knee arthroplasty: development of a predictive model.

Introduction: Total Knee Arthroplasty (TKA) is a widely performed procedure that significantly benefits patients with severe knee degeneration. However, the recovery outcomes post-surgery can vary significantly among patients. Identifying the factors influencing these outcomes is crucial for improving patient care and satisfaction. Methods: In this retrospective study, we analyzed 362 TKA cases performed between January 1, 2018, and July 1, 2022. Multivariate logistic regression was employed to identify key predictors of recovery within the first year after surgery. Results: The analysis revealed that Body Mass Index (BMI), age-adjusted Charlson Comorbidity Index (aCCI), sleep quality, Bone Mineral Density (BMD), and analgesic efficacy were significant predictors of poor recovery (p < 0.05). These predictors were used to develop a clinical prediction model, which demonstrated strong predictive ability with an Area Under the Receiver Operating Characteristic (AUC) curve of 0.802. The model was internally validated. Discussion: The findings suggest that personalized postoperative care and tailored rehabilitation programs based on these predictors could enhance recovery outcomes and increase patient satisfaction following TKA.

Reinforcement learning processes as forecasters of depression remission.

BACKGROUND: Aspects of reinforcement learning have been associated with specific depression symptoms and may inform the course of depressive illness. METHODS: We applied support vector machines to investigate whether blood­oxygen-level dependent (BOLD) responses linked with neural prediction error (nPE) and neural expected value (nEV) from a probabilistic learning task could forecast depression remission. We investigated whether predictions were moderated by treatment use or symptoms. Participants included 55 individuals (n = 39 female) with a depression diagnosis at baseline; 36 of these individuals completed standard cognitive behavioral therapy and 19 were followed during naturalistic course of illness. All participants were assessed for depression diagnosis at a follow-up visit. RESULTS: Both nPE and nEV classifiers forecasted remission significantly better than null classifiers. The nEV classifier performed significantly better than the nPE classifier. We found no main or interaction effects of treatment status on nPE or nEV accuracy. We found a significant interaction between nPE-forecasted remission status and anhedonia, but not for negative affect or anxious arousal, when controlling for nEV-forecasted remission status. LIMITATIONS: Our sample size, while comparable to that of other studies, limits options for maximizing and evaluating model performance. We addressed this with two standard methods for optimizing model performance (90:10 train and test scheme and bootstrapped sampling). CONCLUSIONS: Results support nEV and nPE as relevant biobehavioral signals for understanding depression outcome independent of treatment status, with nEV being stronger than nPE as a predictor of remission. Reinforcement learning variables may be useful components of an individualized medicine framework for depression healthcare.

Machine Learning Prediction of Residual and Recurrent High-Grade CIN Post-LEEP.

Purpose: This study aims to develop a machine learning (ML) model to predict the risk of residual or recurrent high-grade cervical intraepithelial neoplasia (CIN) after loop electrosurgical excision procedure (LEEP), addressing a critical gap in personalized follow-up care. Methods: A retrospective analysis of 532 patients who underwent LEEP for high-grade CIN at Cangzhou Central Hospital (2016-2020) was conducted. In the final analysis, 99 women (18.6%) were found to have residual or recurrent high-grade CIN (CIN2 or worse) within five years of follow-up. Four feature selection methods identified significant predictors of residual or recurrent CIN. Eight ML algorithms were evaluated using performance metrics such as AUROC, accuracy, sensitivity, specificity, PPV, NPV, F1 score, calibration curve, and decision curve analysis. Fivefold cross-validation optimized and validated the model, and SHAP analysis assessed feature importance. Results: The XGBoost algorithm demonstrated the highest predictive performance with the best AUROC. The optimized model included six key predictors: age, ThinPrep cytologic test (TCT) results, HPV classification, CIN severity, glandular involvement, and margin status. SHAP analysis identified CIN severity and margin status as the most influential predictors. An online prediction tool was developed for real-time risk assessment. Conclusion: This ML-based predictive model for post-LEEP high-grade CIN provides a significant advancement in gynecologic oncology, enhancing personalized patient care and facilitating early intervention and informed clinical decision-making.

Kinome state is predictive of cell viability in pancreatic cancer tumor and cancer-associated fibroblast cell lines.

Numerous aspects of cellular signaling are regulated by the kinome-the network of over 500 protein kinases that guides and modulates information transfer throughout the cell. The key role played by both individual kinases and assemblies of kinases organized into functional subnetworks leads to kinome dysregulation driving many diseases, particularly cancer. In the case of pancreatic ductal adenocarcinoma (PDAC), a variety of kinases and associated signaling pathways have been identified for their key role in the establishment of disease as well as its progression. However, the identification of additional relevant therapeutic targets has been slow and is further confounded by interactions between the tumor and the surrounding tumor microenvironment. In this work, we attempt to link the state of the human kinome, or kinotype, with cell viability in treated, patient-derived PDAC tumor and cancer-associated fibroblast cell lines. We applied classification models to independent kinome perturbation and kinase inhibitor cell screen data, and found that the inferred kinotype of a cell has a significant and predictive relationship with cell viability. We further find that models are able to identify a set of kinases whose behavior in response to perturbation drive the majority of viability responses in these cell lines, including the understudied kinases CSNK2A1/3, CAMKK2, and PIP4K2C. We next utilized these models to predict the response of new, clinical kinase inhibitors that were not present in the initial dataset for model devlopment and conducted a validation screen that confirmed the accuracy of the models. These results suggest that characterizing the perturbed state of the human protein kinome provides significant opportunity for better understanding of signaling behavior and downstream cell phenotypes, as well as providing insight into the broader design of potential therapeutic strategies for PDAC.

Development and evaluation of predictive models for pregnancy risk in UK dairy cows.

One suggested approach to improve the reproductive performance of dairy herds is through the targeted management of subgroups of biologically similar animals, such as those with similar probabilities of becoming pregnant, termed pregnancy risk. We aimed to use readily available farm data to develop predictive models of pregnancy risk in dairy cows. Data from a convenience sample of 108 dairy herds in the UK were collated and each herd was randomly allocated, at a ratio of 80:20, to either training or testing data sets. Following data cleaning, there were a total of 78 herds in the training data set and 20 herds in the testing data set. Data were further split by parity into nulliparous, primiparous, and multiparous subsets. An XGBoost model was trained to predict the insemination outcome in each parity subset, with predictors from farm records of breeding, calving and milk recording. Training data comprised 74,511 inseminations in 45,909 nulliparous animals, 86,420 inseminations in 39,439 primiparous animals, and 158,294 inseminations in 32,520 multiparous animals. The final models were evaluated by predicting with the testing data, comprising 31,740 inseminations in 19,647 nulliparous animals, 38,588 inseminations in 16,215 primiparous animals, and 65,049 inseminations in 12,439 multiparous animals. Model discrimination was assessed by calculating the area under receiver operating characteristic curves (AUC); model calibration was assessed by plotting calibration curves and compared across test herds by calculating the expected calibration error (ECE) in each test herd. The models were unable to discriminate between insemination outcomes with high accuracy, with an AUC of 0.63, 0.59 and 0.62 in the nulliparous, primiparous and multiparous subsets, respectively. The models were generally well-calibrated, meaning the model-predicted pregnancy risks were similar to the observed pregnancy risks. The mean (SD) ECE in the test herds was 0.038 (0.023), 0.028 (0.012) and 0.020 (0.008) in the nulliparous, primiparous and multiparous subsets respectively. The predictive models reported here could theoretically be used to identify subgroups of animals with similar pregnancy risk to facilitate targeted reproductive management; or provide information about cows' relative pregnancy risk compared with the herd average, which may support on-farm decision-making. Further research is needed to evaluate the generalizability of these predictive models and understand the source of variation in ECE between herds; however, this study demonstrates that it is possible to accurately predict pregnancy risk in dairy cows using readily available farm data.

The associations of early-life health and performance with subsequent dairy cow longevity, productivity, and profitability.

Selecting and raising dairy animals that are more likely to reach their potential is a strategy to increase milk production efficiency and overall profitability. However, indicators are necessary for the early identification of animals that are less likely to perform well, allowing for their early culling and ensuring that resources are allocated to those with the highest potential. The objective of this study was to analyze the association between early-life animal health and performance with longevity, production, and profitability. After data cleaning, the following early-life measures (i.e., predictors) were available for 363 female calves born between June 2014 and November 2015 in eight dairy herds from New Brunswick, Canada (average: 45 calves/farm; SD: 26.1 calves/farm; median: 42 calves/farm; range: 15-95 calves/farm): birth weight, weaning weight, weaning age, weaning average daily gain (weaning ADG), immunoglobulin G (IgG) serum concentration, the occurrence of navel infection, diarrhea, and pneumonia, and if animals received antibiotic treatment between birth and weaning. Their subsequent length of life (LL), length of productive life (LPL), lifetime cumulative energy-corrected milk (ECM), and lifetime cumulative milk value (i.e., response variables) were provided by the Canadian dairy herd improvement agency. Bayesian Additive Regression Tree models were trained for each response variable using 5-fold cross-validation. Models were evaluated using the RMSE and R2. The three most important predictors were identified using permutation, and the relationship between response variables and important predictors was assessed using accumulated local effect plots. The RMSE for LL, LPL, ECM, and milk value were 1.43 years, 1.37 years, 16 314.94 kg, and $CAD 11 525.68, respectively, whereas the R2 values were 0.30, 0.25, 0.29, and 0.29, respectively, indicating a moderate relationship between predictors and response variables. Non-linear relationships were found between the response variables and important predictors. Animals born with low or high birth weights were associated with decreased LL, LPL, ECM, and milk value. The highest LL, LPL, and milk value was observed for calves weaned between 1.9 and 2.0 months old, followed by a decline for calves weaned at older ages. The lowest LL and ECM were associated with weaning ADG of 0.786 kg/day, while 0.787 kg/day was associated with the lowest LPL. Lastly, both ECM and milk value were highest when serum IgG values were 1 659 mg/dL. These findings provide valuable insights for optimizing early culling decisions and enhancing the productivity and profitability of dairy farms.

A Hierarchical Model to Predict Time of Flowering of Kiwifruit Using Weather Data and Budbreak Dynamics.

Accurate prediction of flowering times is essential for efficient orchard management for kiwifruit, facilitating timely pest and disease control and pollination interventions. In this study, we developed a predictive model for flowering time using weather data and observations of budbreak dynamics for the 'Hayward' and 'Zesy002' kiwifruit. We used historic data of untreated plants collected from 32 previous studies conducted between 2007 and 2022 and analyzed budbreak and flowering timing alongside cumulative heat sum (growing degree days, GDDs), chilling unit (CU) accumulation, and other environmental variables using weather data from the weather stations nearest to the study orchards. We trained/parameterized the model with data from 2007 to 2019, and then evaluated the model's efficacy using testing data from 2020 to 2022. Regression models identified a hierarchical structure with the accumulation of GDDs at the start of budbreak, one of the key predictors of flowering time. The findings suggest that integrating climatic data with phenological events such as budbreak can enhance the predictability of flowering in kiwifruit vines, offering a valuable tool for kiwifruit orchard management.