Multivariate Joint Model Research Articles

Dynamic Prediction of Cardiovascular Events in Incident Patients on Peritoneal Dialysis with Multivariate Joint Models Adjusting for Phosphate, Albumin, and Calcium Trajectories and Competing Risks

Dynamic Prediction of Cardiovascular Events in Incident Patients on Peritoneal Dialysis with Multivariate Joint Models Adjusting for Phosphate, Albumin, and Calcium Trajectories and Competing Risks

Joint modeling of monocyte HLA-DR expression trajectories predicts 28-day mortality in severe SARS-CoV-2 patients.

The recent SarsCov2 pandemic has disrupted healthcare system notably impacting intensive care units (ICU). In severe cases, the immune system is dysregulated, associating signs of hyperinflammation and immunosuppression. In the present work, we investigated, using a joint modeling approach, whether the trajectories of cellular immunological parameters were associated with survival of COVID-19 ICU patients. This study is based on the REA-IMMUNO-COVID cohort including 538 COVID-19 patients admitted to ICU between March 2020 and May 2022. Measurements of monocyte HLA-DR expression (mHLA-DR), counts of neutrophils, of total lymphocytes, and of CD4+ and CD8+ subsets were performed five times during the first month after ICU admission. Univariate joint models combining survival at day 28 (D28), hospital discharge and longitudinal analysis of those biomarkers' kinetics with mixed-effects models were performed prior to the building of a multivariate joint model. We showed that a higher mHLA-DR value was associated with a lower risk of death. Predicted mHLA-DR nadir cutoff value that maximized the Youden index was 5414 Ab/C and led to an AUC = 0.70 confidence interval (95%CI) = [0.65; 0.75] regarding association with D28 mortality while dynamic predictions using mHLA-DR kinetics until D7, D12 and D20 showed AUCs of 0.82 [0.77; 0.87], 0.81 [0.75; 0.87] and 0.84 [0.75; 0.93]. Therefore, the final joint model provided adequate discrimination performances at D28 after collection of biomarker samples until D7, which improved as more samples were collected. After severe COVID-19, decreased mHLA-DR expression is associated with a greater risk of death at D28 independently of usual clinical confounders.

Bayesian modeling of spatially differentiated multivariate enamel defects of the children’s primary maxillary central incisor teeth

BackgroundThe analysis of dental caries has been a major focus of recent work on modeling dental defect data. While a dental caries focus is of major importance in dental research, the examination of developmental defects which could also contribute at an early stage of dental caries formation, is also of potential interest. This paper proposes a set of methods which address the appearance of different combinations of defects across different tooth regions. In our modeling we assess the linkages between tooth region development and both the type of defect and associations with etiological predictors of the defects which could be influential at different times during the tooth crown development.MethodsWe develop different hierarchical model formulations under the Bayesian paradigm to assess exposures during primary central incisor (PMCI) tooth development and PMCI defects. We evaluate the Bayesian hierarchical models under various simulation scenarios to compare their performance with both simulated dental defect data and real data from a motivating application.ResultsThe proposed model provides inference on identifying a subset of etiological predictors of an individual defect accounting for the correlation between tooth regions and on identifying a subset of etiological predictors for the joint effect of defects. Furthermore, the model provides inference on the correlation between the regions of the teeth as well as between the joint effect of the developmental enamel defects and dental caries. Simulation results show that the proposed model consistently yields steady inferences in identifying etiological biomarkers associated with the outcome of localized developmental enamel defects and dental caries under varying simulation scenarios as deemed by small mean square error (MSE) when comparing the simulation results to real application results.ConclusionWe evaluate the proposed model under varying simulation scenarios to develop a model for multivariate dental defects and dental caries assuming a flexible covariance structure that can handle regional and joint effects. The proposed model shed new light on methods for capturing inclusive predictors in different multivariate joint models under the same covariance structure and provides a natural extension to a nested hierarchical model.

Investigating the Association Between Dynamic Driving Pressure and Mortality in COVID-19-Related Acute Respiratory Distress Syndrome: A Joint Modeling Approach Using Real-Time Continuously-Monitored Ventilation Data.

COVID-19-related acute respiratory distress syndrome (ARDS) is associated with high mortality and often necessitates invasive mechanical ventilation (IMV). Previous studies on non-COVID-19 ARDS have shown driving pressure to be robustly associated with ICU mortality; however, those studies relied on "static" driving pressure measured periodically and manually. As "continuous" automatically monitored driving pressure is becoming increasingly available and reliable with more advanced mechanical ventilators, we aimed to examine the effect of this "dynamic" driving pressure in COVID-19 ARDS throughout the entire ventilation period. This retrospective, observational study cohort study evaluates the association between driving pressure and ICU mortality in patients with concurrent COVID-19 and ARDS using multivariate joint modeling. The study cohort (n = 544) included all adult patients (≥ 18 yr) with COVID-19 ARDS between March 1, 2020, and April 30, 2021, on volume-control mode IMV for 12 hours or more in a Mass General Brigham, Boston, MA ICU. Of 544 included patients, 171 (31.4%) died in the ICU. Increased dynamic ΔP was associated with increased risk in the hazard of ICU mortality (hazard ratio [HR] 1.035; 95% credible interval, 1.004-1.069) after adjusting for other relevant dynamic respiratory biomarkers. A significant increase in risk in the hazard of death was found for every hour of exposure to high intensities of driving pressure (≥ 15 cm H2O) (HR 1.002; 95% credible interval 1.001-1.003). Limiting patients' exposure to high intensities of driving pressure even while under lung-protective ventilation may represent a critical step in improving ICU survival in patients with COVID-19 ARDS. Time-series IMV data could be leveraged to enhance real-time monitoring and decision support to optimize ventilation strategies at the bedside.

The Effect of Residence Time of No-tunneled Hemodialysis Catheters on Infection and Thrombosis Outcome. Identification of CVC’s Time Cut-off

Introduction: Permanent vascular access (arteriovenous fistula (AVF), arteriovenous graft (AVG)) is susceptible to acute events that reduce patency. The temporary central venous catheter (CVC) constitutes bridging therapy for primary vascular access dysfunction. The impact of “residence time” on the rate of dysfunction/thrombosis or infection remains to be explored. AIM: 1) To evaluate the impact of CVC residence time on outcomes (infection or Thrombosis/dysfunction) in consecutive temporary CVCs adjusted for the insertion site (upper site vs. lower site). 2) To establish a cut-off resident time. Patients and methods: Seventeen prevalent hemodialysis patients with three consecutive CVCs are followed up prospectively in an observational study for a period equivalent to the permanence of the CVCs. The data is recorded at the beginning of the CVC time. The diagnosis of catheter-related bloodstream infection and thrombosis/dysfunction is made following the K-Doqi 2019 guidelines. Statistical analysis: Seventeen hemodialysis patients (51 CVCs) were included. The ‘CVC resident time’ of each individual patient ((i.e. βcoefficient (log-transformed)*AUC)) was determined using LMM and then inserted into multivariate Cox models to assess infection and dysfunction/thrombosis outcomes (Joint Models). The AUC was calculated at various baseline levels of CVC time (10th……50th percentile). The cut-off point for thrombosis in CVC time corresponds to the mean of the CVC time at the 30th percentile of all CVCs. Results: The CVC time is different for CVC’s site insertion and sequence. From the analysis of multivariate joint models, CVC resident time appears not to be significant for infection, but heterogenicity for the insertion site (ref3-4=upper site) is significant for the outcome of thrombosis/dysfunction. From the study of survival analysis, the free survival from outcomes by CVC site insertion appears to be significant for thrombosis/dysfunction. The average time of CVCs’ calculation at the 30th percentile is 14 days (cut-off). Conclusion: No tunneled hemodialysis Catheter (NTHC) residence time is considered not to be a risk factor for infection, but it represents a risk factor for lower access thrombosis. After the cut-off time of 14 days, the advantage of the higher NTHCs is lost.

Pairwise fitting of piecewise mixed models for the joint modeling of multivariate longitudinal outcomes, in a randomized crossover trial.

Many statistical models have been proposed in the literature for the analysis of longitudinal data. One may propose to model two or more correlated longitudinal processes simultaneously, with a goal of understanding their association over time. Joint modeling is then required to carefully study the association structure among the outcomes as well as drawing joint inferences about the different outcomes. In this study, we sought to model the associations among six nutrition outcomes while circumventing the computational challenge posed by their clustered and high-dimensional nature. We analyzed data from a 2 2 randomized crossover trial conducted in Kenya, to compare the effect of high-dose and low-dose iodine in household salt on systolic blood pressure (SBP) and diastolic blood pressure (DBP) in women of reproductive age and their household matching pair of school-aged children. Two additional outcomes, namely, urinary iodine concentration (UIC) in women and children were measured repeatedly to monitor the amount of iodine excreted through urine. We extended the model proposed by Mwangi etal. (2021, Communications in Statistics: Case Studies, Data Analysis and Applications, 7(3), 413-431) allowing flexible piecewise joint models for six outcomes to depend on separate random effects, which are themselves correlated. This entailed fitting 15 bivariate general linear mixed models and deriving inference for the joint model using pseudo-likelihood theory. We analyzed the outcomes separately and jointly using piecewise linear mixed-effects (PLME) model and further validated the results using current state-of-the-art Jones and Kenward methodology (JKME model) used for analyzing randomized crossover trials. The results indicate that high-dose iodine in salt significantly reduced blood pressure (BP) compared to low-dose iodine in salt. Estimates for the random effects and residual error components showed that SBP and DBP had strong positive correlation, with effect of the random slope indicating that significantly related outcomes are strongly associated in their evolution. There was a moderately strong inverse relationship between evolutions of UIC and BP both in women and children. These findings confirmed the original hypothesis that high-dose iodine salt has significant lowering effect on BP. We further sought to evaluate the performance of our proposed PLME model against the widely used JKME model, within the multivariate joint modeling framework through a simulation study mimicking a crossover design. From our findings, the multivariate joint PLME model performed exceptionally well both in estimation of random-effects matrix (G) and Hessian matrix (H), allowing satisfactory model convergence during estimation. It allowed a more complex fit to the data with both random intercepts and slopes effects compared to the multivariate joint JKME model that allowed for random intercepts only. When a hierarchical viewpoint is adopted, in the sense that outcomes are specified conditionally upon random effects, the variance-covariance matrix of the random effects must be positive definite. In some cases, additional random effects could explain much variability in the data, thus improving precision in estimation of the estimands (effect size) parameters. The key highlight in this evaluation shows that multivariate joint JKME model is a powerful tool especially while fitting mixed models with random intercepts only, in crossover design settings. Addition of random slopes may lead to model complexities in most cases, resulting in unsatisfactory model convergence during estimation. To circumvent convergence pitfalls, extention of JKME model to PLME model allows a more flexible fit to the data (generated from crossover design settings), especially in the multivariate joint modelingframework.

Multivariate joint model under competing risks to predict death of hospitalized patients for SARS-CoV-2 infection.

During the coronavirus disease 2019 (COVID-19) pandemic, several clinical prognostic scores have been proposed and evaluated in hospitalized patients, relying on variables available at admission. However, capturing data collected from the longitudinal follow-up of patients during hospitalization may improve prediction accuracy of a clinical outcome. To answer this question, 327 patients diagnosed with COVID-19 and hospitalized in an academic French hospital between January and July 2020 are included in the analysis. Up to 59 biomarkers were measured from the patient admission to the time to death or discharge from hospital. We consider a joint model with multiple linear or nonlinear mixed-effects models for biomarkers evolution, and a competing risks model involving subdistribution hazard functions for the risks of death and discharge. The links are modeled by shared random effects, and the selection of the biomarkers is mainly based on the significance of the link between the longitudinal and survival parts. Three biomarkers are retained: the blood neutrophil counts, the arterial pH, and the C-reactive protein. The predictive performances of the model are evaluated with the time-dependent area under the curve (AUC) for different landmark and horizon times, and compared with those obtained from a baseline model that considers only information available at admission. The joint modeling approach helps to improve predictions when sufficient information is available. For landmark 6 days and horizon of 30 days, we obtain AUC [95% CI] 0.73 [0.65, 0.81] and 0.81 [0.73, 0.89] for the baseline and joint model, respectively (p = 0.04). Statistical inference is validated through a simulationstudy.

Diagnostic performance and longitudinal analysis of fungal biomarkers in COVID-19 associated pulmonary aspergillosis

ObjectivesGalactomannan lateral flow assay (GM-LFA) is a reliable test for COVID-19 associated pulmonary aspergillosis (CAPA) diagnosis. We aimed to assess the diagnostic performance of GM-LFA with different case definitions, the association between the longitudinal measurements of serum GM-ELISA, GM-LFA, and the risk of death. MethodsSerum and nondirected bronchial lavage (NBL) samples were periodically collected. The sensitivity and specificity analysis for GM-LFA was done in different time periods. Longitudinal analysis was done with the joint model framework. ResultsA total of 207 patients were evaluated. On the day of CAPA diagnosis, serum GM-LFA had a sensitivity of 42 % (95 % CI: 23–63) and specificity of 82 % (95 % CI: 78–84), while NBL GM-LFA had a sensitivity of 73 % (95 % CI: 45–92), specificity of 85 % (95 % CI: 76–91) for CAPA. Sensitivity decreased through the following days in both samples. Univariate joint model analysis showed that increasing GM-LFA and GM-ELISA levels were associated with increased mortality, and that effect remained same with serum GM-ELISA in multivariate joint model analysis. ConclusionGM-LFA, particularly in NBL samples, seems to be a reliable method for CAPA diagnosis. For detecting patients with higher risk of mortality, longitudinal measurement of serum GM-ELISA can be useful.

Evaluation of the Kinetics of Pancreatic Stone Protein as a Predictor of Ventilator-Associated Pneumonia.

Ventilator-associated pneumonia (VAP) is a severe condition. Early and adequate antibiotic treatment is the most important strategy for improving prognosis. Pancreatic Stone Protein (PSP) has been described as a biomarker that increases values 3-4 days before the clinical diagnosis of nosocomial sepsis in different clinical settings. We hypothesized that serial measures of PSP and its kinetics allow for an early diagnosis of VAP. The BioVAP study was a prospective observational study designed to evaluate the role of biomarker dynamics in the diagnosis of VAP. To determine the association between repeatedly measured PSP and the risk of VAP, we used joint models for longitudinal and time-to-event data. Of 209 patients, 43 (20.6%) patients developed VAP, with a median time of 4 days. Multivariate joint models with PSP, CRP, and PCT did not show an association between biomarkers and VAP for the daily absolute value, with a hazard ratio (HR) for PSP of 1.01 (95% credible interval: 0.97 to 1.05), for CRP of 1.00 (0.83 to 1.22), and for PCT of 0.95 (0.82 to 1.08). The daily change of biomarkers provided similar results, with an HR for PSP of 1.15 (0.94 to 1.41), for CRP of 0.76 (0.35 to 1.58), and for PCT of 0.77 (0.40 to 1.45). Neither absolute PSP values nor PSP kinetics alone nor in combination with other biomarkers were useful in improving the prediction diagnosis accuracy in patients with VAP. Registered retrospectively on August 3rd, 2012. NCT02078999.

Semiparametric multivariate joint model for skewed-longitudinal and survival data: A Bayesian approach.

Joint models and statistical inference for longitudinal and survival data have been an active area of statistical research and have mostly coupled a longitudinal biomarker-based mixed-effects model with normal distribution and an event time-based survival model. In practice, however, the following issues may standout: (i) Normality of model error in longitudinal models is a routine assumption, but it may be unrealistically violating data features of subject variations. (ii) Data collected are often featured by the mixed types of multiple longitudinal outcomes which are significantly correlated, ignoring their correlation may lead to biased estimation. Additionally, a parametric model specification may be inflexible to capture the complicated patterns of longitudinal data. (iii) Missing observations in the longitudinal data are often encountered; the missing measures are likely to be informative (nonignorable) and ignoring this phenomenon may result in inaccurate inference. Multilevel item response theory (MLIRT) models have been increasingly used to analyze the multiple longitudinal data of mixed types (ie, continuous and categorical) in clinical studies. In this article, we develop an MLIRT-based semiparametric joint model with skew-t distribution that consists of an extended MLIRT model for the mixed types of multiple longitudinal data and a Cox proportional hazards model, linked through random-effects. A Bayesian approach is employed for joint modeling. Simulation studies are conducted to assess performance of the proposed models and method. A real example from primary biliary cirrhosis clinical study is analyzed to estimate parameters in the joint model and also evaluate sensitivity of parameter estimates for various plausible nonignorable missing data mechanisms.

Prognostic Role of Dynamic Changes in Serological Markers in Metastatic Hormone Naïve Prostate Cancer.

We investigated whether inter-patient variation in the dynamic trajectory of hemoglobin (Hb), neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), lymphocyte to monocyte ratio (LMR), and prostate-specific antigen (PSA) can prognosticate overall survival (OS) in de novo mHSPC. This is a secondary analysis of the LATITUDE trial in which high-risk de novo mHSPC patients were randomly assigned to receive either androgen deprivation therapy (ADT) plus abiraterone or ADT plus placebo. We used a five-fold cross-validated joint model approach to determine the association of temporal changes in the serological markers with OS. Decision curve analysis was applied to determine the net benefit. When dynamic changes in Hb, LMR, NLR, PLR, and PSA were included in a multivariate joint model, an increase in the log of the current value of PSA (HR: 1.24 [1.20-1.28]) was associated with inferior OS. A multivariate joint model that captured dynamic trajectory of Hb, NLR, PLR, LMR, and PSA up to 24 months, showed a net benefit over the "treat all" strategy at a threshold of probability of approximately ≥30% while no net benefit was seen when dynamic change in PSA was omitted. Our joint model could be used for designing future adaptive trials investigating sequential treatment personalization.

Fast and flexible inference for joint models of multivariate longitudinal and survival data using integrated nested Laplace approximations.

Modeling longitudinal and survival data jointly offers many advantages such as addressing measurement error and missing data in the longitudinal processes, understanding and quantifying the association between the longitudinal markers and the survival events, and predicting the risk of events based on the longitudinal markers. A joint model involves multiple submodels (one for each longitudinal/survival outcome) usually linked together through correlated or shared random effects. Their estimation is computationally expensive (particularly due to a multidimensional integration of the likelihood over the random effects distribution) so that inference methods become rapidly intractable, and restricts applications of joint models to a small number of longitudinal markers and/or random effects. We introduce a Bayesian approximation based on the integrated nested Laplace approximation algorithm implemented in the R package R-INLA to alleviate the computational burden and allow the estimation of multivariate joint models with fewer restrictions. Our simulation studies show that R-INLA substantially reduces the computation time and the variability of the parameter estimates compared with alternative estimation strategies. We further apply the methodology to analyze five longitudinal markers (3 continuous, 1 count, 1 binary, and 16 random effects) and competing risks of death and transplantation in a clinical trial on primary biliary cholangitis. R-INLA provides a fast and reliable inference technique for applying joint models to the complex multivariate data encountered in health research.

Fast estimation for generalised multivariate joint models using an approximate EM algorithm

Joint models for longitudinal and survival data have become an established tool for optimally handling scenarios when both types of data co-exist. Multivariate extensions to the classic univariate joint model have started to emerge but are typically restricted to the Gaussian case, deployed in a Bayesian framework or focused on dimension reduction. An approximate EM algorithm is utilised which circumvents the oft-lamented curse of dimensionality and offers a likelihood-based implementation which ought to appeal to clinicians and practitioners alike. The proposed method is validated in a pair of simulation studies, which demonstrate both its accuracy in parameter estimation and efficiency in terms of computational cost. Its clinical use is demonstrated via an application to primary billiary cirrhosis data. The proposed methodology for estimation of these joint models is available in R package gmvjoint.

Novel mathematical modeling approach finds Fortasyn Connect affects memory through reducing hippocampal atrophy in the LipiDiDiet trial

AbstractBackgroundLipiDiDiet is a 6‐year, randomized, double‐blind, placebo‐controlled intervention in 311 individuals with prodromal Alzheimer’s disease (AD)/mild cognitive impairment receiving the specific multinutrient combination Fortasyn Connect (Souvenaid) or a calorie‐matched placebo control once daily.1 AD is characterized by neurodegeneration including significant synapse loss, and the hippocampal formation is particularly vulnerable in the early stages of AD. Results over the first 2 and 3 years of intervention showed that Fortasyn Connect reduced hippocampal atrophy and slowed decline on clinical and other measures related to cognition, function, and disease progression.2,3 Here, possible mechanistic pathways through which the intervention may affect the risk of disease progression are being explored post‐hoc.MethodA new type of multivariate joint model was applied to the 3‐year LipiDiDiet clinical trial data in which the longitudinal outcomes MRI hippocampal volume (HCV) and neuropsychological test battery (NTB)‐memory are allowed to be associated with disease progression and dropout, and affect each other directly. Specifically, for each longitudinal outcome a linear mixed‐effects model is used to estimate its trajectory, where, for the second longitudinal outcome, the linear predictor of the first outcome is included as time‐varying covariate. Results are compared for both pathways (HCV and NTB‐memory and vice versa), and with results of existing multivariate joint models that have the form of so‐called parallel mediator models (i.e. longitudinal outcomes are merely allowed to be correlated but not to influence each other causally).ResultComparing the models, we observed a strong association between changes in NTB‐memory and disease progression and no direct treatment effect on NTB‐memory. Notably, there was a strong direct treatment effect for HCV changes, together with a direct effect of HCV on disease progression. Moreover, indirect effects of HCV changes on disease progression were mediated through NTB‐memory.ConclusionThis new multivariate joint model is useful in modeling possible mechanistic pathways. Overall results from application to the 3‐year data suggest that the observed intervention effects of Fortasyn Connect are partly mediated through reduction of hippocampal atrophy both directly and indirectly via NTB‐memory, in addition to yet to be defined other/complementary pathways. 1 Funding includes EU‐JPND. 2Lancet Neurol. 2017;16:965‐975. 3Alz & Dementia. 2021;17:29‐40.

Association of dynamic changes in serological markers with survival in de novo metastatic hormone-sensitive prostate cancer.

5069 Background: Prognostic association of serological markers of systemic inflammatory response have been demonstrated in metastatic castrate resistant prostate cancer. However, it remains unknown whether dynamic changes in these markers over time are prognostic earlier in the disease process, namely in metastatic hormone sensitive prostate cancer (mHSPC). We performed a secondary analysis of LATITUDE trial to determine if dynamic changes in hemoglobin (Hb), neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR), and lymphocyte to monocyte ratio (LMR) are predictive of prostate cancer-specific survival (PCSS), and overall survival (OS). Methods: We used a joint model approach to determine the association of the dynamic change in the marker levels with PCSS, and OS. For the time-to-event submodel, a multivariable Cox regression model was constructed with treatment arm, skeletal lesion number, liver or lung metastasis, ECOG performance status, and age. For the longitudinal submodel, a linear mixed-effects model was built with an interaction term for treatment arm and time of evaluation in addition to treatment arm, time of evaluation, and baseline value of the serological markers. The two submodels were linked through a shared random effect. Results: Overall 1172 patients were eligible - 580 from the abiraterone plus ADT arm and 592 from the ADT alone group. Median follow-up for surviving patients was 52 months (IQR 47-57). Median number of post-baseline assessments was 16 (IQR 10-28). On univariate joint models, every 10 g/L dynamic increase in Hb was associated with superior PCSS (HR 0.71 [0.67-0.75]) and OS (HR 0.74 [0.68-0.79]) while every 5 points dynamic increase in LMR was associated with a superior PCSS (HR 0.38 [0.26-0.53]) and OS (HR 0.41 [0.29-0.56]). In contrast, dynamic increase in NLR was associated with inferior PCSS (HR 1.29 [1.22-1.36]) and OS (HR 1.29 [1.23-1.36]) while every 10-point dynamic increase in PLR was associated with a small but significant deterioration in PCSS (HR 1.05 [1.04-1.06]) and OS (HR 1.05 [1.04-1.06]), respectively. On multivariate joint modeling, dynamic increase in Hb was also associated with superior PCSS (HR per 10g/L increase 0.74 [0.69-0.79]) and OS (HR per 10g/L rise 0.75 [0.71-0.80]). When dynamic changes in Hb, LMR, NLR, and PLR were included in the same multivariate model with dynamic change in PSA, dynamic increase in Hb continued to show association with significantly superior PCSS (HR per 10g/L rise 0.80 [0.75-0.86]), and OS (HR per 10g/L rise 0.81 [0.75-0.86]), respectively. Conclusions: Our findings suggest that dynamic increase in hemoglobin can predict for superior PCSS, and OS in men with de novo mHSPC treated with ADT with or without abiraterone. These findings need additional validation before implementing routine use of hemoglobin as a prognostic biomarker in de novo mHSPC.

A multivariate joint model to adjust for random measurement error while handling skewness and correlation in dietary data in an epidemiologic study of mortality

PurposeA substantial proportion of global deaths is attributed to unhealthy diets, which can be assessed at baseline or longitudinally. We demonstrated how to simultaneously correct for random measurement error, correlations, and skewness in the estimation of associations between dietary intake and all-cause mortality. MethodsWe applied a multivariate joint model (MJM) that simultaneously corrected for random measurement error, skewness, and correlation among longitudinally measured intake levels of cholesterol, total fat, dietary fiber, and energy with all-cause mortality using US National Health and Nutrition Examination Survey linked to the National Death Index mortality data. We compared MJM with the mean method that assessed intake levels as the mean of a person’s intake. ResultsThe estimates from MJM were larger than those from the mean method. For instance, the logarithm of hazard ratio for dietary fiber intake increased by 14 times (from −0.04 to −0.60) with the MJM method. This translated into a relative hazard of death of 0.55 (95% credible interval: 0.45, 0.65) with the MJM and 0.96 (95% credible interval: 0.95, 0.97) with the mean method. ConclusionsMJM adjusts for random measurement error and flexibly addresses correlations and skewness among longitudinal measures of dietary intake when estimating their associations with death.

Multivariate joint models for the dynamic prediction of psychosis in individuals with clinical high risk

This study attempted to construct and validate dynamic prediction via multivariate joint models and compare the prognostic performance of these models to both static and univariate joint models. Individuals with clinical high risk(CHR)(n=289) were recruited and re-assessed for positive symptoms, general functions, and conversion to psychosis at 2-months, 1-year, and 2-years to develop the dynamic models. A multivariate joint model of positive psychotic symptoms was assessed using the Structured Interview for Prodromal Symptoms(SIPSp) and general function assessed by global assessment of functioning scores(GAFs) with time-to-conversion to psychosis. The area under the receiver operating characteristic(ROC) curve(AUC) was used to test the accuracy of the models. Among 298 CHR individuals, 68 converted to psychosis within 2 years after the initial assessments. Multivariate joint models showed that declining GAFs and increasing SIPSp corresponded to significant and trending to significantly increased risk of psychosis onset and had much higher prognostic accuracy (cross-validated AUC=0.9) compared to the static model(AUC=0.6) and univariate joint models(cross-validated AUC=0.6-0.8). Our results showed that multivariate joint models could be highly efficient in forecasting psychosis onset for CHR individuals. Longitudinal assessments for psychopathology and general functions can be useful for dynamically predicting the prognosis of the pre-morbid phase of psychosis.

Machine Learning-Based Integration of Metabolomics Characterisation Predicts Progression of Myopic Retinopathy in Children and Adolescents

Myopic retinopathy is an important cause of irreversible vision loss and blindness. As metabolomics has recently been successfully applied in myopia research, this study sought to characterize the serum metabolic profile of myopic retinopathy in children and adolescents (4-18 years) and to develop a diagnostic model that combines clinical and metabolic features. We selected clinical and serum metabolic data from children and adolescents at different time points as the training set (n = 516) and the validation set (n = 60). All participants underwent an ophthalmologic examination. Untargeted metabolomics analysis of serum was performed. Three machine learning (ML) models were trained by combining metabolic features and conventional clinical factors that were screened for significance in discrimination. The better-performing model was validated in an independent point-in-time cohort and risk nomograms were developed. Retinopathy was present in 34.2% of participants (n = 185) in the training set, including 109 (28.61%) with mild to moderate myopia. A total of 27 metabolites showed significant variation between groups. After combining Lasso and random forest (RF), 12 modelled metabolites (mainly those involved in energy metabolism) were screened. Both the logistic regression and extreme Gradient Boosting (XGBoost) algorithms showed good discriminatory ability. In the time-validation cohort, logistic regression (AUC 0.842, 95% CI 0.724-0.96) and XGBoost (AUC 0.897, 95% CI 0.807-0.986) also showed good prediction accuracy and had well-fitted calibration curves. Three clinical characteristic coefficients remained significant in the multivariate joint model (p < 0.05), as did 8/12 metabolic characteristic coefficients. Myopic retinopathy may have abnormal energy metabolism. Machine learning models based on metabolic profiles and clinical data demonstrate good predictive performance and facilitate the development of individual interventions for myopia in children and adolescents.

Eat-in or eat-out? A joint model to analyze the new landscape of dinner meal preferences

In this paper, we examine the non-home-cooked meal (NHCM) preferences of individuals for their dinner meal by studying the monthly count of NHCM meals by channel type: eat-out, eat-in takeout, and eat-in delivery. Data from a 2022 online survey collected in Texas is employed to estimate a multivariate joint model. Model estimation results indicate that the most frequent customers of the eat-out channel are white individuals, individuals from 3+ motorized vehicle-owning households, those in non-joint families, those in households with no children, full-time employees who never work from home or do so only for a small fraction of their workdays, and those residing in areas with a high density of restaurants. The distinct consumer segments for the eat-in takeout channel include young individuals, those with high household incomes, those working from home all their workdays or a substantial fraction of their workdays, and urban residents; the most enthusiastic consumers of the eat-in delivery channel are white individuals, those with less than three vehicles in the household, individuals with children, urban residents, and those worried about pandemic-related personal health risks. Older individuals, non-white individuals, individuals with a graduate degree, individuals in fewer motorized vehicle-owning households and in joint families, those with children in the household, and rural residents constitute the most committed population segments of the home-cooked meal (HCM) consumption channel. The results suggest the important impact of workplace location on dining channel choice. The results also show clear evidence of complementary and substitution effects at play; the delivery channel complements eating out but substitutes takeout. Similarly, eat-out has a substitution effect on eat-in takeout. These effects have important implications for activity-travel behavior due to emerging technology-based ordering options for dining choices, especially in the aftermath of the COVID pandemic.

Dynamic Change in Patient Reported Quality of Life is a Predictor for Survival in Localized Prostate Cancer: Exploratory Analysis from a Phase III Randomized Controlled Trial

<h3>Purpose/Objective(s)</h3> Patient reported quality of life (QOL) outcomes are pivotal endpoints to determine treatment efficacy and safety. However, it remains unknown if treatment induced change in patient reported QOL outcomes correlate with survival in localized prostate cancer (LPCa). We performed an exploratory analysis of a phase III randomized controlled study to determine the association of treatment induced dynamic change in patient reported global QOL with long-term overall survival (OS) and disease-free survival (DFS) in LPCa treated with radiotherapy (RT) and androgen deprivation therapy (ADT). <h3>Materials/Methods</h3> LPCa patients with Gleason score ≤7, clinical stage T1b-T3a, and PSA <30 ng/mL were randomized to neoadjuvant and concurrent ADT for 6 mos starting 4 mos before prostate RT or concomitant and adjuvant ADT for 6 mos starting concurrently with RT. Patient reported QOL outcomes were collected using EORTC QOL questionnaires. We used the global QOL (G-QOL) score from C30 questionnaire. Increase in G-QOL score indicated improvement in global QOL. To capture the association between the dynamic change in the G-QOL of each patient and the hazard of outcome, we utilized the framework of multivariate joint modeling for longitudinal and survival outcomes. For the time-to-event submodel, a Cox proportional hazard regression model was built with covariables including treatment arm, baseline PSA, tumor stage, Gleason score, and age. For the longitudinal submodel, a linear mixed effect model was built with an interaction term for treatment arm and time of evaluation in addition to fixed covariables - treatment arm, time of evaluation, baseline G-QOL score and age. Time of assessment was included as a random slope while patients were included as random intercepts in the mixed models. The two submodels were linked through a random effect. Baseline hazard was modeled using penalized splines. A two-sided p-value <0.025 was chosen to be the threshold for statistical significance. <h3>Results</h3> A total of 388 patients were eligible for this study – 191 belonged to the neoadjuvant arm while 197 to the adjuvant arm. At a median follow-up of 148 mos, a total of 133 deaths were recorded. On joint modelling, an increase in G-QOL score by 10 units (the threshold for clinically meaningful change) was associated with a 11.2% decrease in the hazard of death (hazard ratio [HR] per 1 unit increase G-QOL: 0.988, 95% CI: 0.980-0.996; p=0.007). An increase in G-QOL score by 10 units was associated with 10% decrease in the hazard of progression or death (HR per 1 unit increase in G-QOL: 0.990; 95% CI: 0.984-0.999; p=0.023). <h3>Conclusion</h3> Treatment induced improvement in global QOL is a predictor for improved DFS and OS in LPCa. This suggests a significant interplay of patient-reported quality of life and quantity of life in LPCa which could mainly be driven by the non-cancer specific mortality in this patient population. Judicious use of these QOL metrics could guide risk stratification and potentially inform treatment decision in these patients