Coverage Probability Research Articles

Random forest machine learning for maize yield and agronomic efficiency prediction in Ghana

Maize (Zea mays) is an important staple crop for food security in Sub-Saharan Africa. However, there is need to increase production to feed a growing population. In Ghana, this is mainly done by increasing acreage with adverse environmental consequences, rather than yield increment per unit area. Accurate prediction of maize yields and nutrient use efficiency in production is critical to making informed decisions toward economic and ecological sustainability. We trained the random forest machine learning algorithm to predict maize yield and agronomic efficiency in Ghana using soil, climate, environment, and management factors, including fertilizer application. We calibrated and evaluated the performance of the random forest machine learning algorithm using a 5 × 10-fold nested cross-validation approach. Data from 482 maize field trials consisting of 3136 georeferenced treatment plots conducted in Ghana from 1991 to 2020 were used to train the algorithm, identify important predictor variables, and quantify the uncertainties associated with the random forest predictions. The mean error, root mean squared error, model efficiency coefficient and 90 % prediction interval coverage probability were calculated. The results obtained on test data demonstrate good prediction performance for yield (MEC = 0.81) and moderate performance for agronomic efficiency (MEC = 0.63, 0.55 and 0.54 for AE-N, AE-P and AE-K, respectively). We found that climatic variables were less important predictors than soil variables for yield prediction, but temperature was of key importance to yield prediction and rainfall to agronomic efficiency. The developed random forest models provided a better understanding of the drivers of maize yield and agronomic efficiency in a tropical climate and an insight towards improving fertilizer recommendations for sustainable maize production and food security in Sub-Saharan Africa.

Balancing efficacy and computational burden: weighted mean, multiple imputation, and inverse probability weighting methods for item non-response in reliable scales.

Scales often arise from multi-item questionnaires, yet commonly face item non-response. Traditional solutions use weighted mean (WMean) from available responses, but potentially overlook missing data intricacies. Advanced methods like multiple imputation (MI) address broader missing data, but demand increased computational resources. Researchers frequently use survey data in the All of Us Research Program (All of Us), and it is imperative to determine if the increased computational burden of employing MI to handle non-response is justifiable. Using the 5-item Physical Activity Neighborhood Environment Scale (PANES) in All of Us, this study assessed the tradeoff between efficacy and computational demands of WMean, MI, and inverse probability weighting (IPW) when dealing with item non-response. Synthetic missingness, allowing 1 or more item non-response, was introduced into PANES across 3 missing mechanisms and various missing percentages (10%-50%). Each scenario compared WMean of complete questions, MI, and IPW on bias, variability, coverage probability, and computation time. All methods showed minimal biases (all <5.5%) for good internal consistency, with WMean suffered most with poor consistency. IPW showed considerable variability with increasing missing percentage. MI required significantly more computational resources, taking >8000 and >100 times longer than WMean and IPW in full data analysis, respectively. The marginal performance advantages of MI for item non-response in highly reliable scales do not warrant its escalated cloud computational burden in All of Us, particularly when coupled with computationally demanding post-imputation analyses. Researchers using survey scales with low missingness could utilize WMean to reduce computing burden.

Trends in multiple myeloma incidence, prevalence, mortality, and survival rate in South Korea: a nationwide population-based study.

This is the first study presenting the overall descriptive epidemiology of multiple myeloma (MM), including incidence, mortality rate, and prevalence, in South Korea between 2010 and 2018 based on nationwide medical insurance coverage and mortality statistics data. The incidence of MM between 2010 and 2018 was obtained from nationwide medical claims data, and mortality data were obtained from the Korea National Statistical Office. The age-standardized incidence rate (ASIR) and mortality rate (ASMR) and one- and five-year survival rates of patients with MM each year were estimated. There were 10,835 patients with MM aged ≥ 20years in South Korea between 2010 and 2018. The ASIR was 2.42/100,000 in 2010 and increased to 2.71/100,000 in 2018, with an annual percent change (APC) of 1.86% (95% CI = 0.74-2.99%, P = 0.005). While this trend was significant in women, it was not statistically significant in men. The ASMR did not significantly change over time. Furthermore, the median survival time of patients with MM diagnosed between 2010 and 2018 was 3.36years. Notably, the one-year survival rate of patients was increased from 65.3% in 2010 to 76.2% in 2017. Finally, the proportion of patients with MM who received novel therapeutic agents, such as proteasome inhibitors or immunomodulatory drugs, as first-line treatment increased from 37.7% in 2010 to 97.8% in 2018. The ASIR and prevalence of MM in South Korea increased between 2010 and 2018, especially in women and the survival rate of patients with MM has increased.

On soft measurement modeling for predicting photovoltaic power with uncertainty based on the Takagi–Sugeno model

Abstract Accurate solar power forecast is becoming more essential for safe and reliable power grid operation with the increasing number of grid-connected photovoltaic (PV) power production units. However, PV power exhibits significant output fluctuation due to both external inputs and intrinsic stochasticity in system dynamics. Therefore, an efficient and reliable soft measurement model of PV power with uncertainty is demanded in practice applications. The technique described in this paper captures the impacts produced by the fundamental uncertainty observed in the data, instead of relying on unrealistic assumptions about uncertainty. A soft measurement model using the Takagi-Sugeno (TS) Fuzzy Logic System (FLS) based on several input-output time series of PV plants is presented in this study. Chebyshev's inequality from probability theory and statistics is adopted to create the confidence interval-based response envelopes for these time series at each moment. An envelope-based measure of output uncertainty and a center-valued response forecasting model can be obtained by the proposed identification technique. PV datasets are employed to demonstrate the concept, which indicates the proposed soft measurement may outperform existing methods in terms of prediction accuracy. The average values of Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and the Correlation Coefficient (R) are only 0.0787, 0.1113, and 0.9979. The average values of the prediction interval coverage probability (PICP) and the prediction interval normalized average width (PINAW) are 0.9806 and 0.1051.

Mass Conservative Time-Series GAN for Synthetic Extreme Flood-Event Generation: Impact on Probabilistic Forecasting Models

The lack of data on flood events poses challenges in flood management. In this paper, we propose a novel approach to enhance flood-forecasting models by utilizing the capabilities of Generative Adversarial Networks (GANs) to generate synthetic flood events. We modified a time-series GAN by incorporating constraints related to mass conservation, energy balance, and hydraulic principles into the GAN model through appropriate regularization terms in the loss function and by using mass conservative LSTM in the generator and discriminator models. In this way, we can improve the realism and physical consistency of the generated extreme flood-event data. These constraints ensure that the synthetic flood-event data generated by the GAN adhere to fundamental hydrological principles and characteristics, enhancing the accuracy and reliability of flood-forecasting and risk-assessment applications. PCA and t-SNE are applied to provide valuable insights into the structure and distribution of the synthetic flood data, highlighting patterns, clusters, and relationships within the data. We aimed to use the generated synthetic data to supplement the original data and train probabilistic neural runoff model for forecasting multi-step ahead flood events. t-statistic was performed to compare the means of synthetic data generated by TimeGAN with the original data, and the results showed that the means of the two datasets were statistically significant at 95% level. The integration of time-series GAN-generated synthetic flood events with real data improved the robustness and accuracy of the autoencoder model, enabling more reliable predictions of extreme flood events. In the pilot study, the model trained on the augmented dataset with synthetic data from time-series GAN shows higher NSE and KGE scores of NSE = 0.838 and KGE = 0.908, compared to the NSE = 0.829 and KGE = 0.90 of the sixth hour ahead, indicating improved accuracy of 9.8% NSE in multistep-ahead predictions of extreme flood events compared to the model trained on the original data alone. The integration of synthetic training datasets in the probabilistic forecasting improves the model’s ability to achieve a reduced Prediction Interval Normalized Average Width (PINAW) for interval forecasting, yet this enhancement comes with a trade-off in the Prediction Interval Coverage Probability (PICP).

Can supervised deep learning architecture outperform autoencoders in building propensity score models for matching?

PurposePropensity score matching is vital in epidemiological studies using observational data, yet its estimates relies on correct model-specification. This study assesses supervised deep learning models and unsupervised autoencoders for propensity score estimation, comparing them with traditional methods for bias and variance accuracy in treatment effect estimations.MethodsUtilizing a plasmode simulation based on the Right Heart Catheterization dataset, under a variety of settings, we evaluated (1) a supervised deep learning architecture and (2) an unsupervised autoencoder, alongside two traditional methods: logistic regression and a spline-based method in estimating propensity scores for matching. Performance metrics included bias, standard errors, and coverage probability. The analysis was also extended to real-world data, with estimates compared to those obtained via a double robust approach.ResultsThe analysis revealed that supervised deep learning models outperformed unsupervised autoencoders in variance estimation while maintaining comparable levels of bias. These results were supported by analyses of real-world data, where the supervised model’s estimates closely matched those derived from conventional methods. Additionally, deep learning models performed well compared to traditional methods in settings where exposure was rare.ConclusionSupervised deep learning models hold promise in refining propensity score estimations in epidemiological research, offering nuanced confounder adjustment, especially in complex datasets. We endorse integrating supervised deep learning into epidemiological research and share reproducible codes for widespread use and methodological transparency.

Uniformly most accurate confidence intervals under weak restrictions

The natural and commonly used measure of accuracy for a confidence interval (CI) is its length, but it only applies to bounded CI's. More seriously, it is not an invariant measure, creating chaos on selecting CI's. Using the probability of false coverage as a finite and invariant measure of accuracy for a CI, we establish the uniformly most accurate (UMA) CI under weak restriction, which substantially improves the classical UMA unbiased CI for simplicity and optimality.

Fixed-[formula omitted] asymptotics for panel models with two-way clustering

This paper studies a cluster robust variance estimator proposed by Chiang, Hansen and Sasaki (2024) for linear panels. First, we show algebraically that this variance estimator (CHS estimator, hereafter) is a linear combination of three common variance estimators: the one-way unit cluster estimator, the “HAC of averages” estimator, and the “average of HACs” estimator. Based on this finding, we obtain a fixed-b asymptotic result for the CHS estimator and corresponding test statistics as the cross-section and time sample sizes jointly go to infinity. Furthermore, we propose two simple bias-corrected versions of the variance estimator and derive the fixed-b limits. In a simulation study, we find that the two bias-corrected variance estimators along with fixed-b critical values provide improvements in finite sample coverage probabilities. We illustrate the impact of bias-correction and use of the fixed-b critical values on inference in an empirical example on the relationship between industry profitability and market concentration.

The corrosion properties of ferritic stainless steel with varying Cr and Mo contents in the early stages of a simulated proton exchange membrane fuel cell environment investigated using experimental and joint calculation method HKD

The initial corrosion resistance of ferritic stainless steels with various Cr and Mo levels in simulated fuel cell environments is evaluated using the HKD calculation method and electrochemical tests. HKD utilizes high-throughput DFT, KMC, and AIMD statistical coverage calculations to study passivation film development and ion/molecule adsorption behaviors, like O2-, F-, H2O, and OH-. Super-high Cr-containing Mo steel shows reliable corrosion resistance due to their stable structure and effective passivation film integrity at various F- concentrations and pH values. Furthermore, the impact of F- competes with O2-, influencing film formation as pH lowers and concentration rises.

Change point detection in length-biased lognormal distribution

In this paper, we develop two procedures for identifying the dynamic trends in the parameters of length-biased lognormal distribution based on likelihood ratio and modified information criterion. These methods mainly consider the test of the existence of the change point and provide the maximum likelihood estimation of the change point when the change point exists. In addition, the asymptotic distribution of test statistic based on likelihood ratio is derived as an extreme value distribution, while the asymptotic distribution of test statistic based on modified information criterion is derived as a chi-square distribution. And the consistency of parameter estimation is proved. Simulations are conducted to study the performance of the proposed method in terms of power, coverage probabilities and average sizes of confidence sets. The proposed methods are applied to a real data to illustrate the detecting procedures.

Coverage Analysis for High-Speed Railway Communications with Narrow-Strip-Shaped Cells over Suzuki Fading Channels.

Unlike circular cell coverage in public land mobile communications, narrow-strip-shaped cell coverage should be considered in high-speed railway (HSR) communications. Moreover, for the coverage analysis in HSR communications, most works ignore the effect of small-scale fading, which results in an inaccurate coverage performance evaluation. In this paper, we focus on the coverage analysis for HSR communications with narrow-strip-shaped cells over the Suzuki fading channel, where the composite channel fading includes path loss, lognormal shadowing, and Rayleigh-distributed small-scale fading. Based on the channel model, we first analyze the statistical characteristic of the received signal-to-noise ratio. Then, we derive analytical expressions of the edge coverage probability (ECP) and the percentage of cell coverage area (CCA). To link the edge coverage performance and the average coverage performance of a cell, we express the percentage of CCA as a summation of the ECP and a positive increment. As special cases, we also obtain the coverage performance expressions for the systems without small-scale fading. Through Monte Carlo simulations, the accuracy of the derived expressions is verified. Numerical results also show that the small-scale fading has a strong effect on coverage performance and cannot be ignored. In addition, the effects of key parameters are also discussed.

Evaluating energy harvesting UAV-NOMA network with random user pairing in the finite blocklength regime

This study proposes an integrated system that combines energy harvesting (EH) enabled unmanned aerial vehicles (UAVs) with non-orthogonal multiple access (NOMA) to enhance communication system performance within a cellular network. Addressing the limitations of existing analyses that often assume an infinite blocklength scenario, we explore EH-enabled UAV-NOMA systems within a cellular framework under a finite blocklength (FBL) scenario. The study investigates the complex interactions and advantages resulting from the integration of EH, NOMA, and UAV technologies, aiming to assess whether EH can sustain communication within this framework. The network model considers base stations (BSs), UAVs, and terrestrial devices distributed with independent Poisson point processes (PPPs) over a large area. In this network, BSs employ NOMA to serve cell center devices directly, while cell edge devices, which are nor in direct contact with BS, are served via simultaneous wireless information and power transfer (SWIPT) enabled UAVs. The study derives metrics including joint harvesting and decoding probability for a randomly selected UAV, coverage probability (CP) for cell devices, and end-to-end block error rate (BLER) probabilities for typical device pairs. The findings demonstrate that the proposed scheme effectively supplies all the necessary transmit power for communication purposes through EH, achieving reasonable reliability. Additionally, the study highlights the importance of considering a combination of blocklengths from different phases to achieve optimal performance, rather than solely relying on an increment in blocklength. Finally, the effects of parameter variations on network performance are examined.

ZIBGLMM: Zero-Inflated Bivariate Generalized Linear Mixed Model for Meta-Analysis with Double-Zero-Event Studies.

Double-zero-event studies (DZS) pose a challenge for accurately estimating the overall treatment effect in meta-analysis. Current approaches, such as continuity correction or omission of DZS, are commonly employed, yet these ad hoc methods can yield biased conclusions. Although the standard bivariate generalized linear mixed model can accommodate DZS, it fails to address the potential systemic differences between DZS and other studies. In this paper, we propose a zero-inflated bivariate generalized linear mixed model (ZIBGLMM) to tackle this issue. This two-component finite mixture model includes zero-inflation for a subpopulation with negligible or extremely low risk. We develop both frequentist and Bayesian versions of ZIBGLMM and examine its performance in estimating risk ratios (RRs) against the bivariate generalized linear mixed model and conventional two-stage meta-analysis that excludes DZS. Through extensive simulation studies and real-world meta-analysis case studies, we demonstrate that ZIBGLMM outperforms the bivariate generalized linear mixed model and conventional two-stage meta-analysis that excludes DZS in estimating the true effect size with substantially less bias and comparable coverage probability.

Interference-aware scheme to improve distributed caching in cellular networks via D2D underlay communications

Underlay Device-to-Device (D2D) communications is a promising networking technology intended to boost thespectral efficiency of future cellular networks, including 5G and beyond. When used for distributed caching, where cellulardevices store popular files for direct exchange later with other devices away from the cellular infrastructure, the technologybears more fruits such as enhancing throughput, reducing latency and offloading the infrastructure. However, due to theirnon-orthogonality, underlay D2D communications can result in excessive interference to the cellular user. To avoid thisproblem, the present article proposes a scheme with two interference-reduction elements: a guard zone intended to allowD2D communications only for devices far enough from the base station (BS), and a pairing strategy intended to allow D2Dpairing for only devices that are close enough to each other. We assess the performance of the scheme using a stochasticgeometry (SG) model, through which we characterize the coverage probability of the cellular user. This probability is aprincipal indicator of maintaining the quality of service (QoS) of the cellular user and of enabling successful caching for theD2D user. We introduce in the process a novel empirical technique which, given a desired level of interference, identifiesan upper bound for the distance between two devices to be paired without exceeding that level. We finally validate theanalytical findings obtained from the model by intensive simulation to ensure the correctness of both the model and thescheme performance. A salient feature of the scheme is that it requires for its implementation no software or hardwaremodification in the device

Coverage Probability Analysis for Uplink/Downlink Decoupled Access in Heterogeneous Cellular Networks

Coverage Probability Analysis for Uplink/Downlink Decoupled Access in Heterogeneous Cellular Networks

GFR measurement in patients with CKD: Performance and feasibility of simplified iohexol plasma clearance techniques.

Implementing shortened one-compartment iohexol plasma clearance models for GFR measurement is crucial since the gold standard inulin renal clearance technique and the reference two-compartment, 10-hour, 16-samplings iohexol plasma clearance method are clinically unfeasible. Inulin may precipitate anaphylactic shock. Four-hour and 8-hour one-compartment iohexol plasma clearance models with Bröchner-Mortensen correction provide accurate GFR measurements in patients with estimated GFR (eGFR) > or ≤40 mL/min/1.73m2, respectively. We compared the performance of the simplified 5-hour, 4-samplings, two-compartment population pharmacokinetic model (popPK) with the performance of the reference two-compartment 10-hour iohexol method in 16 patients with GFR 15.2 to 56.5 mL/min/1.73 m2. We also compared the performance of shortened (5, 6 and 7-hour) one-compartment models with the performance of the standard 8-hour one-compartment model in 101 patients with eGFR ≤40 mL/min/1.73 m2. The performance of popPK and shortened methods versus reference methods was evaluated by total deviation index (TDI), concordance correlation coefficient (CCC) and coverage probability (CP). TDI <10%, CCC ≥0.9 and CP >90% indicated adequate performance. TDI, CCC and CP of popPK were 11.11%, 0.809 and 54.10%, respectively. All shortened, one-compartment models overestimated the GFR (p <0.0001 for all) as compared to the 8-hour model. TDI, CCC and CP were 7.02%, 0.815, and 75.80% for the 7-hour model, 7.26%, 0.803, and 74.20% for the 6-hour model, and 8.85%, 0.729 and 64.70% for the 5-hour model. The agreement of popPK model was comparable to that obtained with the Chronic-Kidney-Disease-Collaboration-Epidemiology (CKD-Epi) and the Modification-of-Diet-in-Renal-Disease (MDRD) serum-creatinine based equations for GFR estimation. PopPK model is remarkably unreliable for GFR measurement in stage III-IV CKD patients. In patients with eGFR ≤40 mL/min/1.73m2, shortened one-compartment models, in particular the 5-hour model, are less performant than the reference 8-hour model. For accurate GFR measurements, the iohexol plasma clearance should be measured with appropriate protocols. Over-simplified procedures should be avoided.

Different Statistical Inference Algorithms for the New Pareto Distribution Based on Type-II Progressively Censored Competing Risk Data with Applications

In this research, the statistical inference of unknown lifetime parameters is proposed in the presence of independent competing risks using a progressive Type-II censored dataset. The lifetime distribution associated with a failure mode is assumed to follow the new Pareto distribution, with consideration given to two distinct competing failure reasons. Maximum likelihood estimators (MLEs) for the unknown model parameters, as well as reliability and hazard functions, are derived, noting that they are not expressible in closed form. The Newton–Raphson, expectation maximization (EM), and stochastic expectation maximization (SEM) methods are employed to generate maximum likelihood (ML) estimations. Approximate confidence intervals for the unknown parameters, reliability, and hazard rate functions are constructed using the normal approximation of the MLEs and the normal approximation of the log-transformed MLEs. Additionally, the missing information principle is utilized to derive the closed form of the Fisher information matrix, which, in turn, is used with the delta approach to calculate confidence intervals for reliability and hazards. Bayes estimators are derived under both symmetric and asymmetric loss functions, with informative and non-informative priors considered, including independent gamma distributions for informative priors. The Monte Carlo Markov Chain sampling approach is employed to obtain the highest posterior density credible intervals and Bayesian point estimates for unknown parameters and reliability characteristics. A Monte Carlo simulation is conducted to assess the effectiveness of the proposed techniques, with the performances of the Bayes and maximum likelihood estimations examined using average values and mean squared errors as benchmarks. Interval estimations are compared in terms of average lengths and coverage probabilities. Real datasets are considered and examined for each topic to provide illustrative examples.

Deep DNAshape webserver: prediction and real-time visualization of DNA shape considering extended k-mers.

Sequence-dependent DNA shape plays an important role in understanding protein-DNA binding mechanisms. High-throughput prediction of DNA shape features has become a valuable tool in the field of protein-DNA recognition, transcription factor-DNA binding specificity, and gene regulation. However, our widely used webserver, DNAshape, relies on statistically summarized pentamer query tables to query DNA shape features. These query tables do not consider flanking regions longer than two base pairs, and acquiring a query table for hexamers or higher-order k-mers is currently still unrealistic due to limitations in achieving sufficient statistical coverage in molecular simulations or structural biology experiments. A recent deep-learning method, Deep DNAshape, can predict DNA shape features at the core of a DNA fragment considering flanking regions of up to seven base pairs, trained on limited simulation data. However, Deep DNAshape is rather complicated to install, and it must run locally compared to the pentamer-based DNAshape webserver, creating a barrier for users. Here, we present the Deep DNAshape webserver, which has the benefits of both methods while being accurate, fast, and accessible to all users. Additional improvements of the webserver include the detection of user input in real time, the ability of interactive visualization tools and different modes of analyses. URL: https://deepdnashape.usc.edu.

Inferior vena cava ultrasound to predict hypotension after general anesthesia induction: a systematic review and meta-analysis of observational studies.

Hypotension after induction of general anesthesia is common and is associated with significant adverse events. Identification of patients at high risk can inform the use of preoperative mitigation strategies. We conducted a systematic review and meta-analysis to assess the diagnostic accuracy of the inferior vena cava collapsibility index (IVC-CI) and maximal diameter (dIVCmax) in predicting postinduction hypotension and to identify their predictive performance across different threshold ranges. We searched MEDLINE, PubMed®, and Embase from inception to March 2023 for prospective observational studies exploring the performance of IVC-CI and dIVCmax in predicting postinduction hypotension in adults presenting for elective surgery under general anesthesia. We excluded studies reporting on IVC parameters predicting postinduction hypotension in the obstetric patient population or exclusively in patients with obesity. Trials screening and data extraction were conducted independently. We performed meta-analyses to identify the performance of IVC parameters in predicting postinduction hypotension, followed by subgroup analyses that sought the IVC-CI range with the highest hierarchical summary receiver-operating characteristic area under the curve (HSROC-AUC). We used a bivariate random effects model to calculate summary estimates. We evaluated study quality using Newcastle-Ottawa scores and certainty of evidence using the GRADE framework. We included 14 studies involving 1,166 patients. Pooled sensitivity and specificity of the IVC-CI to predict postinduction hypotension was 0.68 (95% confidence interval [CI], 0.55 to 0.79; coverage probability, 0.91) and 0.78 (95% CI, 0.69 to 0.85; coverage probability, 0.9), respectively, with an HSROC-AUC of 0.80 (95% CI, 0.68 to 0.85, high quality of evidence). An IVC-CI threshold range of 40-45% had an HSROC-AUC of 0.86 (95% CI, 0.69 to 0.93, high quality of evidence). Preoperative IVC-CI is a strong predictor of postinduction hypotension. We recommend that future studies use an IVC-CI threshold of 40-45% (low certainty of evidence). Future studies are needed to establish whether ultrasound-guided preoperative optimization improves outcomes in high-risk patients. PROSPERO ( CRD42022316140 ); first submitted 10 March 2022.

Latent classification model for censored longitudinal binary outcome.

Latent classification model is a class of statistical methods for identifying unobserved class membership among the study samples using some observed data. In this study, we proposed a latent classification model that takes a censored longitudinal binary outcome variable and uses its changing pattern over time to predict individuals' latent class membership. Assuming the time-dependent outcome variables follow a continuous-time Markov chain, the proposed method has two primary goals: (1) estimate the distribution of the latent classes and predict individuals' class membership, and (2) estimate the class-specific transition rates and rate ratios. To assess the model's performance, we conducted a simulation study and verified that our algorithm produces accurate model estimates (ie, small bias) with reasonable confidence intervals (ie, achieving approximately 95% coverage probability). Furthermore, we compared our model to four other existing latent class models and demonstrated that our approach yields higher prediction accuracies for latent classes. We applied our proposed method to analyze the COVID-19 data in Houston, Texas, US collected between January first 2021 and December 31st 2021. Early reports on the COVID-19 pandemic showed that the severity of a SARS-CoV-2 infection tends to vary greatly by cases. We found that while demographic characteristics explain some of the differences in individuals' experience with COVID-19, some unaccounted-for latent variables were associated with the disease.