Partially ordered mixed hidden Markov model for the disablement process of older adults

Order selection for heterogeneous semiparametric hidden Markov models

Hidden Markov models (HMMs), which can characterize dynamic heterogeneity, are valuable tools for analyzing longitudinal data. The order of HMMs (ie, the number of hidden states) is typically assumed to be known or predetermined by some model selection criterion in conventional analysis. As prior information about the order frequently lacks, pairwise comparisons under criterion-based methods become computationally expensive with the model space growing. A few studies have conducted order selection and parameter estimation simultaneously, but they only considered homogeneous parametric instances. This study proposes a Bayesian double penalization (BDP) procedure for simultaneous order selection and parameter estimation of heterogeneous semiparametric HMMs. To overcome the difficulties in updating the order, we create a brand-new Markov chain Monte Carlo algorithm coupled with an effective adjust-bound reversible jump strategy. Simulation results reveal that the proposed BDP procedure performs well in estimation and works noticeably better than the conventional criterion-based approaches. Application of the suggested method to the Alzheimer's Disease Neuroimaging Initiative research further supports its usefulness.

Septic Shock: Phenotypes and Outcomes

INTRODUCTION

Sepsis is a heterogeneous syndrome that results in life-threatening organ dysfunction. Our goal was to determine the relevant variables and patient phenotypes to use in predicting sepsis outcomes.

METHODS

We performed an ancillary study concerning 119 patients with septic shock at intensive care unit (ICU) admittance (T0). We defined clinical worsening as having an increased sequential organ failure assessment (SOFA) score of ≥ 1, 48 h after admission (ΔSOFA ≥ 1). We performed univariate and multivariate analyses based on the 28-day mortality rate and ΔSOFA ≥ 1 and determined three patient phenotypes: safe, intermediate and unsafe. The persistence of the intermediate and unsafe phenotypes after T0 was defined as a poor outcome.

RESULTS

At T0, the multivariate analysis showed two variables associated with 28-day mortality rate: norepinephrine dose and serum lactate concentration. Regarding ΔSOFA ≥ 1, we identified three variables at T0: norepinephrine dose, lactate concentration and venous-to-arterial carbon dioxide difference (P(v-a)CO₂). At T0, the three phenotypes (safe, intermediate and unsafe) were found in 28 (24%), 70 (59%) and 21 (18%) patients, respectively. We thus suggested using an algorithm featuring norepinephrine dose, lactate concentration and P(v-a)CO₂ to predict patient outcomes and obtained an area under the curve (AUC) of 74% (63-85%).

CONCLUSION

Our findings highlight the fact that identifying relevant variables and phenotypes may help physicians predict patient outcomes.

Predicting Future Mobility Limitation in Older Adults: A Machine Learning Analysis of Health ABC Study Data

BACKGROUND

Mobility limitation in older adults is common and associated with poor health outcomes and loss of independence. Identification of at-risk individuals remains challenging because of time-consuming clinical assessments and limitations of statistical models for dynamic outcomes over time. Therefore, we aimed to develop machine learning models for predicting future mobility limitation in older adults using repeated measures data.

METHODS

We used annual assessments over 9 years of follow-up from the Health, Aging, and Body Composition study to model mobility limitation, defined as self-report of any difficulty walking a quarter mile or climbing 10 steps. We considered 46 predictors, including demographics, lifestyle, chronic conditions, and physical function. With a split sample approach, we developed mixed models (generalized linear and Binary Mixed Model forest) using (a) all 46 predictors, (b) a variable selection algorithm, and (c) the top 5 most important predictors. Age was included in all models. Performance was evaluated using area under the receiver operating curve in 2 internal validation data sets.

RESULTS

Area under the receiver operating curve ranged from 0.80 to 0.84 for the models. The most important predictors of mobility limitation were ease of getting up from a chair, gait speed, self-reported health status, body mass index, and depression.

CONCLUSIONS

Machine learning models using repeated measures had good performance for identifying older adults at risk of developing mobility limitation. Future studies should evaluate the utility and efficiency of the prediction models as a tool in clinical settings for identifying at-risk older adults who may benefit from interventions aimed to prevent or delay mobility limitation.

Tolerability of bevacizumab and chemotherapy in a phase 3 clinical trial with human epidermal growth factor receptor 2-negative breast cancer: A trajectory analysis of adverse events

BACKGROUND

E5103 was a study designed to evaluate the efficacy and safety of bevacizumab. It was a negative trial for the end points of invasive disease-free survival and overall survival. The current work examines the tolerability of bevacizumab and other medication exposures with respect to clinical outcomes and patient-reported outcomes (PROs).

METHODS

Adverse events (AEs) collected from the Common Terminology Criteria for Adverse Events were summarized to form an AE profile at each treatment cycle. All-grade and high-grade events were separately analyzed. The change in the AE profile over the treatment cycle was delineated as distinct AE trajectory clusters. AE-related and any-reason early treatment discontinuations were treated as clinical outcome measures. PROs were measured with the Functional Assessment of Cancer Therapy-Breast + Lymphedema. The relationships between the AE trajectory and early treatment discontinuation as well as PROs were analyzed.

RESULTS

More than half of all AEs (57.5%) were low-grade. A cluster of patients with broad and mixed AE (all-grade) trajectory grades was significantly associated with any-reason early treatment discontinuation (odds ratio [OR], 2.87; P = .01) as well as AE-related discontinuation (OR, 4.14; P = .001). This cluster had the highest count of all-grade AEs per cycle in comparison with other clusters. Another cluster of patients with primary neuropathic AEs in their trajectories had poorer physical well-being in comparison with a trajectory of no or few AEs (P < .01). A high-grade AE trajectory did not predict discontinuations.

CONCLUSIONS

A sustained and cumulative burden of across-the-board toxicities, which were not necessarily all recognized as high-grade AEs, contributed to early treatment discontinuation. Patients with neuropathic all-grade AEs may require additional attention for preventing deterioration in their physical well-being.

Model-based two-way clustering of second-level units in ordinal multilevel latent Markov models

In this paper, an ordinal multilevel latent Markov model based on separate random effects is proposed. In detail, two distinct second-level discrete effects are considered in the model, one affecting the initial probability vector and the other affecting the transition probability matrix of the first-level ordinal latent Markov process. To model these separate effects, we consider a bi-dimensional mixture specification that allows to avoid unverifiable assumptions on the random effect distribution and to derive a two-way clustering of second-level units. Starting from a general model where the two random effects are dependent, we also obtain the independence model as a special case. The proposal is applied to data on the physical health status of a sample of elderly residents grouped into nursing homes. A simulation study assessing the performance of the proposal is also included.

Generalized linear mixed hidden semi-Markov models in longitudinal settings: A Bayesian approach

Hidden Markov and semi-Markov models (H(S)MMs) constitute useful tools for modeling observations subject to certain dependency structures. The hidden states render these models very flexible and allow them to capture many different types of latent patterns and dynamics present in the data. This has led to the increased popularity of these models, which have been applied to a variety of problems in various domains and settings, including longitudinal data. In many longitudinal studies, the response variable is categorical or count-type. Generalized linear mixed models (GLMMs) can be used to analyze a wide range of variables, including categorical and count. The present study proposes a model that combines HSMMs with GLMMs, leading to generalized linear mixed hidden semi-Markov models (GLM-HSMMs). These models can account for time-varying unobserved heterogeneity and handle different response types. Parameter estimation is achieved using a Monte Carlo Newton-Raphson (MCNR)-like algorithm. In our proposed model, the distribution of the random effects depends on hidden states. We illustrate the applicability of GLM-HSMMs with an example in the field of occupational health, where the response variable consists of count values. Furthermore, we assess the performance of our MCNR-like algorithm through a simulation study.

Longitudinal partially ordered data analysis for preclinical sarcopenia

Sarcopenia is a geriatric syndrome characterized by significant loss of muscle mass. Based on a commonly used definition of the condition that involves three measurements, different subclinical and clinical states of sarcopenia are formed. These states constitute a partially ordered set (poset). This article focuses on the analysis of longitudinal poset in the context of sarcopenia. We propose an extension of the generalized linear mixed model and a recoding scheme for poset analysis such that two submodels-one for ordered categories and one for nominal categories-that include common random effects can be jointly estimated. The new poset model postulates random effects conceptualized as latent variables that represent an underlying construct of interest, that is, susceptibility to sarcopenia over time. We demonstrate how information can be gleaned from nominal sarcopenic states for strengthening statistical inference on a person's susceptibility to sarcopenia.

A non-compensatory analysis of quality of life in breast cancer survivors using multivariate hidden Markov modeling

PURPOSE

Health-related quality of life (HRQoL) is a multidimensional concept comprising multiple domains such as physical, emotional, and social well-being. Many analyses use a sum score to represent the construct. However, this approach implies that gain in one domain can compensate for a deficit in another, and thus such analyses may not capture HRQoL profiles. Additionally, within-individual change over time, such as improvement in one domain but deterioration in another, may not be detected. The objectives of this research are to demonstrate the utility of a non-compensatory approach by (1) evaluating this approach applied to HRQoL data, and (2) comparing the approach to a compensatory method.

METHODS

Data from a sample of 653 breast cancer survivors (BCS) provided five measurement time points over 18 months. We analyzed the scores from five domains on the FACT-B questionnaire (physical, functional, social, and emotional well-being and breast cancer-related concerns) using the multivariate hidden Markov model (MHMM), a non-compensatory approach that identifies different HRQoL states and associated BCS subgroups and their trajectories.

RESULTS

The MHMM delineated six states. States 1 and 2 had low well-being scores across all domains, with state 2 slightly better than state 1. States 3 and 4 had similar overall HRQoL scores, but different profiles with compensation occurring across the domains of both physical and social well-being. States 5 and 6 had almost identical overall scores with compensation occurring between the domains of both social and emotional well-being. Over time, states 3-6 mostly "communicated" with each other (with moderate probabilities of transitioning between states). Compensation across domains could mask subtle changes occurring in BCS. We found that a trend analysis using both compensatory and non-compensatory approaches showed improvement in the HRQoL in BCS over time.

CONCLUSION

The non-compensatory analysis of FACT-B shows differential profiles and trajectories in the HRQoL of BCS not captured by the sum score or one-domain-at-a-time approach.

Bayesian quantile nonhomogeneous hidden Markov models

Hidden Markov models are useful in simultaneously analyzing a longitudinal observation process and its dynamic transition. Existing hidden Markov models focus on mean regression for the longitudinal response. However, the tails of the response distribution are as important as the center in many substantive studies. We propose a quantile hidden Markov model to provide a systematic method to examine the entire conditional distribution of the response given the hidden state and potential covariates. Instead of considering homogeneous hidden Markov models, which assume that the probabilities of between-state transitions are independent of subject- and time-specific characteristics, we allow the transition probabilities to depend on exogenous covariates, thereby yielding nonhomogeneous Markov chains and making the proposed model more flexible than its homogeneous counterpart. We develop a Bayesian approach coupled with efficient Markov chain Monte Carlo methods for statistical inference. Simulations are conducted to assess the empirical performance of the proposed method. The proposed methodology is applied to a cocaine use study to provide new insights into the prevention of cocaine use.

Identifying Social Network Conditions that Facilitate Sedentary Behavior Change: The Benefit of Being a "Bridge" in a Group-based Intervention

Using data from one of the first trials to try to leverage social networks as a mechanism for obesity intervention, we examined which social network conditions amplified behavior change. Data were collected as part of a community-based healthy lifestyle intervention in Nashville, USA, between June 2014 and July 2017. Adults randomized to the intervention arm were assigned to a small group of 10 participants that met in person for 12 weekly sessions. Intervention small group social networks were measured three times; sedentary behavior was measured by accelerometry at baseline and 12 months. Multivariate hidden Markov models classified people into distinct social network trajectories over time, based on the structure of the emergent network and where the individual was embedded. A multilevel regression analysis assessed the relationship between network trajectory and sedentary behavior (N = 261). Being a person that connected clusters of intervention participants at any point during the intervention predicted an average reduction of 31.3 min/day of sedentary behavior at 12 months, versus being isolated [95% CI: (−61.4, −1.07), p = 0.04]. Certain social network conditions may make it easier to reduce adult sedentary behavior in group-based interventions. While further research will be necessary to establish causality, the implications for intervention design are discussed.

Bayesian hidden Markov models for delineating the pathology of Alzheimer's disease

Alzheimer's disease is a firmly incurable and progressive disease. The pathology of Alzheimer's disease usually evolves from cognitive normal, to mild cognitive impairment, to Alzheimer's disease. The aim of this paper is to develop a Bayesian hidden Markov model to characterize disease pathology, identify hidden states corresponding to the diagnosed stages of cognitive decline, and examine the dynamic changes of potential risk factors associated with the cognitive normal-mild cognitive impairment-Alzheimer's disease transition. The hidden Markov model framework consists of two major components. The first one is a state-dependent semiparametric regression for delineating the complex associations between clinical outcomes of interest and a set of prognostic biomarkers across neurodegenerative states. The second one is a parametric transition model, while accounting for potential covariate effects on the cross-state transition. The inter-individual and inter-process differences are taken into account via correlated random effects in both components. Based on the Alzheimer's Disease Neuroimaging Initiative data set, we are able to identify four states of Alzheimer's disease pathology, corresponding to common diagnosed cognitive decline stages, including cognitive normal, early mild cognitive impairment, late mild cognitive impairment, and Alzheimer's disease and examine the effects of hippocampus, age, gender, and APOE- ε 4 on degeneration of cognitive function across the four cognitive states.

Impact of Intensive Lifestyle Intervention on Disability-Free Life Expectancy: The Look AHEAD Study

OBJECTIVE

The impact of weight loss intervention on disability-free life expectancy in adults with diabetes is unknown. We examined the impact of a long-term weight loss intervention on years spent with and without physical disability.

RESEARCH DESIGN AND METHODS

Overweight or obese adults with type 2 diabetes age 45-76 years (n = 5,145) were randomly assigned to a 10-year intensive lifestyle intervention (ILI) or diabetes support and education (DSE). Physical function was assessed annually for 12 years using the 36-Item Short Form Health Survey. Annual incidence of physical disability, mortality, and disability remission were incorporated into a Markov model to quantify years of life spent active and physically disabled.

RESULTS

Physical disability incidence was lower in the ILI group (6.0% per year) than in the DSE group (6.8% per year) (incidence rate ratio 0.88 [95% CI 0.81-0.96]), whereas rates of disability remission and mortality did not differ between groups. ILI participants had a significant delay in moderate or severe disability onset and an increase in number of nondisabled years (P < 0.05) compared with DSE participants. For a 60-year-old, this effect translates to 0.9 more disability-free years (12.0 years [95% CI 11.5-12.4] vs. 11.1 years [95% CI 10.6-11.7]) but no difference in total years of life. In stratified analyses, ILI increased disability-free years of life in women and participants without cardiovascular disease (CVD) but not in men or participants with CVD.

CONCLUSIONS

Long-term lifestyle interventions among overweight or obese adults with type 2 diabetes may reduce long-term disability, leading to an effect on disability-free life expectancy but not on total life expectancy.

Bayesian Diagnostics of Hidden Markov Structural Equation Models with Missing Data

Cocaine is a type of drug that functions to increase the availability of the neurotransmitter dopamine in the brain. However, cocaine dependence or abuse is highly related to an increased risk of psychiatric disorders and deficits in cognitive performance, attention, and decision-making abilities. Given the chronic and persistent features of drug addiction, the progression of abstaining from cocaine often evolves across several states, such as addiction to, moderate dependence on, and swearing off cocaine. Hidden Markov models (HMMs) are well suited to the characterization of longitudinal data in terms of a set of unobservable states, and have increasingly been used to uncover the dynamic heterogeneity in progressive diseases or activities. However, the existence of outliers or influential points may misidentify the hidden states and distort the associated inference. In this study, we develop a Bayesian local influence procedure for HMMs with latent variables in the presence of missing data. The proposed model enables us to investigate the dynamic heterogeneity of multivariate longitudinal data, reveal how the interrelationships among latent variables change from one state to another, and simultaneously conduct statistical diagnosis for the given data, model assumptions, and prior inputs. We apply the proposed procedure to analyze a dataset collected by the UCLA center for advancing longitudinal drug abuse research. Several outliers or influential points that seriously influence estimation results are identified and removed. The proposed procedure also discovers the effects of treatment and individuals' psychological problems on cocaine use behavior and delineates their dynamic changes across the cocaine-addiction states.

Estimating recurrence and incidence of preterm birth subject to measurement error in gestational age: A hidden Markov modeling approach

Prediction of preterm birth as well as characterizing the etiological factors affecting both the recurrence and incidence of preterm birth (defined as gestational age at birth ≤ 37 wk) are important problems in obstetrics. The National Institute of Child Health and Human Development (NICHD) consecutive pregnancy study recently examined this question by collecting data on a cohort of women with at least 2 pregnancies over a fixed time interval. Unfortunately, measurement error due to the dating of conception may induce sizable error in computing gestational age at birth. This article proposes a flexible approach that accounts for measurement error in gestational age when making inference. The proposed approach is a hidden Markov model that accounts for measurement error in gestational age by exploiting the relationship between gestational age at birth and birth weight. We initially model the measurement error as being normally distributed, followed by a mixture of normals that has been proposed on the basis of biological considerations. We examine the asymptotic bias of the proposed approach when measurement error is ignored and also compare the efficiency of this approach to a simpler hidden Markov model formulation where only gestational age and not birth weight is incorporated. The proposed model is compared with alternative models for estimating important covariate effects on the risk of subsequent preterm birth using a unique set of data from the NICHD consecutive pregnancy study.

Longitudinal profiles of eudaimonic well-being in Asian American adolescents

OBJECTIVE

The current study explores whether a well-known model (i.e., Ryff's, 1989, conceptualization of psychological functioning) can be used to examine patterns of eudaimonic well-being among Asian Americans, who are rarely the focus of systematic investigations in positive psychology.

METHOD

Hidden Markov modeling, a form of latent transition analysis, was used to analyze longitudinal data from adolescents (N = 180; 49% female; 75% U.S.-born).

RESULTS

After establishing measurement validity, analyses revealed 4 profiles of well-being: Flourishing (consistently high on all well-being dimensions), Functioning (consistently moderate), Hindered (consistently low), and Self-Driven Success (high on most dimensions, but moderate levels of positive relationships). The Functioning profile was the most prevalent, followed by relatively even distributions of the remaining profiles. Profiles substantially shifted from year to year, with the Functioning and Hindered groups exhibiting the most stability. Profiles reflecting more positive well-being (i.e., Flourishing, Self-Driven) were associated with ethnic and American centrality and regard, and interactive effects suggest compounding benefits of these identities.

CONCLUSIONS

Psychological models of well-being appear malleable, and cultural identity can contribute to such fluctuations. Results also support the utility of a profile approach to continue examining qualities of positive well-being among Asian American youth. (PsycINFO Database Record

Child Feeding Style and Dietary Outcomes in a Cohort of Latino Farmworker Families

BACKGROUND

The high level of obesity in Latino children, especially in farmworker families, may be partly attributed to feeding styles of parents. Feeding styles used in Latino farmworker families have not been well characterized.

OBJECTIVE

This study sought to identify and describe feeding styles used by mothers in farmworker families with 2.5- to 3.5-year-old children, describe how styles change over time, and characterize the relationship of feeding styles to dietary outcomes and measures of overweight and obesity.

DESIGN

This was a longitudinal cohort study, with families participating for a 2-year period; surveys were administered to mothers with varying frequency depending on the instrument, and dietary measurements were collected at baseline and 12 and 24 months.

PARTICIPANTS/SETTING

Eligible participants were self-identified Latino women with a co-resident child aged 2.5 to 3.5 years old and at least one household member engaged in farm work during the previous year. The sample included 248 farmworker families enrolled between 2011 and 2012 in the Niños Sanos study, a longitudinal investigation of Latino mothers and their young children in rural North Carolina. Eleven families provided incomplete dietary data, so the analysis included 237 families. Fifteen families were lost to follow-up and 12 withdrew during the course of the study.

MAIN OUTCOME MEASURES

Feeding style was assessed using items from the Caregiver's Feeding Style Questionnaire, selected dietary components were assessed using the Revised Children's Diet Quality Index, and weight outcomes were determined using body mass index-for-age percentile. Performance on the Caregiver's Feeding Style Questionnaire items was used to assign mothers to one of four feeding style states.

STATISTICAL ANALYSES PERFORMED

Exploratory factor analysis was conducted on baseline data to verify the replicability of the factor structure of the instrument Caregiver's Feeding Style Questionnaire. Hidden Markov Model analysis was used to delineate different subtypes of feeding style. Multivariable mixed-effects regression models were used to assess the impact of feeding style on selected dietary components, energy intake, and body mass index-for-age percentile.

RESULTS

Four distinct states emerged from the Hidden Markov Model: low parent-centered (PC)/moderate child-centered (CC) feeding style (28% at baseline), high PC/CC without physical control (24%), high PC/CC (26%), and moderate PC/CC (22%). The low PC/moderate CC state increased in prevalence over time. Compared to high PC/CC, the low PC/moderate CC state was associated with greater intake of added sugars (P<0.01), lower intake of whole grains and vegetables (P<0.01), and lower overall diet quality (P<0.05). Children in low PC/moderate CC also had higher mean body mass index percentiles (76.2 percentile vs 66.7 percentile in high PC/CC; P<0.001).

CONCLUSIONS

High PC feeding along with high CC feeding is associated with improved diet quality and weight outcomes for children in the study.

Longitudinal Examination of Symptom Profiles Among Breast Cancer Survivors

CONTEXT

Identification of cancer patients with similar symptom profiles may facilitate targeted symptom management.

OBJECTIVES

To identify subgroups of breast cancer survivors based on differential experience of symptoms, examine change in subgroup membership over time, and identify relevant characteristics and quality of life (QOL) among subgroups.

METHODS

Secondary analyses of data from 653 breast cancer survivors recruited within eight months of diagnosis who completed questionnaires at five time points. Hidden Markov modeling was used to 1) formulate symptom profiles based on prevalence and severity of eight symptoms commonly associated with breast cancer and 2) estimate probabilities of changing subgroup membership over 18 months of follow-up. Ordinal repeated measures were used to 3) identify patient characteristics related to subgroup membership and 4) evaluate the relationship between symptom subgroup and QOL.

RESULTS

A seven-subgroup model provided the best fit: 1) low symptom burden, 2) mild fatigue, 3) mild fatigue and mild pain, 4) moderate fatigue and moderate pain, 5) moderate fatigue and moderate psychological, 6) moderate fatigue, mild pain, mild psychological, and 7) high symptom burden. Seventy percent of survivors remained in the same subgroup over time. In multivariable analyses, chemotherapy and greater illness intrusiveness were significantly related to greater symptom burden, while not being married or partnered, no difficulty paying for basics, and greater social support were protective. Higher symptom burden was associated with lower QOL. Survivors who reported psychological symptoms had significantly lower QOL than did survivors with pain symptoms.

CONCLUSION

Cancer survivors can be differentiated by their symptom profiles.

Reversible States of Physical and/or Cognitive Dysfunction: A 9-Year Longitudinal Study

OBJECTIVES

To determine 1) age-adjusted transition probabilities to worsening physical/cognitive function states, reversal to normal cognition/physical function, or maintenance of normal state; 2) whether these transitions are modulated by sex, BMI, education, hypertension (HTN), health status, or APOE4; 3) whether worsening gait speed preceded cognition change, or vice versa.

DESIGN

Analysis of 9-year prospective cohort data from the New Mexico Aging Process Study.

SETTING

Healthy independent-living adults.

PARTICIPANTS

60+ years of age (n= 598).

MEASUREMENTS

Gait speed, cognitive function (3MSE score), APOE4, HTN, BMI, education, health status.

RESULTS

Over 9 years, 2129 one-year transitions were observed. 32.6% stayed in the same state, while gait speed and cognitive function (3MSE scores) improved for 38% and 43% of participants per year, respectively. Transitions to improved function decreased with age (P< 0.001), APOE4 status (P=0.02), BMI (P=0.009), and health status (P=0.009). Transitions to worse function were significantly increased for the same factors (all P<0.05). Times to lower gait speed and cognitive function did not precede each other (P=0.91).

CONCLUSIONS

Transitions in gait speed and cognition were mutable with substantial likelihood of transition to improvement in physical and cognitive function even in oldest-old, which may have clinical implications for treatment interventions.

Analysis of Multiple Partially Ordered Responses to Belief Items with Don't Know Option

Understanding beliefs, values, and preferences of patients is a tenet of contemporary health sciences. This application was motivated by the analysis of multiple partially ordered set (poset) responses from an inventory on layman beliefs about diabetes. The partially ordered set arises because of two features in the data-first, the response options contain a Don't Know (DK) option, and second, there were two consecutive occasions of measurement. As predicted by the common sense model of illness, beliefs about diabetes were not necessarily stable across the two measurement occasions. Instead of analyzing the two occasions separately, we studied the joint responses across the occasions as a poset response. Few analytic methods exist for data structures other than ordered or nominal categories. Poset responses are routinely collapsed and then analyzed as either rank ordered or nominal data, leading to the loss of nuanced information that might be present within poset categories. In this paper we developed a general class of item response models for analyzing the poset data collected from the Common Sense Model of Diabetes Inventory. The inferential object of interest is the latent trait that indicates congruence of belief with the biomedical model. To apply an item response model to the poset diabetes inventory, we proved that a simple coding algorithm circumvents the requirement of writing new codes such that standard IRT software could be directly used for the purpose of item estimation and individual scoring. Simulation experiments were used to examine parameter recovery for the proposed poset model.

Hidden Markov latent variable models with multivariate longitudinal data

Cocaine addiction is chronic and persistent, and has become a major social and health problem in many countries. Existing studies have shown that cocaine addicts often undergo episodic periods of addiction to, moderate dependence on, or swearing off cocaine. Given its reversible feature, cocaine use can be formulated as a stochastic process that transits from one state to another, while the impacts of various factors, such as treatment received and individuals' psychological problems on cocaine use, may vary across states. This article develops a hidden Markov latent variable model to study multivariate longitudinal data concerning cocaine use from a California Civil Addict Program. The proposed model generalizes conventional latent variable models to allow bidirectional transition between cocaine-addiction states and conventional hidden Markov models to allow latent variables and their dynamic interrelationship. We develop a maximum-likelihood approach, along with a Monte Carlo expectation conditional maximization (MCECM) algorithm, to conduct parameter estimation. The asymptotic properties of the parameter estimates and statistics for testing the heterogeneity of model parameters are investigated. The finite sample performance of the proposed methodology is demonstrated by simulation studies. The application to cocaine use study provides insights into the prevention of cocaine use.

Physical Activity States of Preschool-Aged Latino Children in Farmworker Families: Predictive Factors and Relationship With BMI Percentile

BACKGROUND

Obesity disproportionately affects children of Latino farmworkers. Further research is needed to identify patterns of physical activity (PA) in this group and understand how PA affects Body Mass Index (BMI) percentile.

METHODS

Two hundred and forty-four participants ages 2.5 to 3.5 in the Niños Sanos longitudinal study wore accelerometers that measured daily PA. Several PA-related parameters formed a profile for conducting hidden Markov modeling (HMM), which identified different states of PA.

RESULTS

Latino farmworker children were generally sedentary. Two different states were selected using HMM-less active and more active. In the more active state; members spent more minutes in moderate-vigorous physical activity (MVPA). Most children were in the less active state at any given time; however, switching between states occurred commonly. One variable-mother's concern regarding lack of PA-was a marginally significant predictor of membership in the more active state. State did not predict BMI or weight percentile after adjusting for caloric intake.

CONCLUSION

Most children demonstrated high amounts of sedentary behavior, and rates of MVPA fell far below recommended levels for both states. The lack of statistically significant results for risk factors and PA state on weight-related outcomes is likely due to the homogeneous behaviors of the children.

Profiles of Food Security for US Farmworker Households and Factors Related to Dynamic of Change

OBJECTIVES

We recruited 248 farmworker families with preschool-aged children in North Carolina and examined food security indicators over 24 months to identify food security patterns and examine the dynamic of change over time.

METHODS

Participants in the Niños Sanos study, conducted 2011 to 2014, completed quarterly food security assessments. Based on responses to items in the US Household Food Security Survey Module, we identified different states of food security by using hidden Markov model analysis, and examined factors associated with different states. We delineated factors associated with changes in state by using mixed-effect ordinal logistic regression.

RESULTS

About half of the households (51%) consistently stayed in the most food-secure state. The least food-secure state was transient, with only 29% probability of this state for 2 consecutive quarters. Seasonal (vs migrant) work status, having immigration documents (vs not documented), and season predicted higher levels of food security.

CONCLUSIONS

Heterogeneity in food security among farmworker households calls for tailoring intervention strategies. The transiency and unpredictability of low food security suggest that access to safety-net programs could reduce low food security risk in this population.

Realizing the potential of mobile mental health: new methods for new data in psychiatry

Smartphones are now ubiquitous and can be harnessed to offer psychiatry a wealth of real-time data regarding patient behavior, self-reported symptoms, and even physiology. The data collected from smartphones meet the three criteria of big data: velocity, volume, and variety. Although these data have tremendous potential, transforming them into clinically valid and useful information requires using new tools and methods as a part of assessment in psychiatry. In this paper, we introduce and explore numerous analytical methods and tools from the computational and statistical sciences that appear readily applicable to psychiatric data collected using smartphones. By matching smartphone data with appropriate statistical methods, psychiatry can better realize the potential of mobile mental health and empower both patients and providers with novel clinical tools.

Reciprocal Markov Modeling of Feedback Mechanisms Between Emotion and Dietary Choice Using Experience-Sampling Data

With intensively collected longitudinal data, recent advances in the experience-sampling method (ESM) benefit social science empirical research, but also pose important methodological challenges. As traditional statistical models are not generally well equipped to analyze a system of variables that contain feedback loops, this paper proposes the utility of an extended hidden Markov model to model reciprocal the relationship between momentary emotion and eating behavior. This paper revisited an ESM data set (Lu, Huet, & Dube, 2011) that observed 160 participants' food consumption and momentary emotions 6 times per day in 10 days. Focusing on the analyses on feedback loop between mood and meal-healthiness decision, the proposed reciprocal Markov model (RMM) can accommodate both hidden ("general" emotional states: positive vs. negative state) and observed states (meal: healthier, same or less healthy than usual) without presuming independence between observations and smooth trajectories of mood or behavior changes. The results of RMM analyses illustrated the reciprocal chains of meal consumption and mood as well as the effect of contextual factors that moderate the interrelationship between eating and emotion. A simulation experiment that generated data consistent with the empirical study further demonstrated that the procedure is promising in terms of recovering the parameters.

A discrete time event-history approach to informative drop-out in mixed latent Markov models with covariates

Mixed latent Markov (MLM) models represent an important tool of analysis of longitudinal data when response variables are affected by time-fixed and time-varying unobserved heterogeneity, in which the latter is accounted for by a hidden Markov chain. In order to avoid bias when using a model of this type in the presence of informative drop-out, we propose an event-history (EH) extension of the latent Markov approach that may be used with multivariate longitudinal data, in which one or more outcomes of a different nature are observed at each time occasion. The EH component of the resulting model is referred to the interval-censored drop-out, and bias in MLM modeling is avoided by correlated random effects, included in the different model components, which follow common latent distributions. In order to perform maximum likelihood estimation of the proposed model by the expectation-maximization algorithm, we extend the usual forward-backward recursions of Baum and Welch. The algorithm has the same complexity as the one adopted in cases of non-informative drop-out. We illustrate the proposed approach through simulations and an application based on data coming from a medical study about primary biliary cirrhosis in which there are two outcomes of interest, one continuous and the other binary.

Reconciling statistical and systems science approaches to public health

Although systems science has emerged as a set of innovative approaches to study complex phenomena, many topically focused researchers including clinicians and scientists working in public health are somewhat befuddled by this methodology that at times appears to be radically different from analytic methods, such as statistical modeling, to which the researchers are accustomed. There also appears to be conflicts between complex systems approaches and traditional statistical methodologies, both in terms of their underlying strategies and the languages they use. We argue that the conflicts are resolvable, and the sooner the better for the field. In this article, we show how statistical and systems science approaches can be reconciled, and how together they can advance solutions to complex problems. We do this by comparing the methods within a theoretical framework based on the work of population biologist Richard Levins. We present different types of models as representing different tradeoffs among the four desiderata of generality, realism, fit, and precision.

Multi-profile hidden Markov model for mood, dietary intake, and physical activity in an intervention study of childhood obesity

Motivated by an application to childhood obesity data in a clinical trial, this paper describes a multi-profile hidden Markov model (HMM) that uses several temporal chains of measures respectively related to psychosocial attributes, dietary intake, and energy expenditure behaviors of adolescents in a school setting. Using these psychological and behavioral profiles, the model delineates health states from the longitudinal data set. Furthermore, a two-level regression model that takes into account the clustering effects of students within school is used to assess the effects of school-based and community-based interventions and other risk factors on the transition between health states over time. The results from our study suggest that female students tend to decrease their physical activities despite a high level of anxiety about weight. The finding is consistent across intervention and control arms.