Markov Model Research Articles

MMG4: Recognition of G4-Forming Sequences Based on Markov Model.

G-quadruplexes (G4s) are special nucleic acid structures with various important biological functions. Existing tools and technologies for G4-forming sequences recognition are limited to time-consuming and costly methods such as circular dichroism and nuclear magnetic resonance. Developing a fast and accurate model for G4-forming sequences recognition has far-reaching significance. In this study, MMG4, a novel model to recognize G4-forming sequences based on Markov model (MM), was developed and the phenomenon of high recognition accuracy in the central region of the sequence and low accuracy in the two end regions was discovered. It was further found that the differences in base transfer probabilities, ratio distribution, and G4-motif structural content in different regions may be the causes of this phenomenon. The study also explored the impact of sequence length on recognition accuracy and found the optimal recognition interval to be [910-1049], with the highest recognition accuracy reaching 85.95%. By extracting sequence features, the study constructed three types of machine learning models: random forest (RF), support vector machine, and back-propagation neural network. It was found that recognition performance of MM was significantly better than that of the other three machine learning models, proving that the recognition method based on MM can effectively capture the correlation information between adjacent nucleotides of G4. By combining MM with the three machine learning models, the predictive performance of MMG4 improved. Among them, the RF model combined with MM has the best performance, achieving an area under the receiver operating characteristic curve value of 0.93 and an area under the precision-recall curve value of 0.9. Finally, the study validated the model robustness and generalization ability through independent testing dataset.

Artificial Intelligence and the Future of Communication Sciences and Disorders: A Bibliometric and Visualization Analysis.

As artificial intelligence (AI) takes an increasingly prominent role in health care, a growing body of research is being dedicated to its application in the investigation of communication sciences and disorders (CSD). This study aims to provide a comprehensive overview, serving as a valuable resource for researchers, developers, and professionals seeking to comprehend the evolving landscape of AI in CSD research. We conducted a bibliometric analysis of AI-based research in the discipline of CSD published up to December 2023. Utilizing the Web of Science and Scopus databases, we identified 15,035 publications, with 4,375 meeting our inclusion criteria. Based on the bibliometric data, we examined publication trends and patterns, characteristics of research activities, and research hotspot tendencies. From 1985 onwards, there has been a consistent annual increase in publications, averaging 16.51%, notably surging from 2012 to 2023. The primary communication disorders studied include autism, aphasia, dysarthria, Parkinson's disease, and Alzheimer's disease. Noteworthy AI models instantiated in CSD research encompass support vector machine, convolutional neural network, and hidden Markov model, among others. Compared to AI applications in other fields, the adoption of AI in CSD has lagged slightly behind. While CSD studies primarily use classical machine learning techniques, there is a growing trend toward the integration of deep learning methods. AI technology offers significant benefits for both research and clinical practice in CSD, but it also presents certain challenges. Moving forward, collaboration among technological, research, and clinical domains is essential to empower researchers and speech-language pathologists to effectively leverage AI technology for the study, diagnosis, assessment, and rehabilitation of CSD. https://doi.org/10.23641/asha.27162564.

Generative dynamical models for classification of rsfMRI data

ABSTRACT The growing availability of large-scale neuroimaging datasets and user-friendly machine learning tools has led to a recent surge in studies that use fMRI data to predict psychological or behavioral variables. Many such studies classify fMRI data on the basis of static features, but fewer try to leverage brain dynamics for classification. Here, we pilot a generative, dynamical approach for classifying resting-state fMRI (rsfMRI) data. By fitting separate hidden Markov models to the classes in our training data and assigning class labels to test data based on their likelihood under those models, we are able to take advantage of dynamical patterns in the data without confronting the statistical limitations of some other dynamical approaches. Moreover, we demonstrate that hidden Markov models are able to successfully perform within-subject classification on the MyConnectome dataset solely on the basis of transition probabilities among their hidden states. On the other hand, individual Human Connectome Project subjects cannot be identified on the basis of hidden state transition probabilities alone—although a vector autoregressive model does achieve high performance. These results demonstrate a dynamical classification approach for rsfMRI data that shows promising performance, particularly for within-subject classification, and has the potential to afford greater interpretability than other approaches.

Evaluating Medical Entity Recognition in Health Care: Entity Model Quantitative Study.

Named entity recognition (NER) models are essential for extracting structured information from unstructured medical texts by identifying entities such as diseases, treatments, and conditions, enhancing clinical decision-making and research. Innovations in machine learning, particularly those involving Bidirectional Encoder Representations From Transformers (BERT)-based deep learning and large language models, have significantly advanced NER capabilities. However, their performance varies across medical datasets due to the complexity and diversity of medical terminology. Previous studies have often focused on overall performance, neglecting specific challenges in medical contexts and the impact of macrofactors like lexical composition on prediction accuracy. These gaps hinder the development of optimized NER models for medical applications. This study aims to meticulously evaluate the performance of various NER models in the context of medical text analysis, focusing on how complex medical terminology affects entity recognition accuracy. Additionally, we explored the influence of macrofactors on model performance, seeking to provide insights for refining NER models and enhancing their reliability for medical applications. This study comprehensively evaluated 7 NER models-hidden Markov models, conditional random fields, BERT for Biomedical Text Mining, Big Transformer Models for Efficient Long-Sequence Attention, Decoding-enhanced BERT with Disentangled Attention, Robustly Optimized BERT Pretraining Approach, and Gemma-across 3 medical datasets: Revised Joint Workshop on Natural Language Processing in Biomedicine and its Applications (JNLPBA), BioCreative V CDR, and Anatomical Entity Mention (AnatEM). The evaluation focused on prediction accuracy, resource use (eg, central processing unit and graphics processing unit use), and the impact of fine-tuning hyperparameters. The macrofactors affecting model performance were also screened using the multilevel factor elimination algorithm. The fine-tuned BERT for Biomedical Text Mining, with balanced resource use, generally achieved the highest prediction accuracy across the Revised JNLPBA and AnatEM datasets, with microaverage (AVG_MICRO) scores of 0.932 and 0.8494, respectively, highlighting its superior proficiency in identifying medical entities. Gemma, fine-tuned using the low-rank adaptation technique, achieved the highest accuracy on the BioCreative V CDR dataset with an AVG_MICRO score of 0.9962 but exhibited variability across the other datasets (AVG_MICRO scores of 0.9088 on the Revised JNLPBA and 0.8029 on AnatEM), indicating a need for further optimization. In addition, our analysis revealed that 2 macrofactors, entity phrase length and the number of entity words in each entity phrase, significantly influenced model performance. This study highlights the essential role of NER models in medical informatics, emphasizing the imperative for model optimization via precise data targeting and fine-tuning. The insights from this study will notably improve clinical decision-making and facilitate the creation of more sophisticated and effective medical NER models.

Cost-Effectiveness of Teduglutide for Pediatric Patients with Short Bowel Syndrome in Japan, Including Caregiver Burden.

Short bowel syndrome (SBS) is associated with a significant mental and physical burden for patients and caregivers. Standard of care (SOC) for SBS includes parenteral support (PS) to optimize intestinal function. Teduglutide, a recombinant human glucagon-like peptide2 analogue, reduces the need for PS in patients with SBS. In this study, we assessed the cost-effectiveness of teduglutide in pediatric patients with SBS from multiple perspectives, considering the caregiver's burden. A Markov model was used to evaluate cost (Japanese yen, JPY) and effectiveness (quality-adjusted life years, QALYs) of teduglutide compared with SOC for pediatric patients with SBS in Japan. We conducted a base-case analysis and selected sensitivity and scenario analyses from three perspectives: (1) the public healthcare payer, (2) the public healthcare and long-term care payer, and (3) society. In the base-case analysis, the incremental cost-effectiveness ratio (ICER) was 9,533,412 JPY per QALY from the public healthcare payer perspective, 6,335,980 JPY per QALY from the public healthcare and long-term care payer perspective, and 3,510,371 JPY per QALY from the societal perspective. The probability that cost-effectiveness of teduglutide is favorable from a societal perspective was 59.3%. In all scenario analyses, consistent with the base-case analysis, ICERs for teduglutide compared with SOC were different depending on whether caregiver utility and productivity loss were considered. Incorporating the caregiver's burden in the cost-effectiveness analysis of teduglutide for pediatric patients with SBS provided a more comprehensive assessment of the value of teduglutide for patients, their families, and society. This approach enhances our understanding of the overall value of a treatment, especially for diseases with significant caregiver burden.

Budget impact analysis of TPMT and NUDT15 pharmacogenomic testing for 6-mercaptopurine in pediatric acute lymphoblastic leukemia patients.

Pharmacogenomic testing identifies gene polymorphisms impacting drug metabolism, aiding in optimizing treatment efficacy and minimizing toxicity, thus potentially reducing healthcare utilization. 6-Mercaptopurine metabolism is affected by thiopurine methyltransferase (TPMT) and nudix hydrolase 15 (NUDT15) polymorphisms. We sought to estimate the budget impact of preemptive pharmacogenomic testing for these genes in pediatric acute lymphoblastic leukemia (ALL) patients from an institutional perspective. A Markov model was constructed to model the first cycle of the maintenance phase of chemotherapy for pediatric ALL patients transitioning between one of three health states: stable, moderately myelosuppressed, and severely myelosuppressed over 16 weeks, with each health state's associated costs derived from the literature. The patient's likelihood to experience moderate or severe myelosuppression based on metabolism phenotype was calculated from the literature and applied on a weekly basis, and the marginal budget impact of preemptive pharmacogenomic testing vs. no pharmacogenomic testing was calculated. One-way sensitivity analysis was conducted to assess parameter influence on results. Preemptive pharmacogenomic testing of TPMT and NUDT15 provided savings of up to $26 028 per patient during the maintenance phase. In the sensitivity analysis, the cost of outpatient management of moderate myelosuppression had the greatest impact on the budget, resulting in cost savings ranging from $8592 to $30 129 when the minimum and maximum costs of management were used in the model. Preemptive pharmacogenomic testing for TPMT and NUDT15 polymorphisms before initiation of maintenance therapy for pediatric ALL patients yielded considerable cost savings.

Cost-effectiveness of pembrolizumab as an adjuvant treatment of renal cell carcinoma post-nephrectomy in Switzerland

Aims Pembrolizumab has demonstrated significantly prolonged disease-free survival and overall survival (OS) among adult patients post-nephrectomy who have an intermediate-high risk, high-risk, or M1 stage with no evidence of disease (M1 NED) renal cell carcinoma (RCC) with clear cell component. The aim of this study was to evaluate the cost-effectiveness of pembrolizumab for patients with RCC post-nephrectomy versus observation in Switzerland. Materials and methods A previously published Markov model was adapted for the Swiss setting to estimate the cost-effectiveness of adjuvant pembrolizumab versus observation from the Swiss statutory health insurance perspective. Transition probabilities between model states were estimated using survival curves from KEYNOTE-564 (data cut-off: June 14, 2021). Outcomes included costs (2022 Swiss francs [CHF]), quality-adjusted life-years (QALYs), and life-years (LYs), measured over a lifetime horizon. Costs included drug acquisition and administration for adjuvant and subsequent therapy. Both costs and effectiveness were discounted at 3.0% annually. Cost-effectiveness was evaluated at a hypothetical willingness-to-pay (WTP) threshold of CHF 100,000 Sensitivity was assessed through scenario analyses as well as deterministic and probabilistic sensitivity analyses. Results Over a lifetime horizon, the total incremental cost for pembrolizumab versus observation was CHF 59,089, providing incremental gains of 0.90 QALYs (1.07 LYs); the incremental cost-effectiveness ratio was CHF 65,299/QALY. Pembrolizumab was deemed cost-effective versus observation, with a 69.9% probability of cost-effectiveness. Limitations A more recent interim analysis data cut from KEYNOTE-564 with median follow up of 57.2 months has since been published; however, these were not available at the time of analysis. It would likely have minimal impact on transition probabilities from disease-free, and the current approach remains conservative for predicting OS for pembrolizumab. Conclusions As an adjuvant treatment of RCC post-nephrectomy, pembrolizumab was found to be cost-effective versus observation in Switzerland at a WTP threshold of CHF 100,000/QALY. Policy makers should consider pembrolizumab as an adjuvant treatment for patients with RCC post-nephrectomy when making decisions regarding resource allocation.

Mathematics behind the identifying CpG islands

The major objective of the paper is to review the theory for an hidden Markov model, a very general type of probabilistic model for sequences of symbols. In order for the hidden Markov model to be applicable to real-world applications, three key problems about the model must be addressed, and to do this, first we go over how to choose the best state sequence to explain an observation sequence, then we go over how to calculate the probability of an observation sequence, and finally we go over how to compute the maximization of the probability of the observation sequence. From these three angles, we review the mathematical concept behind the identification of CpG islands. The entire process and study of the outcomes have been tackled by examining both hypothetical and real DNA sequences side by side. We use well-known biological sequence analysis servers to carry out the experiment. Analytical and algorithmic approaches are compared while taking the hypothetical DNA sequence example into consideration.

Accurate identification of locally aneuploid cells by incorporating cytogenetic information in single cell data analysis

Single-cell RNA sequencing is a powerful tool to investigate the cellular makeup of tumor samples. However, due to the sparse data and the complex tumor microenvironment, it can be challenging to identify neoplastic cells that play important roles in tumor growth and disease progression. This is especially relevant for blood cancers, where neoplastic cells may be highly similar to normal cells. To address this challenge, we have developed partCNV and partCNVH, two methods for rapid and accurate detection of aneuploid cells with local copy number deletion or amplification. PartCNV uses an expectation-maximization (EM) algorithm with mixtures of Poisson distributions and incorporates cytogenetic information to guide the classification. PartCNVH further improves partCNV by integrating a hidden Markov model for feature selection. We have thoroughly evaluated the performance of partCNV and partCNVH through simulation studies and real data analysis using three scRNA-seq datasets from blood cancer patients. Our results show that partCNV and partCNVH have favorable accuracy and provide more interpretable results compared to existing methods. In the real data analysis, we have identified multiple biological processes involved in the oncogenesis of myelodysplastic syndromes and acute myeloid leukemia.

Are information criteria good enough to choose the right the number of regimes in hidden Markov models?

Selecting the number of regimes in Hidden Markov models is an important problem. Many criteria are used to do so, such as Akaike information criterion (AIC), Bayesian information criterion (BIC), integrated completed likelihood (ICL), deviance information criterion (DIC), and Watanabe-Akaike information criterion (WAIC), to name a few. In this article, we introduced goodness-of-fit tests for general Hidden Markov models with covariates, where the distribution of the observations is arbitrary, i.e. continuous, discrete, or a mixture of both. A selection procedure is proposed based on this goodness-of-fit test. The main aim of this article is to compare the classical information criterion with the new criterion when the outcome is either continuous, discrete or zero-inflated. Numerical experiments assess the finite sample performance of the goodness-of-fit tests, and comparisons between the different criteria are made.

Single-vesicle imaging reveals actin-dependent spatial restriction of vesicles at the active zone, essential for sustained transmission

Synaptic-vesicle (SV) recruitment is thought to maintain reliable neurotransmitter release during high-frequency signaling. However, the mechanism underlying the SV reloading for sustained neurotransmission at central synapses remains unknown. To elucidate this, we performed direct observations of SV reloading and mobility at a single-vesicle level near the plasma membrane in cerebellar mossy fiber terminals using total internal reflection fluorescence microscopy, together with simultaneous recordings of membrane fusion by capacitance measurements. We found that actin disruption abolished the rapid SV recruitment and reduced sustained release. In contrast, induction of actin polymerization and stabilization did not affect vesicle recruitment and release, suggesting that the presence of actin filaments, rather than actin dynamics, was required for the rapid recruitment. Single-particle tracking experiments of quantum dot-labeled vesicles, which allows nanoscale resolution of vesicle mobility, revealed that actin disruption caused vesicles to diffuse more rapidly. Hidden Markov modeling with Bayesian inference revealed that SVs had two diffusion states under normal conditions: free-diffusing and trapped. After disruption of the actin filament, vesicles tended to have only the free-diffusing state. F-actin staining showed that actin filaments were localized outside the active zones (AZs) and surrounded some SV trajectories. Perturbation of SV mobility, possibly through interference with biomolecular condensates, also suggested that the restricted diffusion state determined the rate of SV recruitment. We propose that actin filaments confined SVs near the AZ to achieve rapid and efficient recruitment followed by priming and sustained synaptic transmission.

Nonhomogeneous hidden semi-Markov models for toroidal data

Abstract A nonhomogeneous hidden semi-Markov model is proposed to segment bivariate time series of wind and wave directions according to a finite number of latent regimes and, simultaneously, estimate the influence of time-varying covariates on the process’ survival under each regime. The model is a mixture of toroidal densities, whose parameters depend on the evolution of a semi-Markov chain, which is in turn modulated by time-varying covariates. It includes nonhomogeneous hidden Markov models and hidden semi-Markov models as special cases. Parameter estimates are obtained using an Expectation-Maximization algorithm that relies on an efficient augmentation of the latent process. Fitted on a time series of wind and wave directions recorded in the Adriatic Sea, the model offers a clear-cut description of sea state dynamics in terms of latent regimes and captures the influence of time-varying weather conditions on the duration of such regimes.

Predator-prey space use and landscape features influence movement behaviors in a large-mammal community.

Predator hunting strategies, such as stalking versus coursing behaviors, are hypothesized to influence antipredator behaviors of prey and can describe the movement behaviors of predators themselves. Predators and prey may alter their movement in relation to predator hunting modes, yet few studies have evaluated how these strategies influence movement behaviors of free-ranging animals in a multiple-predator, multiple-prey system. We fit hidden Markov models (HMM) with movement data derived from >400 GPS-collared ungulates and large predators in eastern Washington, USA. We used these models to test our hypotheses that stalking (cougars [Puma concolor]) and coursing (gray wolves [Canis lupus]) predators would exhibit different broad-scale movement behaviors consistent with their respective hunting strategies in areas that increased the likelihood of encountering or capturing ungulate prey (e.g., habitats selected by deer [Odocoileus spp.]). Similarly, we expected that broadscale movement behaviors of prey would change in response to background levels of predation risk associated with each predator's hunting strategy. We found that predators and ungulate prey adjusted their broadscale movements in response to one another's long-term patterns of habitat selection but not based on differences in predator-hunting strategies. Predators changed their movement behaviors based on the type of prey, whereas ungulates generally reduced movement in areas associated with large predators, regardless of the predator's hunting strategy. Both predator and prey movements varied in response to landscape features but not necessarily based on habitat that would facilitate specific hunting behaviors. Our results suggest that predators and prey adjust their movements at broad temporal scales in relation to long-term patterns of risk and resource distributions, potentially influencing their encounter rates with one another at finer spatiotemporal scales. Habitat features further influenced changes in movement, resulting in a complex combination of movement behaviors in multiple-predator, multiple-prey systems.

Three-State Hidden Markov Model for Spectrum Prediction in Cognitive Radio Networks

The exponential growth and proliferation of wireless devices for different wireless applications have led to the emergence of cognitive radio network (CRN) for optimal utilization of scarce spectrum resources. However, these resources have grossly been under-utilized due to the inaccurate spectrum predictions. Existing spectrum occupancy and prediction techniques which rely on 2-state hidden Markov model (HMM) results in false alarm or missed detection caused by noisy or incomplete observable effects. In this paper, a 3-state HMM spectrum occupancy and prediction technique in CRNs is proposed. The transmission, emission and initial state probabilities of the proposed 3-state HMM parameters were derived based on the three canonical problems associated with HMM. The evaluation, decoding and learning problems were solved using Forward algorithm, Viterbi algorithm and the Baum-Welch algorithm, respectively. The performance of the proposed 3-state HMM spectrum prediction technique was evaluated using prediction accuracy, probability of detection and spectrum utilization efficiency. The simulation results obtained revealed that the 3-state HMM outperformed the 2-state HMM spectrum prediction technique by 24.1% in prediction accuracy.

Comparing Recurrence Rates and the Cost-Effectiveness of Arthroscopic Labral Repair and Nonoperative Management for Primary Anterior Shoulder Dislocations in Young Patients: A Decision-Analytic Markov Model-Based Analysis

Background: Value-based decision-making regarding nonoperative management versus early surgical stabilization for first-time anterior shoulder instability (ASI) events remains understudied. Purpose: To perform (1) a systematic review of the current literature and (2) a Markov model–based cost-effectiveness analysis comparing an initial trial of nonoperative management to arthroscopic Bankart repair (ABR) for first-time ASI. Study Design: Economic and decision analysis; Level of evidence, 3. Methods: A Markov chain Monte Carlo probabilistic model was developed to evaluate the outcomes and costs of 1000 simulated patients (mean age, 20 years; range, 12-26 years) with first-time ASI undergoing nonoperative management versus ABR. Utility values, recurrence rates, and transition probabilities were derived from the published literature. Costs were determined based on the typical patient undergoing each treatment strategy at the authors’ institution. Outcome measures included costs, quality-adjusted life-years (QALYs), and the incremental cost-effectiveness ratio (ICER). Results: The Markov model with Monte Carlo microsimulation demonstrated mean (± standard deviation) 10-year costs for nonoperative management and ABR of $38,649 ± $10,521 and $43,052 ± $9352, respectively. Total QALYs acquired over the 10-year time horizon were 7.67 ± 0.43 and 8.44 ± 0.46 for nonoperative management and ABR, respectively. The ICER comparing ABR with nonoperative management was found to be just $5725/QALY, which falls substantially below the $50,000 willingness-to-pay (WTP) threshold. The mean numbers of recurrences were 2.55 ± 0.31 and 1.17 ± 0.18 for patients initially assigned to the nonoperative and ABR treatment groups, respectively. Of 1000 samples run over 1000 trials, ABR was the optimal strategy in 98.7% of cases, with nonoperative management the optimal strategy in 1.3% of cases. Conclusion: ABR reduces the risk for recurrent dislocations and is more cost-effective despite higher upfront costs when compared with nonoperative management for first-time ASI in the young patient. While all these factors are important to consider in surgical decision-making, ultimate treatment decisions should be made on an individual basis and occur through a shared decision-making process.

Prevention Lab: a predictive model for estimating the impact of prevention interventions in a simulated Italian cohort

BackgroundA large fraction of the disease burden in the Italian population is due to behavioral risk factors. The objective of this work is to provide a tool to estimate the impact of preventive interventions that reduce the exposure to smoking and sedentary lifestyle of the Italian population, with the goal of selecting optimal interventions.MethodsWe construct a Markovian model that simulates the state of each subject of the Italian population. The model predicts the distribution of subjects in each health status and risk factor status for every year of the simulation. Based on this distribution, the model provides a rich output summary, such as the number of incident and prevalent cases for each tracing disease and the Disability Adjusted Life Years (DALY), used to assess the impact of preventive interventions, and how this impact is shaped in time.ResultsThis paper focuses on the methodological aspects of the model. The proposed model is flexible and can be applied to estimate the impact of complex interventions on the two risk factors and adapted to consider different cohorts. We validate the model by simulating the evolution of the Italian population from 2009 to 2017 and comparing the output with historical data. Furthermore, as a case-study, we simulate a counterfactual scenario where both tobacco and sedentary lifestyle are eradicated from the Italian population in 2019 and estimate the impact of such intervention over the following 20 years.ConclusionsWe propose a Markovian model to estimate how interventions on smoking and sedentary lifestyle can affect the reduction of the disease burden, and validate the model on historical data. The model is flexible and allows to extend the analysis to consider more risk factors in future research. However, we are aware that, given the ever-increasing availability of data, it is necessary in the future to increase the complexity of the model, to be closer to reality and to provide decision-making support to the policy-makers.

Does the Vaccination against Tick-Borne Encephalitis Offer Good Value for Money for Incidence Rates below the WHO Threshold for Endemicity? A Case Study for Germany.

A closed-cohort Markov model was developed to estimate the cost-effectiveness of TBE vaccination. We compared primary vaccination applied to the whole population (aged above 1 year) and to a subpopulation aged between 60 and 85 years to a scenario without vaccination. Since TBE incidence is often underestimated, we included under-ascertained TBE cases and non-CNS TBE infections. Germany was used as a case study due to the availability of detailed incidence data. Our incidence threshold analysis showed that TBE vaccination offers good value for money well below the WHO threshold in most of the analyzed scenarios. Our results support a recommendation for TBE vaccination even in settings with low numbers of reported cases, especially for older patients. Furthermore, this analysis identified major research gaps regarding the costs, utilities, and clinical progression of TBE.

Information and Communications Technology-Based Monitoring Service for Tailored Chronic Disease Management in Primary Care: Cost-Effectiveness Analysis Based on ICT-CM Trial Results.

Information and communications technology-based tailored management (TM) intervention is a novel automatic system in which a smartphone app for the management of patients with hypertension and diabetes, the provider web, and Bluetooth devices are linked. However, little evidence exists regarding the cost-effectiveness of the interventions using mobile apps. This study aimed to assess the cost-effectiveness of TM intervention for adult patients with hypertension or diabetes in primary care compared with usual care (UC). Cost-effectiveness analysis using a Markov model was conducted from the Korean health care system perspective. Based on 6-month outcome data from an information and communications technology-based tailored chronic disease management (ICT-CM) trial, effectiveness over a lifetime beyond the trial periods was extrapolated using a cardiovascular disease risk prediction model. Costs were estimated using ICT-CM trial data and national health insurance claims data. Health utility weights were obtained from the Korea National Health and Nutrition Examination Survey. In the base-case analysis, compared with UC, TM was more costly (US $23,157 for TM vs US $22,391 for UC) and more effective (12.006 quality-adjusted life-years [QALYs] for TM vs 11.868 QALYs for UC). The incremental cost-effectiveness ratio was US $5556 per QALY gained. Probabilistic sensitivity analysis showed that the probability of TM being cost-effective compared with UC was approximately 97% at an incremental cost-effectiveness ratio threshold of US $26,515 (KRW 35 million) per QALY gained. Compared with UC, TM intervention is a cost-effective option for patients with hypertension or diabetes in primary care settings. The study results can assist policy makers in making evidence-based decisions when implementing accessible chronic disease management services.

Deconstructing Deep Active Inference: A Contrarian Information Gatherer.

Active inference is a theory of perception, learning, and decision making that can be applied to neuroscience, robotics, psychology, and machine learning. Recently, intensive research has been taking place to scale up this framework using Monte Carlo tree search and deep learning. The goal of this activity is to solve more complicated tasks using deep active inference. First, we review the existing literature and then progressively build a deep active inference agent as follows: we (1) implement a variational autoencoder (VAE), (2) implement a deep hidden Markov model (HMM), and (3) implement a deep critical hidden Markov model (CHMM). For the CHMM, we implemented two versions, one minimizing expected free energy, CHMM[EFE] and one maximizing rewards, CHMM[reward]. Then we experimented with three different action selection strategies: the ε-greedy algorithm as well as softmax and best action selection. According to our experiments, the models able to solve the dSprites environment are the ones that maximize rewards. On further inspection, we found that the CHMM minimizing expected free energy almost always picks the same action, which makes it unable to solve the dSprites environment. In contrast, the CHMM maximizing reward keeps on selecting all the actions, enabling it to successfully solve the task. The only difference between those two CHMMs is the epistemic value, which aims to make the outputs of the transition and encoder networks as close as possible. Thus, the CHMM minimizing expected free energy repeatedly picks a single action and becomes an expert at predicting the future when selecting this action. This effectively makes the KL divergence between the output of the transition and encoder networks small. Additionally, when selecting the action down the average reward is zero, while for all the other actions, the expected reward will be negative. Therefore, if the CHMM has to stick to a single action to keep the KL divergence small, then the action down is the most rewarding. We also show in simulation that the epistemic value used in deep active inference can behave degenerately and in certain circumstances effectively lose, rather than gain, information. As the agent minimizing EFE is not able to explore its environment, the appropriate formulation of the epistemic value in deep active inference remains an open question.

Introduction of a hexavalent vaccine containing acellular pertussis into the national immunization program for infants in Peru: a cost-consequence analysis of vaccination coverage

BackgroundInfant vaccination coverage rates in Peru have declined in recent years, exacerbated by the COVID-19 pandemic. Introduction of the fully-liquid diphtheria, tetanus, and acellular pertussis (DTaP)-inactivated polio vaccine (IPV)-hepatitis B (HB)-Haemophilus influenzae type B (Hib) hexavalent vaccine (DTaP-IPV-HB-Hib) in Peru’s infant National Immunization Program may help improve coverage. We evaluated costs and healthcare outcomes, including coverage, of switching from a pentavalent vaccine containing whole-cell pertussis component (DTwP-HB-Hib) plus IPV/oral polio vaccine (IPV/OPV) to the hexavalent vaccine for the primary vaccination scheme (2, 4 and 6 months).MethodsThe analysis was performed over a 5-year period on a cohort of children born in Peru in 2020 (N = 494,595). Four scenarios were considered: the pentavalent plus IPV/OPV scheme (S1); replacing the pentavalent plus IPV/OPV scheme with the hexavalent scheme (S2); expanded delivery of the pentavalent plus IPV/OPV scheme (S3); expanded delivery of the hexavalent scheme (S4). Vaccine coverage and incidence of adverse reactions (ARs) were estimated using Monte Carlo simulations and previous estimates from the literature. Cases of vaccine-preventable diseases were estimated using a Markov model. Logistical and healthcare costs associated with these outcomes were estimated. Impact of key variables (including coverage rates, incidence of ARs and vaccine prices) on costs was evaluated in sensitivity analyses.ResultsThe overall cost from a public health payer perspective associated with the pentavalent plus IPV/OPV vaccine scheme (S1) was estimated at $56,719,350, increasing to $61,324,263 (+ 8.1%), $59,121,545 (+ 4.2%) and $64,872,734 (+ 14.4%) in scenarios S2, S3 and S4, respectively. Compared with the status quo (S1), coverage rates were estimated to increase by 3.1% points with expanded delivery alone, and by 9.4 and 14.3% points, if the hexavalent vaccine is deployed (S2 and S4, respectively). In both scenarios with the hexavalent vaccine (S2 and S4), pertussis cases would also be 5.7% and 8.7% lower, and AR rates would decrease by 32%. The cost per protected child would be reduced when the hexavalent vaccine scheme. Incidence of ARs was an important driver of cost variability in the sensitivity analysis.ConclusionsImplementation of the hexavalent vaccine in Peru’s National Immunization Program has a positive public health cost consequence.