Multivariate Beta Research Articles

Healthcare contact days experienced by patients with advanced gastrointestinal cancer.

477 Background: Frequent visits to healthcare facilities can be time consuming and resource intensive for people with cancer. We assessed patterns of in-person healthcare contact days among veterans with advanced gastrointestinal (GI) cancer. Methods: We conducted a retrospective cohort analysis using the Veterans Affairs Clinical Cancer Registry (VACCR) and Corporate Data Warehouse (CDW) to identify decedents with stage IV GI cancer between 2011 and 2021 at the Minneapolis VA Hospital, Minnesota, USA. We determined the nature of contact days (clinician visit, imaging, inpatient, etc.) from diagnosis to death by reviewing encounters in the electronic medical record. The primary outcome was the percentage of days with health care contact (contact days/overall survival). We calculated descriptive statistics and visualized the trajectory of percentage of contact days over time by standardizing differential survival from cancer diagnosis to death and fitting the observations to a cubic smoothing spline. Using multivariable beta regression, we analyzed for clinicodemographic factors associated with contact days. Results: We analyzed data for 468 patients (median [IQR] age at diagnosis 69 [64-76] years, 99% men, 82% white, 31% with gastroesophageal cancer and 47% with receipt of any cancer directed treatment). Median [IQR] overall survival was 126 [49-318] days and median [IQR] contact days were 44 [20-95], or 35% of survival days. Ambulatory clinician visits represented the largest source of contact days (61.4%). A majority (88%) of patients were hospitalized at least once, with a median of 13 total inpatient days. The highest percentage of contact days occurred immediately after diagnosis (45%) and prior to death (32%) with a more modest middle phase, forming a U-shaped curve. This pattern was consistent when stratifying by length of survival. Beta regression results are shown in Table 1. Conclusions: Veterans with advanced GI cancer had a median survival of four months and spent just over one third of their days receiving in-person healthcare. Contact days were highest post-diagnosis and pre-death. Higher co-morbidity burden, urban residence, later diagnosis period, and no cancer directed treatment were associated with increased contact days. These findings highlight the need to understand sources of variation, benchmark appropriate care, and improve efficiency of visits, especially at critical phases of care.[Table: see text]

Health Care Contact Days Experienced by Decedents With Advanced GI Cancer.

Frequent visits to health care facilities can be time intensive and all-consuming for people with cancer. We measured health care contact days (days with healthcare contact outside the home) among decedents with advanced GI cancer and examined sources of contact days, their associations with demographic and clinical factors, and their temporal patterns over the course of illness. We conducted a retrospective cohort study using a tumor registry and electronic medical record data for decedents with stage IV GI cancer between 2011 and 2019 in a large health care network in MN. We determined contact days from diagnosis to death using chart review. Using multivariable beta regression adjusted for sociodemographic and clinical characteristics offset by survival, we calculated adjusted estimates of contact days and determined patient-level factors associated with percentage of contact days. We identified 809 patients eligible for analysis (median [IQR] age at diagnosis, 65 [56-73] years). The median (IQR) overall survival was 175 (56-459) days. Patients spent a median (IQR) of 25.8% (17.4%-39.1%) of these as contact days. Of these days, 83.6% were spent on outpatient visits. In the multivariable analysis, older age, Black race, and never receiving systemic cancer-directed treatment were associated with a higher percentage of contact days. The percentage of contact days was highest in the first month after diagnosis (39.6%) and before death (32.2%), with a more moderate middle phase (U-shaped curve). Decedents with advanced GI cancer spend 1 in 4 days alive with health care contact, despite a median survival of under 6 months. This is even higher immediately postdiagnosis and near death. These findings highlight the need to understand sources of variation, benchmark appropriate care, and deliver more efficient care for this vulnerable population with limited time.

The art of Schwinger and Feynman parametrizations

We present a derivation of the Schwinger parametrization and the Feynman parametrization in detail and their elementary applications. Although the parametrizations are essential to computing the loop integral arising in relativistic quantum field theory, their detailed derivations are not presented in usual textbooks. Beginning with an integral representation of the unity, we derive the Schwinger parametrization by performing multiple partial derivatives and utilizing the analyticity of the gamma function. The Feynman parametrization is derived by the partial-fraction decomposition and the change of variables introducing an additional delta function. Through the extensive employment of the analyticity of a complex function, we show the equivalence of those parametrizations. As applications of the parametrizations, we consider the combinatorial factor arising in the Feynman parametrization integral and the multivariate beta function. The combinatorial factor corresponds to an elementary integral embedded in the time-ordered product of the Dyson series in the time-dependent perturbation theory. We believe that the derivation presented here can be a good pedagogical example that students enhance their understanding of complex variables and train the use of the Dirac delta function in coordinate transformation.

An Individualized Low-Pneumoperitoneum-Pressure Strategy May Prevent a Reduction in Liver Perfusion during Colorectal Laparoscopic Surgery.

High intra-abdominal pressure (IAP) during laparoscopic surgery is associated with reduced splanchnic blood flow. It is uncertain whether a low IAP prevents this reduction. We assessed the effect of an individualized low-pneumoperitoneum-pressure strategy on liver perfusion. This was a single-center substudy of the multicenter 'Individualized Pneumoperitoneum Pressure in Colorectal Laparoscopic Surgery versus Standard Therapy II study' (IPPCollapse-II), a randomized clinical trial in which patients received an individualized low-pneumoperitoneum strategy (IPP) or a standard pneumoperitoneum strategy (SPP). Liver perfusion was indirectly assessed by the indocyanine green plasma disappearance rate (ICG-PDR) and the secondary endpoint was ICG retention rate after 15 min (R15) using pulse spectrophotometry. Multivariable beta regression was used to assess the association between group assignment and ICG-PDR and ICG-R15. All 29 patients from the participating center were included. Median IAP was 8 (25th-75th percentile: 8-10) versus 12 (12,12) mmHg, in IPP and SPP patients, respectively (p < 0.001). ICG-PDR was higher (OR 1.42, 95%-CI 1.10-1.82; p = 0.006) and PDR-R15 was lower in IPP patients compared with SPP patients (OR 0.46, 95%-CI 0.29-0.73; p = 0.001). During laparoscopic colorectal surgery, an individualized low pneumoperitoneum may prevent a reduction in liver perfusion.

Avifaunal diversity in Indian Institute of Technology Guwahati Campus, Assam, India

Indian Institute of Technology - Guwahati (IITG), Assam, is an ecologically rich campus hosting different species of birds, butterflies and mammals. It accommodates several migratory and resident species of birds across different seasons. However, information is scanty on avian diversity with respect to the different habitats of the campus. Therefore, the present study attempts to gain insight into avian diversity with respect to habitat heterogeneity by considering the species presence-absence dataset collected for three years (2017–2020). A multivariate Beta (β) diversity analysis is carried out for the IITG campus constituted of five primary habitats, viz., secondary growth, eco–forest, water bodies, swampy-marshy area, and constructions. Of 152 bird species observed in the IITG campus, the highest number is reported from secondary growth, followed by eco-forest. The multivariate analysis shows that the average β–diversity for the IITG campus is approximately equal to 79%, which is in accordance with another published study. These observations are examined in light of hypotheses and phenomena documented in the literature, such as habitat heterogeneity hypothesis, niche-based hypothesis and anthropogenic impact on habitats. The study also establishes that the IITG is among the educational institutes and campuses that host many migratory bird species. Lastly, based on the outcomes of β–diversity analysis, it is suggested that the conservation effort for avian species in the campus should be directed towards individual habitats uniformly.

Unequal lives: a sociodemographic analysis of COVID-19 transmission and mortality in India

ObjectivesExisting socio-economic inequalities shape, in very particular and measurable ways, the differential impact that a disease has on different sections of the same society. This is particularly true of COVID-19, which has rapidly exhausted the public health system in India, and magnified the gradient of vulnerability in an underserved populace. Using publicly available data, we have aimed to deconstruct this gradient into individual variables of inequality and quantify their impact on the transmission and mortality outcomes of COVID-19 in India. Study designSociodemographic analysis. MethodsWe quantify doubling times and case fatality ratios for all districts in India, then correlate them to 20 variables of socio-economic vulnerability and demographic structure. Variables that exhibit persistent correlation are then analysed using multivariate beta regression models to validate their impact on COVID-19 outcomes in India. ResultsThe transmission of COVID-19 in India is enhanced by the lack of access to indoor latrines, drainage facilities, electricity, and proximate sources of drinking water. Transmission is slowed by the presence of an elderly population. Fatality rates relate negatively to an area's medical infrastructure and the presence of a college-educated populace. ConclusionsAn interactive matrix of social inequalities, cultural practices, and behavioural patterns determines the path of COVID-19 through a community. Specific variables exhibit patterns of persistent vulnerability; others indicate a resistance to infection and mortality. This body of evidence, when incorporated into policy design, may lead to localised, need-sensitive models of intervention, both for preventive measures and medical care.

Limitations introduced by a low participation rate of SARS-CoV-2 seroprevalence data.

There has been a large influx of COVID-19 seroprevalence studies, but comparability between the seroprevalence estimates has been an issue because of heterogeneities in testing platforms and study methodology. One potential source of heterogeneity is the response or participation rate. We conducted a review of participation rates (PR) in SARS-CoV-2 seroprevalence studies collected by SeroTracker and examined their effect on the validity of study conclusions. PR was calculated as the count of participants for whom the investigators had collected a valid sample, divided by the number of people invited to participate in the study. A multivariable beta generalized linear model with logit link was fitted to determine if the PR of international household and community-based seroprevalence studies was associated with the factors of interest, from 1 December 2019 to 10 March 2021. We identified 90 papers based on screening and were able to calculate the PR for 35 out of 90 papers (39%), with a median PR of 70% and an interquartile range of 40.92; 61% of the studies did not report PR. Many SARS-CoV-2 seroprevalence studies do not report PR. It is unclear what the median PR rate would be had a larger portion not had limitations in reporting. Low participation rates indicate limited representativeness of results. Non-probabilistic sampling frames were associated with higher participation rates but may be less representative. Standardized definitions of participation rate and data reporting necessary for the PR calculations are essential for understanding the representativeness of seroprevalence estimates in the population of interest.

Association between environmental air pollution and olfactory functioning among Italian adolescents and young adults in Northern Italy

BACKGROUND AND AIM Olfaction is a critical physiological process with various functions in humans and olfactory impairments are associated with neurodegenerative disease and decreased quality of life. Olfactory receptors are in direct contact to ambient air, and thus, olfactory function is uniquely vulnerable to airborne pollutants. Here, we investigate associations between ambient air pollutant exposure and olfactory function in adolescents and young adults in the province of Brescia, Italy. METHODS This study included 207 subjects (52.7% female, 15-25 years) enrolled in the ongoing longitudinal Public Health Impact of Metals Exposure (PHIME) study. All participants completed a baseline (2008-2014) and follow-up (2017-2021) olfactory performance assessment using the &amp;#x22;Sniffin’ Sticks”, a psychophysical test to assess odor threshold, discrimination and identification. Spatial annual average concentrations of particulate matter (PM) PM10, PM2.5 and nitrogen dioxide (NO2) were estimated for the target area between 2016-2019 on a 4 x 4 km2 spatial resolution grid by means of the Chemical Transport Model ARIA Regional. We applied multivariable beta regression models to test for associations between each pollutant and olfactory functions. RESULTS All air pollution annual average values were below European air quality standards (mean +/- SD g/m3 PM2.5 = 18.4 +/- 5.1; PM10 = 21.0 +/- 5.3; NO2 = 32.1 +/- 9.8). We observed a significant negative association between NO2 and olfactory threshold score; a 14% decrement (approximately 2 score points) for each 10 unit increase in NO2 (p = 0.042), adjusting for age, sex, Sniffin’ identification baseline score, having a cold and/or allergic rhinitis, IQ and passive smoking exposure. CONCLUSIONS These results suggest increased NO2 exposure is associated with reduced olfactory function in these Italian adolescents and young adults. The identification of risk factors for olfactory functions can help prevent reduced quality of life and improve recovery. KEYWORDS Olfactory functions; air pollution; NO2; PM2.5; PM10

Spatial clusters, social determinants of health and risk of COVID-19 mortality in Brazilian children and adolescents: A nationwide population-based ecological study

BackgroundData regarding the geographical distribution of cases and risk factors for COVID-19 death in children and adolescents are scarce. We describe the spatial distribution of COVID-19 cases and deaths in paediatric population and their association with social determinants of health in Brazil.MethodsThis is a population-based ecological study with a spatial analysis of all cases and deaths due to COVID-19 in Brazil among children and adolescents aged 0–19 years from March 2020 to October 2021. The units of analysis were the 5570 municipalities. Data on COVID-19 cases and deaths, social vulnerability, health inequities, and health system capacity were obtained from publicly available databases. Municipalities were stratified from low to very high COVID-19 incidence and mortality using K-means clustering procedures, and spatial clusters and relative risks were estimated using spatial statistics with Poisson probability models. The relationship between COVID-19 estimates and social determinants of health was explored by using multivariate Beta regression techniques.FindingsA total of 33,991 COVID-19 cases and 2424 deaths among children and adolescents aged 0–19 years were recorded from March 2020 to October 2021. There was a spatial dependence for the crude mortality coefficient per 100,000 population in the paediatric population aged 0–19 years (I Moran 0·10; P < 0·001). Forty municipalities had higher mortality rates, of which 20 were in states from the Northeast region. Seven spatial clusters were identified for COVID-19 mortality, with four clusters in the Northeast region and three in the North region. Municipalities with higher social inequality and vulnerability had higher COVID-19 mortality in the paediatric population.InterpretationThe main clusters of risk for mortality among children and adolescents were identified in municipalities in the North and Northeast regions, which are the regions with the worst socioeconomic indicators and greatest health disparities in the country. Our findings confirmed the higher burden of COVID-19 for Brazilian paediatric population in municipalities with higher social inequality and vulnerability and worse socioeconomic indicators. To reduce the burden of COVID-19 on children, mass immunisation is necessary.FundingNone.

A Non-parametric Bayesian Learning Model Using Accelerated Variational Inference on Multivariate Beta Mixture Models for Medical Applications

Clustering as an exploratory technique has been a promising approach for performing data analysis. In this paper, we propose a non-parametric Bayesian inference to address clustering problem. This approach is based on infinite multivariate Beta mixture models constructed through the framework of Dirichlet process. We apply an accelerated variational method to learn the model. The motivation behind proposing this technique is that Dirichlet process mixture models are capable to fit the data where the number of components is unknown. For large-scale data, this approach is computationally expensive. We overcome this problem with the help of accelerated Dirichlet process mixture models. Moreover, the truncation is managed using kd-trees. The performance of the model is validated on real medical applications and compared to three other similar alternatives. The results show the outperformance of our proposed framework.

Health-related quality of life of female sex workers living with HIV in South Africa: a cross-sectional study.

IntroductionHealth‐related quality of life (HRQoL) is an important HIV outcome beyond viral suppression. However, there are limited data characterizing HRQoL of key populations, including female sex workers (FSW) living with HIV.MethodsWe used baseline data (22 June 2018–23 March 2020) of FSW who were diagnosed with HIV and enrolled into a randomized trial in Durban, South Africa. HRQoL information was collected by a generic preference‐accompanied tool with five domains (EQ‐5D), and summarized into a single score (range 0–1), which represents health utility. We employed multivariable beta regression models to identify determinants of HRQoL and to estimate subgroup‐specific HRQoL score. Using external estimates of life expectancy and population size, we estimated the number of quality adjusted life years reduced among FSW living with HIV in South Africa associated with violence and drug use.ResultsOf 1,363 individuals (mean age: 32.4 years; mean HRQoL score: 0.857) in our analysis, 62.6% used drugs, 61.3% experienced physical or sexual violence and 64.6% self‐reported taking antiretroviral treatment (ART). The following were associated with a reduction in the average marginal HRQoL score: older age (per decade: 0.018 [95% confidence interval (CI): 0.008, 0.027]), drug use (0.022 [0.007, 0.036]), experience of violence (0.024 [0.010, 0.038]) and moderate (vs. no) level of internalized stigma (0.023 [0.004, 0.041]). Current ART use was associated with a 0.015‐point (–0.001, 0.031) increase in the HRQoL score. The estimated mean (95% CI) HRQoL scores ranged from 0.838 (0.816, 0.860) for FSW who used drugs, experienced violence and were not on ART; to 0.899 (0.883, 0.916) for FSW who did not use drugs nor experience violence and were on ART. Our results can be translated into a reduction in 37,184 and 39,722 quality adjusted life years related to drug use and experience of violence, respectively, in South Africa.ConclusionsThese results demonstrate the association of ART with higher HRQoL among FSW and the need to further address structural risks, including drug use, violence and stigma. Population‐specific estimates of HRQoL score can be further used to calculate quality‐adjusted life years in economic evaluations of individual and structural interventions addressing the needs of FSW living with HIV.Clinical Trial RegistrationNCT03500172 (April 17, 2018).

Expectation propagation learning of finite multivariate Beta mixture models and applications

Expectation propagation learning of finite multivariate Beta mixture models and applications

A Nomogram Predicting Early Cervical Cancer Distant Recurrence

Distant recurrence is the most common cause of treatment failure in locally advanced cervical cancer (LACC) following chemoradiation (CRT). We explored risk factors and developed a clinical nomogram predicting distant recurrence within 6 months of CRT completion.This was a retrospective case-control study of LACC patients treated with definitive CRT at a tertiary academic center from 1997-2020. Patients underwent FDG-PET at initial staging and approximately 3 months post treatment and had ≥ 3-month clinical follow-up. Patients were prescribed 50.4 Gy external beam radiation (EBRT) to the pelvic ± para-aortic (PA) lymph nodes (PA if involved on PET staging); both brachytherapy and cisplatin were delivered concurrently with EBRT. Patients who developed distant recurrence within 6 months were designated as cases and all others as controls. The dataset was randomly divided into 2/3 training and 1/3 testing. Binary logistic regression was applied to training data to determine significant predictor variables. Continuous variables were also assessed as categorical variables with cut points determined by Youden's method. Patients with missing data for significant predictors were excluded (110 training, 54 testing). The best model was defined by highest c-statistic and was applied to the testing data set.514 patients were identified of whom 45 (9%) developed distant recurrence within 6 months of CRT completion (15 biopsy proven, 30 clinically determined); 44 (98%) identified on 3 month post treatment PET. The dataset was divided into training (20 cases, 214 controls) and testing (10 cases, 106 controls). Age, treatment duration, histology (87% squamous), cervix SUVmax, pre/post squamous cell antigen did not differ across cases and controls in both training and testing data. Logistic regression applied to training data showed histology grade (poorly vs well/mod differentiated), PET positive lymph nodes (pelvic/para-aortic vs none) and PET initial tumor volume (≥50 vs < 50 cm3) had the most predictive multivariate beta coefficients (1.2, P = 0.025; 1.2, P = 0.016; 1.3, P = 0.028 respectively) with an overall c-statistic of 0.76. In the nomogram, each variable was assigned one point resulting in a total score range of 0-3 points, with higher scores indicating increasing risk. 3 points carried 29% positive predictive value (PPV), and 0 points indicated 100% negative predictive value (NPV) (Table 1). This nomogram was applied to the testing data and showed c-statistic 0.82 and 3 points PPV of 27% and 0 points NPV of 100%.More than a quarter of LACC patients with poorly differentiated histology, PET positive lymph nodes, and initial primary tumor volume greater than 50 cm3 experience distant recurrence within 6 months of completing CRT.

Batch and online variational learning of hierarchical Dirichlet process mixtures of multivariate Beta distributions in medical applications

Batch and online variational learning of hierarchical Dirichlet process mixtures of multivariate Beta distributions in medical applications

A hidden Markov model reliably characterizes ketamine-induced spectral dynamics in macaque local field potentials and human electroencephalograms.

Ketamine is an NMDA receptor antagonist commonly used to maintain general anesthesia. At anesthetic doses, ketamine causes high power gamma (25-50 Hz) oscillations alternating with slow-delta (0.1-4 Hz) oscillations. These dynamics are readily observed in local field potentials (LFPs) of non-human primates (NHPs) and electroencephalogram (EEG) recordings from human subjects. However, a detailed statistical analysis of these dynamics has not been reported. We characterize ketamine’s neural dynamics using a hidden Markov model (HMM). The HMM observations are sequences of spectral power in seven canonical frequency bands between 0 to 50 Hz, where power is averaged within each band and scaled between 0 and 1. We model the observations as realizations of multivariate beta probability distributions that depend on a discrete-valued latent state process whose state transitions obey Markov dynamics. Using an expectation-maximization algorithm, we fit this beta-HMM to LFP recordings from 2 NHPs, and separately, to EEG recordings from 9 human subjects who received anesthetic doses of ketamine. Our beta-HMM framework provides a useful tool for experimental data analysis. Together, the estimated beta-HMM parameters and optimal state trajectory revealed an alternating pattern of states characterized primarily by gamma and slow-delta activities. The mean duration of the gamma activity was 2.2s([1.7,2.8]s) and 1.2s([0.9,1.5]s) for the two NHPs, and 2.5s([1.7,3.6]s) for the human subjects. The mean duration of the slow-delta activity was 1.6s([1.2,2.0]s) and 1.0s([0.8,1.2]s) for the two NHPs, and 1.8s([1.3,2.4]s) for the human subjects. Our characterizations of the alternating gamma slow-delta activities revealed five sub-states that show regular sequential transitions. These quantitative insights can inform the development of rhythm-generating neuronal circuit models that give mechanistic insights into this phenomenon and how ketamine produces altered states of arousal.

The relationship between motor vehicle speed and active school transportation at elementary schools in Calgary and Toronto, Canada

BackgroundHigher traffic speeds increase the risk of serious or fatal injury to child pedestrians and may deter parents from encouraging their children to walk to school. This cross-sectional study examined the association between measured vehicle speed and active school transportation. MethodsThe study sample included public elementary schools in Calgary (n = 46) and Toronto (n = 42), Canada. Speed limits in school zones were 30 km/h in Calgary but ranged from 30 to 60 km/h in Toronto. In May and June 2018, trained observers counted children arriving to school and recorded their mode of travel. Vehicle speeds around schools were measured between 07:30 and 18:00 (school activity hours) and quantified as the 85th percentile speed, the percentage of vehicles exceeding 30 km/h, and the percentage of vehicles exceeding the speed limit. Multivariable beta regression was used to examine the relationship between vehicle speed and active transportation to school for each city, adjusting for school environment features. ResultsA high percentage of vehicles were travelling over the speed limit in front of schools: 45% in Calgary and 42% in Toronto. No association between vehicle speed and active school transportation was found in Calgary (mean 85th percentile speed OR: 0.99, 95% CI: 0.93, 1.06). There was an inverse association between vehicle speed and active school transportation in Toronto, where for every 1 km/h increase in the mean 85th percentile speed, the odds of active school transportation decreased by 3% (OR: 0.97, 95% CI: 0.95, 0.99). ConclusionThere was a high prevalence of speeding in front of schools, which could increase the risk of traffic-related injuries and fatalities to children. Speed limit reductions to 30 km/h on roads near schools may be a useful preventive intervention in Toronto during school activity hours. Such targeted interventions can help improve child pedestrian safety and promote active school transportation.

Online variational inference on finite multivariate Beta mixture models for medical applications

Online variational inference on finite multivariate Beta mixture models for medical applications

Understanding eating choices among university students: A study using data from cafeteria cashiers’ transactions

ObjectiveTo illustrate the use of automatically collected data from cashier transactions to understand eating habits among university students using cafeteria and to identify individual characteristics associated with the diverse behaviors. MethodsThe study was carried out at a large university located in Pisa, central Italy, using data about meals automatically recorded from cashier transaction meals during the academic year 2015−16 as well as data from the administrative archive of the university. A model-based clustering relying on multivariate beta distribution was used to cluster eating choices while multivariate multinomial logistic regressions were applied to identify variables associated to diverse clusters identified. ResultsConsidering 4643 students and about 200,000 meals consumed, results suggest that healthy eaters represented a minority (11.2 %) of the study population while the large part of students composed their meals combining grains with processed food or proteins (32.7 %) and limiting the choice of fruit (42.9 %). Male gender and younger age were associated with eating behavior not in line with recommendations for a healthy diet. ConclusionsEating choice resulted to be “compromised” in most of students and specific characteristics associated with unhealthy choice were also identified that can help inform and target specific policy. The use of routinely collected data gives the opportunity to both cafeterias and university to take an active role in policy development.

A ratio of finitely many gamma functions and its properties with applications

In the paper, the authors establish an inequality involving exponential functions and sums, introduce a ratio of finitely many gamma functions, discuss properties, including monotonicity, logarithmic convexity, (logarithmically) complete monotonicity, and the Bernstein function property, of the newly introduced ratio, and construct two inequalities of multinomial coefficients and multivariate beta functions.

Nonparametric variational learning of multivariate beta mixture models in medical applications

AbstractClustering as an essential technique has matured into a capable solution to address the gap between the growing availability of data and deriving the knowledge from them. In this paper, we propose a novel clustering method “variational learning of infinite multivariate Beta mixture models.” The motivation behind proposing this technique is the flexibility of mixture models to fit the data. This approach has the capability to infer the model complexity and estimate model parameters from the observed data automatically. Moreover, as a label‐free method, it could also address the problem of high costs of medical data labeling, which can be undertaken just by medical experts. The performance of the model is evaluated on real medical applications and compared with other similar alternatives. We demonstrate the ability of our proposed method to outperform widely used methods in the field as it has been shown in experimental results.