Category: Statistics

J R Soc Interface. 2023 Jun;20(203):20220877. doi: 10.1098/rsif.2022.0877. Epub 2023 Jun 21.

ABSTRACT

With a view towards artificial cells, molecular communication systems, molecular multiagent systems and federated learning, we propose a novel reaction network scheme (termed the Baum-Welch (BW) reaction network) that learns parameters for hidden Markov models (HMMs). All variables including inputs and outputs are encoded by separate species. Each reaction in the scheme changes only one molecule of one species to one molecule of another. The reverse change is also accessible but via a different set of enzymes, in a design reminiscent of futile cycles in biochemical pathways. We show that every positive fixed point of the BW algorithm for HMMs is a fixed point of the reaction network scheme, and vice versa. Furthermore, we prove that the ‘expectation’ step and the ‘maximization’ step of the reaction network separately converge exponentially fast and compute the same values as the E-step and the M-step of the BW algorithm. We simulate example sequences, and show that our reaction network learns the same parameters for the HMM as the BW algorithm, and that the log-likelihood increases continuously along the trajectory of the reaction network.

PMID:37340782 | DOI:10.1098/rsif.2022.0877

J Audiol Otol. 2023 Jun 22. doi: 10.7874/jao.2022.00542. Online ahead of print.

ABSTRACT

BACKGROUND AND OBJECTIVES: : This study was aimed at assessing recovery from otitis media (OM) and variables associated with it among 1- to 6-year-old children.

SUBJECTS AND METHODS: : We assessed 87 children with OM otologically and audiologically. Medicines were prescribed, and medication compliance was ensured. The children were followed up after 3 months to judge the status of OM as resolved or recurrent. Data were statistically analyzed to derive the risk of recurrence of OM with effusion (OME) and acute OM by degree of hearing loss, type of tympanogram, age group, and sex.

RESULTS: : The overall recurrence rate was 26%. The risk of recurrence was higher for OME (odds ratio [OR]=4.33; 95% confidence interval [CI]: 1.90 to 9.83); at AC auditory brainstem peak V responses up to 40 dBnHL (OR=5.20; 95% CI: 2.05 to 13), 50 dBnHL (OR=3.47; 95% CI: 0.5 to 23), and 60 dBnHL (OR=16.09; 95% CI: 4.36 to 1.2); in B (OR= 3.16; 95% CI: 1.36 to 7.33) and C tympanograms (OR=2.83; 95% CI: 0.70 to 11.41); and in the age group of 5-6 years (OR=8, 95% CI: 2.23 to 28). The risk of recurrence of OM did not differ between male and female patients.

CONCLUSIONS: : The rate of recurrence was comparable to or lower than that reported in the pediatric population of other countries. The findings suggest that children with OME, severe pathology, or age of 5-6 years require more attention and frequent monitoring to minimize the risk of recurrence.

PMID:37340774 | DOI:10.7874/jao.2022.00542

BMJ Mil Health. 2023 Jun 20:e002431. doi: 10.1136/military-2023-002431. Online ahead of print.

NO ABSTRACT

PMID:37339828 | DOI:10.1136/military-2023-002431

Addiction. 2023 Jun 20. doi: 10.1111/add.16278. Online ahead of print.

ABSTRACT

BACKGROUND AND AIMS: Reduction of alcohol consumption is important for people undergoing treatment for HIV. We tested the efficacy of a brief intervention for reducing the average volume of alcohol consumed among patients on HIV antiretroviral therapy (ART).

DESIGN, SETTING AND PARTICIPANTS: This study used a two-arm multi-centre randomized controlled trial with follow-up to 6 months. Recruitment occurred between May 2016 and October 2017 at six ART clinics at public hospitals in Tshwane, South Africa. Participants were people living with HIV, mean age 40.8 years [standard deviation (SD) = 9.07], 57.5% female, and on average 6.9 years (SD = 3.62) on ART. At baseline (BL), the mean number of drinks consumed over the past 30 days was 25.2 (SD = 38.3). Of 756 eligible patients, 623 were enrolled.

INTERVENTION: Participants were randomly assigned to a motivational interviewing (MI)/problem-solving therapy (PST) intervention arm (four modules of MI and PST delivered over two sessions by interventionists) or a treatment as usual (TAU) comparison arm. People assessing outcomes were masked to group assignment.

MEASUREMENTS: The primary outcome was the number of standard drinks (15 ml pure alcohol) consumed during the past 30 days assessed at 6-month follow-up (6MFU).

FINDINGS: Of the 305 participants randomized to MI/PST, 225 (74%) completed the intervention (all modules). At 6MFU, retention was 88% for the control and 83% for the intervention arm. In support of the hypothesis, an intention-to-treat-analysis for the primary outcome at 6MFU was -0.410 (95% confidence interval = -0.670 to -0.149) units lower on log scale in the intervention group than in the control group (P = 0.002), a 34% relative reduction in the number of drinks. Sensitivity analyses were undertaken for patients who had alcohol use disorders identification test (AUDIT) scores ≥ 8 at BL (n = 299). Findings were similar to those of the whole sample.

CONCLUSIONS: In South Africa, a motivational interviewing/problem-solving therapy intervention significantly reduced drinking levels in HIV-infected patients on antiretroviral therapy at 6-month follow-up.

PMID:37339811 | DOI:10.1111/add.16278

Life Sci Alliance. 2023 Jun 20;6(9):e202201550. doi: 10.26508/lsa.202201550. Print 2023 Sep.

ABSTRACT

The identification of condition-specific gene sets from transcriptomic experiments is important to reveal regulatory and signaling mechanisms associated with a given cellular response. Statistical methods of differential expression analysis, designed to assess individual gene variations, have trouble highlighting modules of small varying genes whose interaction is essential to characterize phenotypic changes. To identify these highly informative gene modules, several methods have been proposed in recent years, but they have many limitations that make them of little use to biologists. Here, we propose an efficient method for identifying these active modules that operates on a data embedding combining gene expressions and interaction data. Applications carried out on real datasets show that our method can identify new groups of genes of high interest corresponding to functions not revealed by traditional approaches. Software is available at https://github.com/claudepasquier/amine.

PMID:37339804 | DOI:10.26508/lsa.202201550

J Am Med Dir Assoc. 2023 Jun 17:S1525-8610(23)00478-4. doi: 10.1016/j.jamda.2023.05.014. Online ahead of print.

ABSTRACT

OBJECTIVES: Anticholinergic burden is detrimental to cognitive health. Multiple studies found that a high anticholinergic burden is associated with an increased risk for dementia, changes to the brain structure, function, and cognitive decline. We performed a post hoc analysis of a randomized controlled deprescribing trial. We compared the effect of the intervention on baseline anticholinergic burden across the treatment and control groups and the time of recruitment before and after a lockdown due to the COVID pandemic with subgroup analyses by baseline frailty index.

DESIGN: Randomized controlled trial.

SETTINGS AND PARTICIPANTS: We analyzed data from a de-prescribing trial of older adults (>65 years) previously conducted in New Zealand that was focused on reducing the Drug Burden Index (DBI).

METHODS: We used the anticholinergic cognitive burden (ACB) to quantify the impact of the intervention on reducing the anticholinergic burden. Participants not taking anticholinergics at the start of the trial were excluded. The primary outcome for this subgroup analysis was a change in ACB, measured with the ĝ_Hedges statistic describing the difference in standard deviation units of this change between intervention and control. For this analysis, the trial participants were stratified into low, medium, and high frailty and timing into prior- and post-lockdown (public health measures for COVID-19).

RESULTS: Among the 295 participants in this analysis, the median (IQR) age was 79 (74, 85), and 67% were women. For the primary outcome ĝ_Hedges = -0.04 (95% CI -0.26 to 0.19) with a -0.23 mean reduction in ACB in the intervention arm and -0.19 in the control arm. Before lockdown ĝ_Hedges = -0.38 (95% CI -0.84 to 0.04) and post-lockdown ĝ_Hedges = 0.07 (95% CI -0.19 to 0.33). The mean change in ACB for each of the frailty strata was as follows: low frailty (-0.02; 95% CI -0.65 to 0.18); medium frailty (0.05; 95% CI -0.28 to 0.38); high frailty (0.08; 95% CI -0.40 to 0.56).

CONCLUSIONS AND IMPLICATIONS: The study did not provide evidence for the effect of pharmacist deprescribing intervention on reducing the anticholinergic burden. However, this post hoc analysis examined the impact of COVID on the effectiveness of the intervention, and further research in this area may be warranted.

PMID:37339754 | DOI:10.1016/j.jamda.2023.05.014

Int J Antimicrob Agents. 2023 Jun 18:106895. doi: 10.1016/j.ijantimicag.2023.106895. Online ahead of print.

ABSTRACT

INTRODUCTION: Although approximately 40 years have passed since Japanese spotted fever (JSF) was first reported in Japan, its treatment has not been standardized yet. As in other rickettsial infections, tetracycline (TC) is the first-line treatment, but successful cases of fluoroquinolone (FQ) combination therapy in severe cases have been reported. However, the effectiveness of TC plus FQ combined treatment (TC+FQ) remains controversial. Therefore, we evaluated the antipyretic effect of TC+FQ.

METHODS: A comprehensive search of published JSF case reports was conducted to extract individual patient data. In the cases where we were able to extract temperature data, after homogenizing patient characteristics, we evaluated the time-dependent changes in the fever type from the date of the first visit for the TC and TC+FQ groups.

RESULTS: The primary search yielded 182 cases, with the individual data evaluations resulting in a final analysis of 102 cases (84 in the TC group and 18 in the TC+FQ group), which included the temperature data. The TC+FQ group had significantly lower body temperatures as compared to the TC group from day 3 to day 4.

CONCLUSIONS: Although TC monotherapy for JSF can produce defervescence after a while, the fever duration is longer as compared to other rickettsial infections, such as scrub typhus. Results suggest that TC+FQ was more effective for the antipyretic effect, with a potential shortening of the duration that patients suffer from the febrile symptoms.

PMID:37339710 | DOI:10.1016/j.ijantimicag.2023.106895

J Shoulder Elbow Surg. 2023 Jun 18:S1058-2746(23)00456-1. doi: 10.1016/j.jse.2023.05.018. Online ahead of print.

ABSTRACT

BACKGROUND: Current treatment paradigms for anterior shoulder instability are based on radiological measurements of glenohumeral bone defects, and mathematical calculation of glenoid track is used to classify lesions into “on-track”, and “off-track” morphology. However, radiological measurements have shown high variability, and glenoid track widths under dynamic conditions have been reported to be significantly smaller than those under static radiological conditions. The purpose of this study was to assess the reliability, reproducibility, and diagnostic validity of dynamic arthroscopic standardized tracking (DAST) in comparison with the gold-standard radiological track measurement method for identification of on/off-track bony lesions in patients with anteroinferior shoulder instability METHODS: Between January 2018 and 2022, 114 patients with traumatic anterior shoulder instability were evaluated using 3T MRI or CT scans; glenoid bone loss (GBL), Hill-Sachs interval (HSI), glenoid track (GT), and Hill-Sach’s occupancy ratio (HSO) were measured, and defects were classified as on- or off-track, and peripheral-track (HSO%) by two independent researchers. During arthroscopy, a standardized method (DAST) was used to classify defects into on-track (central and peripheral), and off-track by two independent observers. Interobserver reliability for DAST and radiological method was calculated using Kappa statistics and reported as percent agreement. Diagnostic validity (sensitivity, specificity, positive predictive value, and negative predictive value) of DAST was calculated using the radiological (HSO%) track as gold standard.

RESULTS: Radiologically measured mean GBL, HSI, GT, and HSO for off-track lesions were lower in arthroscopically classified off-track lesions (DAST) as compared with radiological method. DAST showed near-perfect agreement between 2 observers for on-off track classification (k=0.96, p<0.001) as well as for on-peripheral-off track classification (k=0.88, p<0.001). Radiological method showed greater interobserver variability (0.31, 0.24) with only fair agreement for both classifications. Inter-method agreement varied between 71% and 79% (CI 62-86%) between the 2 observers, and reliability was assessed as slight to fair (K=0.38, 0.16). Overall, for identification of an off-track lesion, DAST showed maximum specificity (81%, 78%) when radiological peripheral-track lesions (HSO% 75-100) were considered as off-track, and maximum sensitivity when arthroscopic peripheral track lesions were classified as off-track.

CONCLUSION: Although the intermethod agreement was low, a standardized arthroscopic tracking method (DAST) showed superior interobserver agreement and reliability for lesion classification in comparison with radiological track method. Incorporating DAST in current algorithms may help reduce variability in surgical decision-making.

PMID:37339701 | DOI:10.1016/j.jse.2023.05.018

Neuroimage. 2023 Jun 18:120228. doi: 10.1016/j.neuroimage.2023.120228. Online ahead of print.

ABSTRACT

Functional gradients, in which response properties change gradually across a brain region, have been proposed as a key organising principle of the brain. Recent studies using both resting-state and natural viewing paradigms have indicated that these gradients may be reconstructed from functional connectivity patterns via “connectopic mapping” analyses. However, local connectivity patterns may be confounded by spatial autocorrelations artificially introduced during data analysis, for instance by spatial smoothing or interpolation between coordinate spaces. Here, we investigate whether such confounds can produce illusory connectopic gradients. We generated datasets comprising random white noise in subjects’ functional volume spaces, then optionally applied spatial smoothing and/or interpolated the data to a different volume or surface space. Both smoothing and interpolation induced spatial autocorrelations sufficient for connectopic mapping to produce both volume- and surface-based local gradients in numerous brain regions. Furthermore, these gradients appeared highly similar to those obtained from real natural viewing data, although gradients generated from real and random data were statistically different in certain scenarios. We also reconstructed global gradients across the whole-brain – while these appeared less susceptible to artificial spatial autocorrelations, the ability to reproduce previously reported gradients was closely linked to specific features of the analysis pipeline. These results indicate that previously reported gradients identified by connectopic mapping techniques may be confounded by artificial spatial autocorrelations introduced during the analysis, and in some cases may reproduce poorly across different analysis pipelines. These findings imply that connectopic gradients need to be interpreted with caution.

PMID:37339700 | DOI:10.1016/j.neuroimage.2023.120228

Am J Perinatol. 2023 Jun 20. doi: 10.1055/a-2113-8528. Online ahead of print.

ABSTRACT

OBJECTIVE: Cerebral sonography through the anterior fontanelle is a neonatal brain imaging method that has become an integral part of modern neonatal bedside care for both screening and diagnostic purposes. Premature infants with cognitive delay have a reduction in cerebellar size at term corrected age on MRI. We aimed to determine the level of agreement between postnatal MRI and cerebral sonography for cerebellar biometry and to determine the level of agreement within one and between different examiners.

STUDY DESIGN: Cerebellar sonography and MRI measurements of the cerebellum from 30 term infants was assessed by Bland-Altman plots. Measurements between both modalities were compared using Wilcoxon Signed-Rank Test. A p-value < 0.01 was considered statistically significant. Intra-class correlation coefficients (ICC) for intra-rater and inter-rater reliability of cerebral sonography measurements were calculated.

RESULTS: There was no statistically significant difference between cerebral sonography and MRI for linear measurements, but the measurements of perimeter and surface area differed significantly between the two techniques. There was a systematic bias between both modalities for most measurements except for anterior-posterior width and vermis height. For measurements that were not statistically different from MRI we found excellent intra-rater ICC for the anterior-posterior width, vermis height and cerebellar width. The inter-rater ICC was excellent for the anterior-posterior width and vermis height, but poor for the transverse cerebellar width.

CONCLUSIONS: With a strict imaging protocol cerebellar measurements of the anterior-posterior width and the vermis height can be used as an alternative to MRI for diagnostic-screening purposes in a neonatal department where multiple clinicians perform bedside cerebral sonography.

PMID:37339678 | DOI:10.1055/a-2113-8528