Introduction to statistical modelling: linear regression

How fragile the positive results of Chinese herbal medicine randomized controlled trials on irritable bowel syndrome are?

OBJECTIVE

The fragility index (FI), which is the minimum number of changes in status from "event" to "non-event" resulting in a loss of statistical significance, serves as a significant supplementary indicator for clinical physicians in interpreting clinical trial results and aids in understanding the outcomes of randomized controlled trials (RCTs). In this systematic literature survey, we evaluated the FI for RCTs evaluating Chinese herbal medicine (CHM) for irritable bowel syndrome (IBS), and explored potential associations between study characteristics and the robustness of RCTs.

METHODS

A comprehensive search was conducted in four databases in Chinese and four databases in English from their inception to January 1, 2023. RCTs encompassed 1:1 ratio into two parallel arms and reported at least one binary outcome that demonstrated statistical significance were included. FI was calculated by the iterative reduction of a target outcome event in the treatment group and concomitant subtraction of a non-target event from that group, until positive significance (defined as P < 0.05 by Fisher's exact test) is lost. The lower the FI (minimum 1) of a trial outcome, the more fragile the positive result of the outcome was. Linear regression models were adopted to explore influence factors of the value of FI.

RESULTS

A total of 30 trials from 2 4118 potentially relevant citations were finally included. The median FI of total trials included was 1.5 (interquartile range [IQR], 1-5), and half of the trials (n = 15) had a FI equal to 1. In 12 trials (40%), the total number of participants lost to follow-up surpassed the respective FI. The study also identified that increased FI was significantly associated with no TCM syndrome differentiation for inclusion criteria of the patients, larger total sample size, low risk of bias, and larger numbers of events.

CONCLUSIONS

The majority of CHM IBS RCTs with positive results were found to be fragile. Ensuring adequate sample size, scientifically rigorous study design, proper control of confounding factors, and a quality control calibration for consistency of TCM diagnostic results among clinicians should be addressed to increase the robustness of the RCTs. We recommend reporting the FI as one of the components of sensitivity analysis in future RCTs to facilitate the assessment of the fragility of trials.

A narrative review of the application of machine learning in venous thromboembolism

OBJECTIVE

To summarize the current research progress of machine learning and venous thromboembolism.

METHODS

The literature on risk factors, diagnosis, prevention and prognosis of machine learning and venous thromboembolism in recent years was reviewed.

RESULTS

Machine learning is the future of biomedical research, personalized medicine, and computer-aided diagnosis, and will significantly promote the development of biomedical research and healthcare. However, many medical professionals are not familiar with it. In this review, we will introduce several commonly used machine learning algorithms in medicine, discuss the application of machine learning in venous thromboembolism, and reveal the challenges and opportunities of machine learning in medicine.

CONCLUSION

The incidence of venous thromboembolism is high, the diagnostic measures are diverse, and it is necessary to classify and treat machine learning, and machine learning as a research tool, it is more necessary to strengthen the special research of venous thromboembolism and machine learning.

Can feature structure improve model's precision? A novel prediction method using artificial image and image identification

Objectives

This study aimed to develop an approach to enhance the model precision by artificial images.

Materials and Methods

Given an epidemiological study designed to predict 1 response using f features with M samples, each feature was converted into a pixel with certain value. Permutated these pixels into F orders, resulting in F distinct artificial image sample sets. Based on the experience of image recognition techniques, appropriate training images results in higher precision model. In the preliminary experiment, a binary response was predicted by 76 features, the sample set included 223 patients and 1776 healthy controls.

Results

We randomly selected 10 000 artificial sample sets to train the model. Models' performance (area under the receiver operating characteristic curve values) depicted a bell-shaped distribution.

Conclusion

The model construction strategy developed in the research has potential to capture feature order related information and enhance model predictability.

Daily activity patterns in older adults receiving initial support: the association between daily steps and sitting in bouts of at least 60 min

BACKGROUND

Aging has a significant impact on health, underlining the importance of maintaining physical function and reducing time spent sitting among older adults. To understand how to reduce prolonged sitting or increase physical activity, factors related to the daily living and observed daily activity patterns should be explored. This study aimed to investigate the association between daily steps, self-rated health, physical activity, sedentary behavior, motivation to exercise and fear of falling among older adults receiving initial support.

METHOD

Cross-sectional design with total population questionnaire data from adults aged ≥ 60 years (n = 917), living at home with initial support from municipal care in southern Sweden. The older adults were offered to participate in a follow-up study measuring daily activity patterns with accelerometers (n = 72). Linear regression was used to analyze associations between daily steps and possible predictors.

RESULTS

The linear model ([Formula: see text]0.478) showed that sitting in unbroken bouts of > 60 min (β = -0.313, p < 0.05), walking independently outdoors (β = 0.301, p < 0.05), intending to increase physical activity (β = -0.294, p < 0.05), sex (β = 0.279, p < 0.05), relative autonomy index (β = 0.258, p < 0.05), fear of falling (β = -0.238, p < 0.05), and self-rated health (β = 0.213, p < 0.05) predicted daily steps.

CONCLUSION

The model of predictors brings new understanding regarding daily steps among community-dwelling older adults. The association between sitting in bouts of > 60 min and daily steps is interesting as 35% of participants had a number of sitting bouts that on average, showed 30% less steps taken. Minimizing long sitting bouts and maintaining physical functioning to promote independence when walking outdoors can be tools for clinical practitioners devising interventions to break prolonged sitting among community-dwelling older adults. Future research should prioritize studying older adults' outdoor walking independence, including its relation to walking with or without assistive devices and its impact on physical activity and sedentary behavior.

Remote vs. face-to-face activities in the teaching of renal pathophysiology in the context of social isolation during the early phase of the COVID-19 pandemic

The advent of the COVID-19 pandemic forced medical schools around the world to adopt emergency remote learning as a resort to avoid interruption of courses. However, the effectiveness of online classes as an educational strategy has been questioned by medical educators and students. In a prospective observational study design, students enrolled in a renal physiology and pathophysiology course were exposed to either face-to-face or remote synchronous classes. Students taught online obtained significantly higher mean scores than the group who had in-person classes, both groups assessed with identical exams. Appropriate screening tests suggested that fraud is unlikely to have significantly influenced these results and that the observed differences in performance reflected increased learning by the remote group. These observations suggest that online classes can help to maintain the continuity of physiology and pathophysiology courses during periods of social isolation and may contribute to improving learning under normal conditions.NEW & NOTEWORTHY In this study, we were able to make a rare direct comparison of face-to-face and remote strategies for the teaching of undergraduate medical students in a specific area, namely, renal pathophysiology. Unexpectedly, students who attended the remote course had significantly higher grades than those who had mostly in-person classes.

Association of exposures to serum terpenes with the prevalence of dyslipidemia: a population-based analysis

This study sought to examine hitherto unresearched relationships between serum terpenes and the prevalence of dyslipidemia. Serum terpenes such as limonene, α-pinene, and β-pinene from the 2013-2014 National Health and Nutrition Examination Survey (NHANES) were used as independent variables in this cross-sectional study. Continuous lipid variables included total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), non-HDL-C, triglycerides (TG), low-density lipoprotein cholesterol (LDL-C), residual cholesterol (RC), and apolipoprotein B (Apo B). Binary lipid variables (elevated TC, ≥5.18 mmol/L; lowered HDL-C, <1.04 mmol/L in men, and <1.30 mmol/L in women; elevated non-HDL-C, ≥4.2 mmol/L; elevated TG, ≥1.7 mmol/L; elevated LDL-C, ≥3.37 mmol/L; elevated RC, ≥1.0 mmol/L; and elevated Apo B, ≥1.3 g/L) suggest dyslipidemia. The relationships between the mixture of serum terpenes with lipid variables were investigated using weighted quantile sum (WQS) regression and Bayesian kernel machine regression (BKMR). The study for TC, HDL-C, and non-HDL-C included a total of 1,528 people, whereas the analysis for TG, LDL-C, RC, and Apo B comprised 714 participants. The mean age of the overall participants was 47.69 years, and 48.77% were male. We found that tertiles of serum terpene were positively associated with binary (elevated TC, non-HDL-C, TG, LDL-C, RC, Apo B, and lowered HDL-C) and continuous (TC, non-HDL-C, TG, LDL-C, RC, and Apo B, but not HDL-C) serum lipid variables. WQS regression and BKMR analysis revealed that the mixture of serum terpenes was linked with the prevalence of dyslipidemia. According to our data, the prevalence of dyslipidemia was correlated with serum concentrations of three terpenes both separately and collectively.

A Genome-Wide Association Study and Machine-Learning Algorithm Analysis on the Prediction of Facial Phenotypes by Genotypes in Korean Women

Purpose

Changes in facial appearance are affected by various intrinsic and extrinsic factors, which vary from person to person. Therefore, each person needs to determine their skin condition accurately to care for their skin accordingly. Recently, genetic identification by skin-related phenotypes has become possible using genome-wide association studies (GWAS) and machine-learning algorithms. However, because most GWAS have focused on populations with American or European skin pigmentation, large-scale GWAS are needed for Asian populations. This study aimed to evaluate the correlation of facial phenotypes with candidate single-nucleotide polymorphisms (SNPs) to predict phenotype from genotype using machine learning.

Materials and Methods

A total of 749 Korean women aged 30-50 years were enrolled in this study and evaluated for five facial phenotypes (melanin, gloss, hydration, wrinkle, and elasticity). To find highly related SNPs with each phenotype, GWAS analysis was used. In addition, phenotype prediction was performed using three machine-learning algorithms (linear, ridge, and linear support vector regressions) using five-fold cross-validation.

Results

Using GWAS analysis, we found 46 novel highly associated SNPs (p < 1×10-05): 3, 20, 12, 6, and 5 SNPs for melanin, gloss, hydration, wrinkle, and elasticity, respectively. On comparing the performance of each model based on phenotypes using five-fold cross-validation, the ridge regression model showed the highest accuracy (r2 = 0.6422-0.7266) in all skin traits. Therefore, the optimal solution for personal skin diagnosis using GWAS was with the ridge regression model.

Conclusion

The proposed facial phenotype prediction model in this study provided the optimal solution for accurately predicting the skin condition of an individual by identifying genotype information of target characteristics and machine-learning methods. This model has potential utility for the development of customized cosmetics.

The impact of timetable on student's absences and performance

Student’s academic performance is the point of interest for both the student and the academic institution in higher education. This performance can be affected by several factors and one of them is student absences. This is mainly due to the missed lectures and other class activities. Studies related to university timetabling investigate the different techniques and algorithms to design course timetables without analyzing the relationship between student attendance behavior and timetable design. This article first aimed at demonstrating the impact of absences and timetabling design on student’s academic performance. Secondly, this study showed that the number of absences can be caused by three main timetable design factors: namely, (1) the number of courses per semester, (2) the average number of lectures per day and (3) the average number of free timeslots per day. This was demonstrated using Educational Data Mining on a large dataset collected from Prince Sultan University. The results showed a high prediction performance reaching 92% when predicting student’s GPA based on absences and the factors related to timetabling design. High prediction performance reaching 87% was also obtained when predicting student absences based on the three timetable factors mentioned above. The results demonstrated the importance of designing course timetables in view of student absence behavior. Some suggestions were reported such as limiting the number of enrolled courses based on student’s GPA, avoiding busy and almost free days and using automated timetabling to minimize the number of predicted absences. This in turn will help in generating balanced student timetables, and thus improving student academic performance.

Prediction of venous thromboembolism with machine learning techniques in young-middle-aged inpatients

Accumulating studies appear to suggest that the risk factors for venous thromboembolism (VTE) among young-middle-aged inpatients are different from those among elderly people. Therefore, the current prediction models for VTE are not applicable to young-middle-aged inpatients. The aim of this study was to develop and externally validate a new prediction model for young-middle-aged people using machine learning methods. The clinical data sets linked with 167 inpatients with deep venous thrombosis (DVT) and/or pulmonary embolism (PE) and 406 patients without DVT or PE were compared and analysed with machine learning techniques. Five algorithms, including logistic regression, decision tree, feed-forward neural network, support vector machine, and random forest, were used for training and preparing the models. The support vector machine model had the best performance, with AUC values of 0.806-0.944 for 95% CI, 59% sensitivity and 99% specificity, and an accuracy of 87%. Although different top predictors of adverse outcomes appeared in the different models, life-threatening illness, fibrinogen, RBCs, and PT appeared to be more consistently featured by the different models as top predictors of adverse outcomes. Clinical data sets of young and middle-aged inpatients can be used to accurately predict the risk of VTE with a support vector machine model.

Age-related DNA methylation analysis for forensic age estimation using post-mortem blood samples from Japanese individuals

As one of external visible characteristics (EVCs) in forensic phenotyping, age estimation is essential to providing additional information about a sample donor. With the development of epigenetics, age-related DNA methylation may be used as a reliable predictor of age estimation. With the aim of building a feasible age estimation model for Japanese individuals, 53 CpG sites distributed between 11 candidate genes were selected from previous studies. The DNA methylation level of each target CpG site was identified and measured on a massive parallel platform (synthesis by sequencing, Illumina, California, United States) from 60 forensic blood samples during the initial training phase. Multiple linear regression and quantile regression analyses were later performed to build linear and quantile age estimation models, respectively. Four CpG sites on four genes- ASPA, ELOVL2, ITGA2B, and PDE4C -, were found to be highly correlated with chronological age in DNA samples from Japanese individuals (|R| > 0.75). Subsequently, an independent validation dataset (n = 30) was used to verify and evaluate the performance of the two models. Comparison of mean absolute deviation (MAD) with other indicators showed that both models provide accurate age predictions (MAD: linear = 6.493 years; quantile = 6.243 years). The quantile model, however, can provide the changeable prediction intervals that grow wider with increasing age, and this tendency is consistent with the natural aging process in humans. Hence, the quantile model is recommended in this study.

Machine Learning Improvements to Human Motion Tracking with IMUs

Inertial Measurement Units (IMUs) have become a popular solution for tracking human motion. The main problem of using IMU data for deriving the position of different body segments throughout time is related to the accumulation of the errors in the inertial data. The solution to this problem is necessary to improve the use of IMUs for position tracking. In this work, we present several Machine Learning (ML) methods to improve the position tracking of various body segments when performing different movements. Firstly, classifiers were used to identify the periods in which the IMUs were stopped (zero-velocity detection). The models Random Forest, Support Vector Machine (SVM) and neural networks based on Long-Short-Term Memory (LSTM) layers were capable of identifying those periods independently of the motion and body segment with a substantially higher performance than the traditional fixed-threshold zero-velocity detectors. Afterwards, these techniques were combined with ML regression models based on LSTMs capable of estimating the displacement of the sensors during periods of movement. These models did not show significant improvements when compared with the more straightforward double integration of the linear acceleration data with drift removal for translational motion estimate. Finally, we present a model based on LSTMs that combined simultaneously zero-velocity detection with the translational motion of sensors estimate. This model revealed a lower average error for position tracking than the combination of the previously referred methodologies.