Bayesian Information Criterion Research Articles

Sleep Encephalography‐Based Brain Age Index as a Potential Biomarker for Incident Cognitive Impairment in Community‐Dwelling Older Men

AbstractBackgroundIndividuals with dementia frequently exhibit disturbed sleep electroencephalography (EEG) microstructures, including spindles and slow oscillations. We have previously developed a sleep EEG‐based brain age index (BAI), which integrates multiple sleep EEG microstructures and has been associated with prevalent dementia in clinical populations. Here, we aim to determine if BAI is associated with incident cognitive impairment (CI) in a community population of older men, and its discrimination capability as a digital biomarker.MethodThis longitudinal study included men older than 65 years who completed overnight polysomnography (PSG) between 2003 and 2005 and had incident CI followed until 2016, as part of the Osteoporotic Fractures in Men study (MrOS). BAI was based on a model trained on participants without neurological or psychiatric diseases. Incident CI was defined as self‐reported dementia, dementia medication use, Modified Mini‐Mental State Examination (3MS)<80, or a decline in 3MS from baseline to any following visit ≥1.5 standard deviations below the mean. We fit logistic regression to assess the associations between quintiles of BAI and incident CI.ResultThere were 2,609 men without dementia at baseline (mean age [sd] = 76.0 [5.3] years), of whom 416 developed incident CI during a maximum follow‐up of 11 years. After adjustment for demographics, education, baseline cognition, medication use, comorbidities, and lifestyle, the highest quintile of BAI (BAI>+2.3 years) was associated with incident CI with an odds ratio of 1.66 (95% confidence interval 1.17–2.38, p=0.0051) when compared to the lowest quintile (BAI<‐6.7 years). Other quintiles were not significantly associated with incident CI. BAI showed a cross‐validated accuracy of 60%, sensitivity of 66%, specificity of 55%, and area under the receiver operative curve of 61% for incident CI. When looking at the brain age itself, the brain age performed better than chronological age in predicting incident CI, since the model with brain age achieved lower (i.e., better) Bayesian Information Criterion than another model with age, together with other covariates.ConclusionHigher BAI was associated with incident CI in community‐dwelling older men. These results highlight the potential of BAI as a digital biomarker of future CI in older men.

PsiPartition: Improved Site Partitioning for Genomic Data by Parameterized Sorting Indices and Bayesian Optimization.

Phylogenetics has been widely used in molecular biology to infer the evolutionary relationships among species. With the rapid development of sequencing technology, genomic data with thousands of sites become increasingly common in phylogenetic analysis, while heterogeneity among sites arises as one of the major challenges. A single homogeneous model is not sufficient to describe the evolution of all sites and partitioned models are often employed to model the evolution of heterogeneous sites by partitioning them into distinct groups and utilizing distinct evolutionary models for each group. It is crucial to determine the best partitioning, which greatly affects the reconstruction correctness of phylogeny. However, the best partitioning is usually intractable to obtain in practice. Traditional partitioning methods rely on heuristic algorithms or greedy search to determine the best ones in their solution space, are usually time consuming, and with no guarantee of optimality. In this study, we propose a novel partitioning approach, termed PsiPartition, based on the parameterized sorting indices of sites and Bayesian optimization. We apply our method to empirical datasets, and it performs significantly better compared to existing methods, in terms of Bayesian information criterion (BIC) and the corrected Akaike information criterion (AICc). We test PsiPartition on the simulated datasets with different site heterogeneity, alignment lengths, and number of loci. It is demonstrated that PsiPartition evidently and stably outperforms other methods in terms of the Robinson-Foulds (RF) distance between the true simulated trees and the reconstructed trees, especially on the data with more site heterogeneity. More importantly, our proposed Bayesian optimization-based method, for the first time, provides a new general framework to efficiently determine the optimal number of partitions. The corresponding reproducible source code and data are available at http://github.com/xu-shi-jie/PsiPartition .

Bayesian information criteria for multiple regression models: A study of robustness and comparisons

It is recommended to employ robust versions of selection criteria for variables in linear regression when extreme outliers and high leverage points present. This article investigates the robust version of the Bayesian information criterion based on the MM-estimation scale BICMM . Furthermore, the robustness properties of the proposed BICMM criterion are examined via its breakdown point, influence function, and gross-error sensitivity. Performance of the proposed criterion is compared to existing non-robust and robust Bayesian information criterion, robust Akaike’s information criterion, robust Mallows’ C p , and robust Schwarz information criterion, based on the least-trimmed squares estimator. Simulation studies and applications on real data reveal that BICMM selects more suitable models in situations where there are outliers or leverage points. Consequently, when multicollinearity exists, the proposed robust BICMM can be used to select the tuning parameter in shrinkage regression methods such as ridge and lasso regression.

Forecasting Ambient Air Temperature for Solar Photovoltaic Panel Predictive Maintenance in Kuala Kangsar, Perak, Malaysia

This study presents a forecasting approach to predict ambient air temperatures for the predictive maintenance of solar photovoltaic (PV) panels in Kuala Kangsar, Perak, Malaysia. Accurate temperature forecasting is critical as ambient air temperature significantly influences solar panel efficiency, performance, and maintenance requirements. A comprehensive dataset spanning 19 years (2005–2023) of hourly solar and weather variables, obtained from the Photovoltaic Geographical Information System (PVGIS), was analysed using advanced smoothing techniques, including exponential smoothing models. The study identified the Damped-Trend Linear Exponential Smoothing model as the most effective method based on Akaike and Bayesian Information Criteria. Forecast results for 2024 demonstrated good predictive accuracy, aiding the optimisation of maintenance schedules and the performance of solar PV systems. The findings underscore the importance of integrating advanced predictive techniques to enhance the sustainability and reliability of renewable energy projects.

Clustering of cognitive domains among older persons with HIV.

Age and HIV are synergistic risk factors for conditions such as HIV-associated neurocognitive disorders (HAND). Yet, it is unclear whether older persons with HIV (OPWH) display different cognitive profiles for HAND. To describe the cognitive patterns of OPWH treated with combined antiretroviral therapy (cART). Cross-sectional study that included 330 participants with HIV, aged 50 years or older, cared for at a tertiary care hospital in Mexico City. A short neuropsychological test battery was used to assess a wide spectrum of cognitive functions. The optimal number of cognitive clusters was determined by the silhouette method and a minimization of the Bayesian information criterion. Participants' mean age was 58.8 years (standard deviation = 6.6), and 12.1% were women. A 3-cluster solution yielded stable Jaccard coefficients (p > 0.70). Cluster 1 showed more significant impairment in visual and verbal memory domains, whereas participants in cluster 3 showed significant impairment in language, and abstraction. Cluster 2 showed no predominance of any domain for alterations. There are different cognitive profiles among OAWH with HAND. These differences may be due to individual patterns of HIV-related and non-HIV-related factors.

Clustering of cognitive domains among older adults with HIV

Age and HIV are synergistic risk factors for conditions such as HIV-associated neurocognitive disorders (HAND). Yet, it is unclear whether OAWH display different cognitive profiles for HAND. To describe the cognitive patterns of OAWH treated with combined antiretroviral therapy (cART). Cross-sectional study that included 330 participants with HIV, aged 50 years or older, cared for at a tertiary care hospital in Mexico City. A short neuropsychological test battery was used to assess a wide spectrum of cognitive functions. The optimal number of cognitive clusters was determined by the silhouette method and a minimization of the Bayesian information criterion. Participants' mean age was 58.8 years (SD = 6.6), and 12.1% were women. A 3-cluster solution yielded stable Jaccard coefficients (p > 0.70). Cluster 1 showed more significant impairment in visual and verbal memory domains, whereas participants in cluster 3 showed significant impairment in language, and abstraction. Cluster 2 showed no predominance of any domain for alterations. There are different cognitive profiles among OAWH with HAND. These differences may be due to individual patterns of HIV-related and non-HIV-related factors.

Quantile Regression and Homogeneity Identification of a Semiparametric Panel Data Model

In this article, we delve into the quantile regression and homogeneity detection of a varying index coefficient panel data model, which incorporates fixed individual effects and exhibits nonlinear time trends. Utilizing spline approximation, we obtain estimators for the trend functions, link functions, and index parameters, and subsequently establish the corresponding convergence rates and asymptotic normality. Observing that subjects within a group may share identical trend functions, we are motivated to further explore potential homogeneity in these trends. To this end, we propose a homogeneity identification algorithm based on binary segmentation. For the determination of the thresholding parameter in homogeneity identification, we propose a generalized Bayesian information criterion, following the approach outlined in Chen (2019). Furthermore, we introduce a penalized method to discern the constant and linear structures within the nonparametric functions of our model. By leveraging grouped observations, we achieve more efficient estimation and improve the asymptotic properties of the estimators. To demonstrate the finite sample performance of our proposed approach, we conduct simulation studies and apply our methodology to a real-world dataset comprising Air Pollution Data and Integrated Surface Data (APD&ISD).

Blockchain‐based smart and secure manufacturing systems

AbstractIn the current data sharing environment of smart manufacturing systems, there is a lack of a transparent, open, and fair interaction environment, and there exists a serious imbalance in data sharing and transmission, which leads to data trust issues. To address this challenge, we propose a fine‐grained access control method for IoT data sharing based on blockchain technology (the Bayesian information criterion [BIC] scheme for IoT data sharing). This approach includes ex‐ante supervision and authentication of data sources, fine‐grained access control during data usage, and ex‐post tracking of malicious users. Firstly, a system model is built, and symmetric encryption is used to encrypt the data in order to improve computational efficiency. Subsequently, scheme design and algorithm design are carried out to encrypt the symmetric key using ciphertext‐policy attribute‐based encryption to achieve one‐to‐many data sharing, which greatly reduces the number of repeated encryptions and computational overheads. The data owner uploads the encrypted data to the cloud server, and the data user must hold the token generated by the blockchain to interact with the cloud and obtain the corresponding information. The results show that the BIC scheme not only ensures the legitimacy and security of data access but also locks up and traces malicious users who may have compromised their private keys, preventing illegal data sharing and key misuse; it offers significant advantages in terms of storage efficiency and time consumption, and also facilitates a flexible and secure data sharing process.

A Multi-Model Survival Analysis of Lung Cancer Using Parametric Techniques

Lung cancer remains one of the leading causes of cancer-related mortality worldwide, underscoring the critical need for effective prognostic tools. This study utilizes survival analysis to explore the factors that influence the survival outcomes of North Central Cancer Treatment data related to lung cancer. The main goal of this study is to compare and contrast various statistical models, including the Weibull, Exponential, Log-gaussian, Gumbel, and Rayleigh models. We have computed important functions, such as the survival function, the hazard function, and the cumulative hazard function, for all the considered distributions. The Anderson-Darling and Cramer Von-Mises tests, which are Goodness of fit tests, effectively compare and assess various parametric regression survival models. The Weibull survival model is interpreted to be the most effective and efficient way to study the lung cancer dataset, which is concluded upon evaluating the results of Anderson-Darling statistic 0.28745, Cramer Von-Mises statistic 0.0450, Mean Survival Probability 0.9697, Mean Cumulative Survival Probability 0.0303, Akaike Information Criterion 1,650.753 and Bayesian Information Criterion 1623.329 of the Weibull, Exponential, Log-gaussian, Gumbel, and Rayleigh parametric regression survival models.

Historical memory in remotely sensed soil moisture can enhance flash flood modeling for headwater catchments in Germany

The wetness precondition of a catchment affects available soil water storage capacity and infiltration rate, thus influences flash flood generation. Remotely sensed (RS) soil moisture (SM) can provide valuable information on catchment wetness, but typically only represents the top 5 cm of the land surface. However, hydrological models for flash flood simulation need to consider deeper layers to calculate the total soil water storage. Therefore, a key challenge is to link RS SM to total soil water storage and assimilate RS SM into flash flood models to correctly describe initial catchment wetness. In this study, we developed an approach to combine present and antecedent RS SM to infer present soil water storage based on four regression models. The inferred soil water storage from SMAP (soil moisture active passive) SM was assimilated into the operational LARSIM (Large Area Runoff Simulation Model) hydrological model. We tested this new approach with 12 events in the headwater catchments Körsch, Adenauer Bach and Fischbach in Germany. Results show that random forest regression performs the best among the four regression models. The BIC (Bayesian Information Criterion) score suggests that regressions considering antecedent RS SM can well infer soil water storage, resulting in R2 of 0.85, 0.94 and 0.93 for the Körsch, Adenauer Bach and Fischbach catchments, respectively. Compared to the open loop (without data assimilation) simulations, our approach enhanced the general performance of event simulations with average KGE increases of 0.09, 0.24 and 0.33 for the Körsch, Adenauer Bach and Fischbach, respectively; and the mean error in the 12 simulated event peaks is reduced 15 %. Moreover, the simulation uncertainty is reduced, too. The transferability of the proposed approach to other RS products is also discussed. Although assimilating RS SM can enhance flash flood modeling, it is primarily affected by the uncertainty in precipitation. In the future, the proposed approach should be tested with more catchments and events to verify its general validity.

Linking Meteorological Variables and Particulate Matter PM2.5 in the Aburrá Valley, Colombia

Environmental pollution indicated by the presence of PM2.5 particulate matter varies based on prevailing atmospheric conditions described by certain meteorological variables. Consequently, it is important to understand atmospheric behavior in areas such as the Aburrá Valley, which experiences recurrent pollution events twice a year. This study examines the behavior of specific meteorological variables and PM2.5 particulate matter in the Aburrá Valley. By using statistical analysis tools such as correlation coefficients, principal component analysis (PCA), and multiple linear regression models, the research identifies relationships between PM2.5 and daily cycles of temperature, rainfall, radiation, and wind speed and direction. Datasets were analyzed considering periods before and after the COVID-19 lockdown (pre-pandemic and pandemic, respectively), and specific pollution events were also analyzed. Furthermore, this work considers the relationships between PM2.5 and meteorological variables, contrasting the pre-pandemic and pandemic periods. This study characterizes diurnal cycles of meteorological variables and their relationship with PM2.5. There are consistent patterns among temperature, atmospheric boundary layer (ABL) height, and solar radiation, whereas precipitation and relative humidity show the opposite behavior. PM2.5 exhibits similar relative frequency functions during both daytime and nighttime, regardless of rainfall. An inverse relationship is noted between PM2.5 levels and ABL height at different times of the day. Moreover, the PCA results show that the first principal component explains around 60% of the total variance in the hydrometeorological data. The second PC explains 10%, and the rest of the variance is distributed among the other three to eight PCs. In this sense, there is no significant difference between the two PCAs with hydrometeorological data from a pre-pandemic period and a COVID-19 pandemic period. Multiple regression analysis indicates a significant and consistent dependence of PM2.5 on temperature and solar radiation across both analyzed periods. The application of Generalized Additive Models (GAMs) to our dataset yielded promising results, reflecting the complex relationship between meteorological variables and PM2.5 concentrations. The metrics obtained from the GAM were as follows: Mean Squared Error (MSE) of 98.04, Root Mean Squared Error (RMSE) of 9.90, R-squared (R2) of 0.24, Akaike Information Criterion (AIC) of 110,051.34, and Bayesian Information Criterion (BIC) of 110,140.63. In comparison, the linear regression model exhibited slightly higher MSE (100.49), RMSE (10.02), and lower R-squared (0.22), with AIC and BIC values of 110,407.45 and 110,460.67, respectively. Although the improvement in performance metrics from GAM over the linear model is not conclusive, they indicate a better fit for the complexity of atmospheric dynamics influencing PM2.5 levels. These findings underscore the intricate interplay of meteorological factors and particulate matter concentration, reinforcing the necessity for advanced modeling techniques in environmental studies. This work presents new insights that enhance the diagnosis, understanding, and modeling of environmental pollution, thereby supporting informed decision-making and strengthening management efforts.

Development of growth curve and estimation of genetic parameters of Nili-Ravi buffalo calves from birth to one year of age through non-linear models.

The current study aimed to construct the growth curve and find the best-fitting non-linear model for the Nili-Ravi buffalo calves. The pedigree and monthly body weight data for 6644 calves born between 2010 and 2020 (inclusive) in six herds, maintained in different geographical regions of Punjab were collected. The study was performed under a longitudinal design and various non-linear models were used to associate the age with weight by using the easyreg package in R studio. Finally, the genetic parameters of growth curve were estimated through the bi-variate sire model in ASReml (v4.2). The model with the highest value of coefficient of determination and the lowest values of Akaike's information criterion, Bayesian information criterion, and root-mean-square error was considered as the best fit for defining the growth curve. The Brody model was found as the best fitted model with the values of 0.6648, 627871.80, 627908.10, and 30.793 for the R2adj, AIC, BIC, & RMSE respectively, for the combined dataset. The values of growth curve parameters for the Brody model were 943.99 ± 101.38Kg (A), 0.96 ± 0.004Kg (B), and 0.0005 ± 0.00Kg (K) for all animals. A higher K-value of females indicates their early maturity compared with male animals in this breed. The heritability estimates for the growth curve traits were low, while the values of genetic correlations were higher than those of phenotypic correlations. The data revealed that Asymptotic weight (A) and birth weight (B) were positively correlated with each other, while the rate of maturity (K) was negatively correlated with initial and final body weights.

Performance Assessment of Some Count Data Models to Immunization Coverage Data

This research evaluates the performance of various count data models, including Poisson Regression (PR), Zero-Inflated Poisson Regression (ZIP), Zero-Truncated Poisson Regression (ZTP), Truncated Negative Binomial Poisson Regression (TNBP), and Negative Binomial Poisson Regression (NBP), using immunization coverage data from the National Primary Health Care Development Agency (NPHCDA). The study focuses on children under 12 months, assessing model fit using Likelihood Ratio (LR), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) criteria. Analysis conducted with STATA indicates that the Truncated Negative Binomial Poisson Regression (TNBP) outperformed other models in fit and efficiency. Both the ZTeeP and TNBP models demonstrated the best fit, with lower AIC (1959.107) and BIC (2037.649) values and higher Pseudo R-squared values (0.0677 for ZTP and 0.0590 for TNBP), compared to standard models. Age was identified as a significant predictor, negatively associated with immunization status, implying that older infants in the under-12-month category are less likely to receive all vaccinations. The ZTP model showed significant positive effects for antigens such as HepB0, OPV0, BCG, and Measles, with age having a significant negative association. The findings highlight the importance of selecting appropriate statistical models for accurate public health data analysis, enhancing decision-making in immunization programs.

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.

Optimized technique for speaker changes detection in multispeaker audio recording using pyknogram and efficient distance metric.

Segmentation process is very popular in Speech recognition, word count, speaker indexing and speaker diarization process. This paper describes the speaker segmentation system which detects the speaker change point in an audio recording of multi speakers with the help of feature extraction and proposed distance metric algorithms. In this new approach, pre-processing of audio stream includes noise reduction, speech compression by using discrete wavelet transform (Daubechies wavelet 'db40' at level 2) and framing. It is followed by two feature extraction algorithms pyknogram and nonlinear energy operator (NEO). Finally, the extracted features of each frame are used to detect speaker change point which is accomplished by applying dissimilarity measures to find the distance between two frames. To realize it, a sliding window is moved across the whole data stream to find the highest peak which corresponds to the speaker change point. The distance metrics incorporated are standard "Bayesian Information Criteria (BIC)", "Kullback Leibler Divergence (KLD)", "T-test" and proposed algorithm to detect the speaker boundaries. At the end, threshold value is applied and their results are evaluated with Recall, Precision and F-measure. Best result of 99.34% is shown by proposed distance metric with pyknogram as compare to BIC, KLD and T-test algorithms.

An Arima Model-Based Approach to Improve Electricity Reliability

Objective: The objective of this study was to forecast the number of critical days, for the period from July 2020 to December 2022, based on time series analysis, to model changes in the number of critical power days in a prospective way. Related Studies: This section provides an overview of the use of forecasting models and methods for measuring electricity demand, applied as decision-making support and as a tool for identifying anomalous phenomena. Method: The methodology for this research involves analysis and forecasting using the Autoregressive Integrated Moving Average (ARIMA) model of order (p,d,q)(p,d,q), applied to the time series of critical days. Model development and performance evaluation followed these steps and strategies: Specification (Pre-processing), Identification and Estimation, Verification, and Forecasting. Data collection was performed by analyzing time series observations from a Brazilian electric utility company responsible for energy supply in Rio Grande do Sul. Results and Discussion: The results indicated that the seasonal ARIMA (2,1,1)(2,1,1) model performed best, with the lowest Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) values, as well as the lowest mean square error among the tested models. With the selected ARIMA model, forecasts were made for the following 30 months, estimating the number of monthly critical days. Research Implications: The paper discusses the practical and theoretical implications of forecasting critical days for power quality. Practically, it presents an ARIMA model to estimate these days, aiding power companies in projecting interruptions and improving preventive management and resource allocation. Theoretically, it contributes to the time series forecasting literature in the power sector, confirming the effectiveness of ARIMA models with seasonal components for grid management and adverse event forecasting in complex systems. Originality/Value: The study contributes to the literature by applying the ARIMA model with seasonal components to predict critical days of interruption in the distribution of electricity, an approach that has not yet been explored for specific adverse events. The originality lies in the application of the model in a context of quality and reliability in the supply and detailed seasonal analysis. The methodology proposes a solution for forecasting and managing these events, bringing new perspectives for proactive management in the sector. The study stands out for its potential to improve professional practice, allowing informed decisions for resource allocation and maintenance, increasing reliability and reducing costs and financial compensation.

Fitting COVID-19 Incidences in Kenya Using Fractional Polynomials and Linear Splines

The study provides an in-depth analysis of COVID-19 infections in Kenya, aiming to model the non-linear trajectory of daily cases. The research explores two statistical techniques: fractional polynomials and linear splines, to fit the growth of infection rates over time. COVID-19, which first appeared in Kenya in March 2020, exhibited fluctuating trends in daily infections. The study utilizes infection data collected from March 13, 2020, to June 6, 2021. Descriptive statistics and exploratory data analysis revealed significant variability in daily cases, with the infection trajectory characterized by multiple waves. Fractional polynomial models, known for their flexibility in fitting non-linear relationships, were evaluated at varying degrees to identify the best model for COVID-19 incidence trends. The analysis showed that a second-degree fractional polynomial with powers (1, 2) provided the most accurate fit for the data. The closed test algorithm was applied to confirm the model's suitability. Additionally, linear spline models were employed, partitioning the data into segments and fitting linear splines at each knot point. The model with 19 knots demonstrated superior performance based on Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC), outperforming the fractional polynomial model. The comparison of the two methods concluded that linear splines provided a more precise fit for the infection data, capturing the complex nature of COVID-19's spread in Kenya. The study's findings offer critical insights into the infection dynamics and can aid policymakers in resource allocation and mitigation planning during pandemics. The study recommends further analysis by incorporating more covariates and extending the models to other countries for a comparative understanding of pandemic management strategies.

Prediction of Ki-67 expression and malignant potential in gastrointestinal stromal tumors: novel models based on CE-CT and serological indicators

PurposeTo identify more reliable imaging and serological indicators for predicting Ki-67 expression and malignant potential in gastrointestinal stromal tumors, as well as to develop a preoperative prediction model with clinical utility.Patients and methodsThis study retrospectively analyzed patients with gastrointestinal stromal tumors (GIST) diagnosed at the First Affiliated Hospital of Jinzhou Medical University between May 2018 and May 2024. Univariate logistic analyses, two-way stepwise regression, P-value stepwise regression, and LASSO regression were employed to screen for Ki-67 high expression and high malignant potential risk factors associated with GIST. Models were established using various regression methods; Nomograms, calibration curves, and clinical decision curves were generated for the two best prediction models.ResultsTwo-way stepwise regression analysis revealed that diameter (P=0.037; OR=1.22; 95% CI: 1.01 - 1.46), growth pattern (extraluminal type: P=0.028; OR=3.54; 95% CI: 1.14 - 10.94), enhancement model (P=0.099; OR=0.39; 95% CI: 0.12 - 1.20), EVFDM (P=0.069; OR=0.43; 95% CI: 0.17 - 1.07), PLR (P=0.099; OR=3.06; 95% CI: 0.81 - 11.59), and OPNI (P=0.058; OR=2.38; 95% CI: 0.97 - 5.84) are identified as independent risk factors for Ki-67 expression. Utilizing the two-way stepwise regression model to predict Ki-67 expression, the area under the curve (AUC) for the training group was 0.865 (95% CI: 0.807-0.922), while for the validation group it was 0.784 (95% CI: 0.631-0.937). The Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) for the training group were 153.360 and 174.619, respectively. Two-way stepwise regression analysis revealed that volume (P < .001, OR = 1.06; 95% CI: 1.03 - 1.09), contour (P = 0.066; OR = 0.17; 95% CI: 0.05 - 0.62), ulcer (P = 0.094; OR = 0.16; 95% CI: 0.03 - 0.98), IBSC (P = 0.008; OR = 5.27; 95% CI: 1.57 - 17.69), and OPNI (P = 0.045; OR = 0.22; 95% CI: 0.05 - 0.96) are independent risk factors for malignant potential. Utilizing the two-way stepwise regression model to predict malignant potential, the AUC for the training group was 0.950 (95% CI: 0.920 - 0.980), while for the validation group it was 0.936 (95% CI: 0.867 - 1.000). The AIC and BIC values for the training group were 96.330 and 114.552, respectively.ConclusionDiameter, growth pattern, enhancement pattern, EVFDM, PLR, and OPNI are independent risk factors for GIST with high Ki-67 expression. Additionally, volume, contour, ulceration, IBSC, and OPNI serve as independent risk factors for GIST with high malignant potential. The preoperative models developed using CT images can predict the malignant potential and Ki-67 expression status of GIST to a certain extent. When combined with serological indicators, these models' predictive performance can be further enhanced.

Time to Cure and Predictive Factors Among Patients with Multi-Drug Resistance Tuberculosis in Puntland, Somalia: A Retrospective Cohort Study

Background: Multi-drug-resistant tuberculosis, also known as MDR-TB, is caused by bacteria that are resistant to the most effective first-line anti-tuberculosis medications, which are rifampicin and isoniazid. MDR-TB is an increasing global concern, and its spread has varying cure times for patients affected. Thus, this study was intended to investigate the median time to cure and identify predictive factors for patients with MDR-TB in Puntland, Somalia. Methods: A retrospective cohort study was carried out at the MDR-TB center in Galkayo, Puntland, Somalia between the years of July 2017 and August 2023. The study utilized data that were collected from the inquiry form and monitoring cards of a randomly selected sample of 130 patients over a 6-year period. The data were entered into Epi-Data version 4.3 and analyzed using R Programming version 4.4.0. Non-parametric methods were used to estimate the median time to cure for patients with MDR-TB. Semi-parametric and parametric models were employed to determine predictive factors that influence the time to cure for MDR-TB patients. The Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) of different survival models were compared to select the model that offers the best fit. Results: Out of the total of MDR-TB patients, 109(83.85%) were cured and 21(16.15%) were censored during the follow-up period. The median time to cure from MDR-TB was found to be 619 days which is approximately 20 months. In accordance with the AIC and BIC, the Log-logistic accelerated failure time model was the best fit for the data as compared to the other AFT models. The result of the Log-logistic AFT model revealed that age, sex, disease site, and comorbidities significantly affect the time to cure for MDR-TB. According to this study, older age with MDR-TB patients, male MDR-TB patients, MDR-TB patients with comorbid conditions, and extrapulmonary MDR-TB patients had a longer time to cure than their reference categories. Conclusion: Based on the findings of the study, the median time to cure for patients with MDR-TB in Puntland was found to be 20 months. Sex, disease site, comorbidities and age were predictive factors of time to cure from MDR-TB.

Statistical Models and Evaluation Criteria Used in Poultry Farming

This thesis study examined the statistical models and evaluation criteria used in poultry breeding, account considering live weight gain, egg weight and egg numbers. For this purpose, as an example of live weight increase; 9-week live weight gains of Ross-308 broiler line, as an example of egg weight; the average of 20-week egg weights of brown Lohmann laying hens and 9-week egg yields of Japanese quail were used as an example of the number of eggs. In modeling, for live weight gain; 10 different models for egg production; 11 different models and 8 different models for egg weight were considered. In evaluating the models; error mean squares, coefficient of determination, corrected coefficient of determination, Akaike Information Criterion, Corrected Akaike Information Criterion, Bayesian Information Criterion, Accuracy Factor, Bias Factor and Durbin-Watson autocorrelation values were taken into consideration. As a result of the study, in terms of live weight gain; The Cubic Piecewise regression model is the best in terms of egg yields; It was determined that the Modified Compartmental model and the Logistic model gave better results than the others in terms of egg weight. The worst models are in live weight gains; Brody and egg yields; It was concluded that there was a Quadratic Linear and Von Bertalanffy model for egg weight.