Prediction of sepsis mortality in ICU patients using machine learning methods.

PROBLEM: Sepsis, a life-threatening condition, accounts for the deaths of millions of people worldwide. Accurate prediction of sepsis outcomes is crucial for effective treatment and management. Previous studies have utilized machine learning for prognosis, but have limitations in feature sets and model interpretability. AIM: This study aims to develop a machine learning model that enhances prediction accuracy for sepsis outcomes using a reduced set of features, thereby addressing the limitations of previous studies and enhancing model interpretability. METHODS: This study analyzes intensive care patient outcomes using the MIMIC-IV database, focusing on adult sepsis cases. Employing the latest data extraction tools, such as Google BigQuery, and following stringent selection criteria, we selected 38 features in this study. This selection is also informed by a comprehensive literature review and clinical expertise. Data preprocessing included handling missing values, regrouping categorical variables, and using the Synthetic Minority Over-sampling Technique (SMOTE) to balance the data. We evaluated several machine learning models: Decision Trees, Gradient Boosting, XGBoost, LightGBM, Multilayer Perceptrons (MLP), Support Vector Machines (SVM), and Random Forest. The Sequential Halving and Classification (SHAC) algorithm was used for hyperparameter tuning, and both train-test split and cross-validation methodologies were employed for performance and computational efficiency. RESULTS: The Random Forest model was the most effective, achieving an area under the receiver operating characteristic curve (AUROC) of 0.94 with a confidence interval of ±0.01. This significantly outperformed other models and set a new benchmark in the literature. The model also provided detailed insights into the importance of various clinical features, with the Sequential Organ Failure Assessment (SOFA) score and average urine output being highly predictive. SHAP (Shapley Additive Explanations) analysis further enhanced the model's interpretability, offering a clearer understanding of feature impacts. CONCLUSION: This study demonstrates significant improvements in predicting sepsis outcomes using a Random Forest model, supported by advanced machine learning techniques and thorough data preprocessing. Our approach provided detailed insights into the key clinical features impacting sepsis mortality, making the model both highly accurate and interpretable. By enhancing the model's practical utility in clinical settings, we offer a valuable tool for healthcare professionals to make data-driven decisions, ultimately aiming to minimize sepsis-induced fatalities.

Prediction of 30-day mortality for ICU patients with Sepsis-3.

BACKGROUND: There is a growing demand for advanced methods to improve the understanding and prediction of illnesses. This study focuses on Sepsis, a critical response to infection, aiming to enhance early detection and mortality prediction for Sepsis-3 patients to improve hospital resource allocation. METHODS: In this study, we developed a Machine Learning (ML) framework to predict the 30-day mortality rate of ICU patients with Sepsis-3 using the MIMIC-III database. Advanced big data extraction tools like Snowflake were used to identify eligible patients. Decision tree models and Entropy Analyses helped refine feature selection, resulting in 30 relevant features curated with clinical experts. We employed the Light Gradient Boosting Machine (LightGBM) model for its efficiency and predictive power. RESULTS: The study comprised a cohort of 9118 Sepsis-3 patients. Our preprocessing techniques significantly improved both the AUC and accuracy metrics. The LightGBM model achieved an impressive AUC of 0.983 (95% CI: [0.980-0.990]), an accuracy of 0.966, and an F1-score of 0.910. Notably, LightGBM showed a substantial 6% improvement over our best baseline model and a 14% enhancement over the best existing literature. These advancements are attributed to (I) the inclusion of the novel and pivotal feature Hospital Length of Stay (HOSP_LOS), absent in previous studies, and (II) LightGBM's gradient boosting architecture, enabling robust predictions with high-dimensional data while maintaining computational efficiency, as demonstrated by its learning curve. CONCLUSIONS: Our preprocessing methodology reduced the number of relevant features and identified a crucial feature overlooked in previous studies. The proposed model demonstrated high predictive power and generalization capability, highlighting the potential of ML in ICU settings. This model can streamline ICU resource allocation and provide tailored interventions for Sepsis-3 patients.

"Who are You and What are You Doing Here?": Social Capital and Barriers to Movement along the HIV Care Cascade among Tajikistani Migrants with HIV to Russia.

Tajikistani migrants who work in Russia and acquire HIV seldom receive HIV treatment while in Russia. Barriers to engagement in the HIV care cascade were identified from in-depth, semi-structured interviews with purposefully sampled Tajikistani migrants (n = 34) with HIV who had returned from Russia. Data were analyzed using thematic analysis, drawing from Putnam's theory of social capital, showing how bridging and bonding social capital relate to poor engagement in HIV care. We identified three barriers to Tajikistani migrants' movement through the HIV care cascade: (1) Russia's migration ban on people with HIV interrupts social capital accumulation and prevents access to HIV treatment within Russia; (2) mistrust of authority figures, including healthcare providers, leads to avoiding treatment and harm-reduction services upon their return to Tajikistan; and (3) because of pervasive discrimination, Tajikistani migrants form weak social ties while in Russia, which exacerbates risk, including with Russian citizens, and deters engagement with HIV care. Deploying a treatment as prevention strategy and abolishing Russia's ban on people with HIV would improve both individual and public health.

Association between time spent in the Russian Federation and late presentation for HIV among Tajikistani migrants.

BACKGROUND: Between 700 thousand and 1.2 million citizens of Tajikistan currently live in the Russian Federation, one of the only countries where the HIV epidemic continues to worsen. Given the previously reported barriers to healthcare access for migrants to the Russian Federation, and the rapidly expanding HIV epidemic in Eastern Europe and Central Asia, this present study set out to determine whether these barriers impact late presentation with HIV among Tajikistani migrants upon their return to Tajikistan. METHOD: This study uses data from the Tajikistan Ministry of Health surveillance system (2006 - 2019). At time of diagnosis, patients are interviewed by staff of AIDS centers, and doctors complete routine intake forms and complete medical exams. Descriptive characteristics of migrants with HIV who had lived in the Russian Federation (n=503) were calculated and compared with those of non-migrants with HIV (n=9519). Missing data were imputed using multiple imputation (predictive means matching, logistic regression imputation, and polytomous regression imputation). Two logistic models were created to model the probability of late presentation for HIV. The first model shows unadjusted associations between predictor variables and late presentation for HIV. The second model shows multivariable associations between significant study variables identified in the univariate model, and late presentation. RESULTS: Compared to non-migrants, migrants with HIV are more likely to be from Gorno-Badakhshan region, are less likely to use illicit drugs, and are more likely to have purchased the services of sex workers. The unadjusted logistic model found that for every year spent in the Russian Federation, the risk of late presentation for a Tajikistani migrant with HIV increases by 4.0% (95% CI: 0.3-7.7). The multivariate model showed that when age, sex, and region of origin are held constant, the risk of late presentation for a Tajikistani migrant with HIV increases by 4.0% (95% CI: 0.1-7.8) for each year spent in the Russian Federation. CONCLUSION: The results of this paper suggest that if the Russian Federation were to loosen its restrictions on HIV care for foreign nationals, it might improve the treatment outcomes of migrant laborers. As this analysis is only correlational in nature, further research is needed to explicate the causal pathways of the associations found in the present analysis.

Progress in Early Detection of HIV in Tajikistan.

HIV early detection (CD4 counts ≥350 cells/µL) is correlated with higher life expectancy among people living with HIV (PLHIV). Several factors, including physical, cultural, structural, and financial barriers, may limit early detection of HIV. This is a first-of-its-kind study on population-level differences in early detection of HIV across time within Tajikistan and any country in the Central Asia region. Utilizing the Tajikistan Ministry of Health's national HIV data (N = 10,700) spanning 2010 to 2023, we developed median regression models with the median CD4 cell count as the outcome and with the following predictors: time (years), region, age, gender, and area (urban/rural status). Individuals younger than 19 years old were detected early for HIV, whereas those older than 39 years were detected late. Females were detected earlier compared to their male counterparts regardless of region of residence. Rural populations were detected earlier in most years compared to their urban counterparts. The COVID-19 pandemic accelerated HIV early detection in 2021 but most regions have returned to near pre-pandemic levels of detection in 2022 and 2023. There were differences identified among different demographic and geographic groups which warrant further attention.

Predicting Violent Reoffending in Individuals Released From Prison in a Lower-Middle-Income Country: A Validation of OxRec in Tajikistan.

Background: Although around 70% of the world's prison population live in low- and middle-income countries (LMICs), risk assessment tools for criminal recidivism have been developed and validated in high-income countries (HICs). Validating such tools in LMIC settings is important for the risk management of people released from prison, development of evidence-based intervention programmes, and effective allocation of limited resources. Methods: We aimed to externally validate a scalable risk assessment tool, the Oxford Risk of Recidivism (OxRec) tool, which was developed in Sweden, using data from a cohort of people released from prisons in Tajikistan. Data were collected from interviews (for predictors) and criminal records (for some predictors and main outcomes). Individuals were first interviewed in prison and then followed up over a 1-year period for post-release violent reoffending outcomes. We assessed the predictive performance of OxRec by testing discrimination (area under the receiver operating characteristic curve; AUC) and calibration (calibration statistics and plots). In addition, we calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for different predetermined risk thresholds. Results: The cohort included 970 individuals released from prison. During the 12-month follow-up, 144 (15%) were reincarcerated for violent crimes. The original model performed well. The discriminative ability of OxRec Tajikistan was good (AUC = 0.70; 95% CI 0.66-0.75). The calibration plot suggested an underestimation of observed risk probabilities. However, after recalibration, model performance was improved (Brier score = 0.12; calibration in the large was 1.09). At a selected risk threshold of 15%, the tool had a sensitivity of 60%, specificity of 65%, PPV 23% and NPV 90%. In addition, OxRec was feasible to use, despite challenges to risk prediction in LMICs. Conclusion: In an external validation in a LMIC, the OxRec tool demonstrated good performance in multiple measures. OxRec could be used in Tajikistan to help prioritize interventions for people who are at high-risk of violent reoffending after incarceration and screen out others who are at lower risk of violent reoffending. The use of validated risk assessment tools in LMICs could improve risk stratification and inform the development of future interventions tailored at modifiable risk factors for recidivism, such as substance use and mental health problems.

HIV and associated factors among female sex workers in Tajikistan: Analysis from a national bio-behavioral survey.

Tajikistan, a country of approximately nine million people, has a relatively small but quickly growing HIV epidemic. No peer-reviewed study has assessed factors associated with HIV, or associated risk factors, among female sex workers (FSWs) in Tajikistan. The purpose of the current study is to elucidate the factors associated with HIV status and risk factors in the Tajikistani context and add to the scant literature on risk factors among FSWs in Tajikistan and Central Asia. We used cross-sectional data from an HIV bio-behavioral survey (BBS) conducted among FSWs in the Republic of Tajikistan (n = 2174) in 2017. Using Respondent Driven Sampling Analysis Tool software, we calculated the prevalence of HIV, diagnosed cases, linkage to antiretroviral therapy (ART), and the prevalence of syphilis for FSWs in Tajikistan. Prevalence data were adjusted for network size and any clustering effects in the network. Further, using univariate and multivariable logistic regression, we determined correlates of HIV-positive status. Results were as follows: Of all FSWs in Tajikistan, 2.6% (95% CI: 1.7-3.8%) are HIV positive, 2.3% (95% CI: 1.4-3.5%) are diagnosed and aware of their status, and 2.0% (95% CI: 1.2-3.1%) are on ART. About 5.7% (95% CI: 4.5-7.4%) of FSWs in Tajikistan have ever had syphilis, and 0.8% (95% CI: 0.4-1.3%) have active syphilis infections. The epidemic of injection drug use was found to be strongly synergistic with HIV infection as having had sex with a person who injects drugs was shown to be strongly associated with HIV-positive status (OR: 5.2; 95% CI: 2.6-10.2) in the multivariable model. While this study estimates that HIV prevalence among Tajikistani FSWs is relatively low, it is likely an underestimated due to selection and social desirability biases. To curb the small, but potentially volatile, HIV epidemic among FSWs, the government should consider targeted testing and linkage-to-care efforts for FSWs who inject drugs or who have people who inject drugs partners. Services should also be prioritized in Gorno-Badakhshan, which has a higher number of FSWs per capita relative to other regions. Additionally, the link between HIV and experiences of stigma, violence, and discrimination against FSWs should motivate advocates to protect Tajikistani FSWs from these experiences.

Cross-country analysis of correlation between protection of women's economic and social rights, health improvement and sustainabledevelopment.

OBJECTIVE: The goal of this study is to assess the correlation between protection of women's economic and social rights (WESR), health improvement and sustainable development. METHODS: A cross-country analysis of 162 countries was employed to assess development, health and human rights of the countries by measuring associated variables. Data sets for the health, human rights and economic and social rights of these countries were from 2004 to 2010. The dependent variables are health and human development and the independent variables are the human rights variables. Regression analysis and principle axis factoring were used for extraction and varimax method for rotation. Country grouping was made using cluster analysis. Potential biases, resulting from measurement differences in human rights values, were eliminated by using z-transformation to standardise variables. RESULTS: Regression results reveal that WESR variable is correlated with the health outcomes. Cluster analysis separated the countries into three clusters, based on the WESR variable. Countries where WESR were 'highly respected' (44 countries) are categorised into cluster 1; countries where WESR were 'moderately respected' (51 countries) are categorised into cluster 2 and countries where WESR were 'poorly respected' (63 countries) are categorised into cluster 3. Countries were then compared in their respective clusters based on health and human development variables. It was found that the countries which 'highly respected' WESR had better average health values compared with the second and third clusters. Our findings demonstrate that countries with a strong women's rights status ultimately had better health outcomes. CONCLUSION: WESR status has correlation with the health and human development. When women's rights are highly respected, the nation is more likely to have higher health averages and accelerated development.

The Urgency to Mitigate the Spread of Hepatitis C in Pakistan Through Blood Transfusion Reform.

Blood transfusions are contributing to a higher rate of hepatitis C virus (HCV) in Pakistan. Half of all blood transfusions in Pakistan are not screened for hepatitis C, hepatitis B or HIV. Family members donate blood that is likely not tested due to social stigma attached to HCV. Paid donations are also quite common in the country, especially by people who inject drugs (PWID), which increases the population's exposure to HCV. Most of the population utilizes the private sector for their health needs; this sector has lax regulation due to the lack of oversight by the government or any other regulatory body. In addition, groups who are at most need for blood transfusions, such as hemophiliacs and those with thalassemia, have a higher rate of hepatitis C. This fact reinforces the need for blood transfusion reform in Pakistan, which includes improving oversight, upgrading infrastructure and promoting health literacy through cultural norms, according to the World Health Organization (WHO) recommendations. The lessons learned in Pakistan can be adapted to countries facing similar issues.

Association between antiretroviral therapy and antitubercular drug resistance in TB treatment outcome among Kazakh TB/HIV co-infected patients.

OBJECTIVES: Drug-resistant tuberculosis (TB) poses a serious threat to public health in Kazakhstan. This paper presents findings related to TB treatment outcome and drug resistance status among people co-infected with human immunodeficiency virus (HIV) and TB in Kazakhstan. METHODS: A cohort study using data provided by the Kazakhstan Ministry of Health's National Tuberculosis Program for 2014 and 2015 was performed. The χ2 test and logistical regression were performed to understand factors associated with drug-resistant TB status and TB treatment outcome. RESULTS: In the bivariate analysis, drug-resistant TB status was significantly associated with year of TB diagnosis (P=0.001) and viral load (P=0.03). TB treatment outcome was significantly associated with age at diagnosis (p=0.01), antiretroviral (ARV) treatment (P<0.0001) and drug-resistant TB status (P=0.02). In the adjusted analysis, drug-resistant TB status was associated with an increased likelihood of successful completion of treatment with a successful outcome compared with treatment failure (odds ratio=6.94, 95% confidence interval 1.39-34.44). CONCLUSIONS: These results suggest that having drug-resistant TB is associated with a higher likelihood of completing treatment with successful outcome, even when controlling for receipt of ARV therapy.

Factors associated with linkage to care for HIV patients in Tajikistan.

There is a growing human immunodeficiency virus (HIV) epidemic in Tajikistan. This paper presents factors associated with linkage to HIV care among people aged 15 years and older in Tajikistan. This retrospective cross-sectional study used the Tajikistan Ministry of Health HIV registry data from patients diagnosed with HIV at age 15 years or older from 2000 to 2016. Chi squared tests and logistic regression models tested factors associated with linkage to care. A multivariable logistic regression model examined effect modifications. While linkage to care had an overall increase from 2000 to 2016, the odds of linkage were lower among certain sub-groups including among people in Dushanbe, men, people engaging in sex work, injection drug users, and older people. Regional differences exist with linkage to care, occurring least frequently in Dushanbe. While access to care and quality of care have increased significantly over time, findings suggest that linkage to care is low, especially in the capital city where many services are provided. Evaluation focusing on acceptability of HIV services should be undertaken to understand why certain people do not link with services. Additional research about the types of barriers to linking with HIV care is needed to increase linkage to HIV care.

Correlates of injection drug use among individuals admitted to public and private drug treatment facilities in Turkey.

BACKGROUND: The number of individuals seeking treatment for drug use has been increasing in recent years in Turkey. However, existing research on patterns and risk factors for drug use and how they vary by age and location in Turkey is limited. METHODS: We examined the socio-demographic characteristics, drug use behaviors, and treatment history of citizens admitted to inpatient substance use treatment at public and private facilities in Turkey during 2012 and 2013 and identified correlates of lifetime and current injection drug use. RESULTS: Of the 11,247 patients at the 22 public treatment centers in 2012-2013, a majority were male, lived with family, were unemployed, and had an average age of 27 years. Within private clinics (n=663), a higher proportion was female (9.7% private vs. 5.7% public), aged 11-17 years old (13% vs. 7.4%), used cannabis as their primary drug (18.4% vs. 13.2%), and had previously received drug treatment (57% vs. 47.2%). Within public centers, 40.4% reported ever injecting drugs and 33.7% reported injecting in the past 30 days; the corresponding percentages at private clinics were 22.5% and 18.1%. Significant predictors of injection drug use included being homeless, being a temporal employee or unemployed, having higher education, heroin as a preferred drug, having a longer duration of drug use, and prior drug treatment. CONCLUSION: Prevention and intervention efforts are needed to reduce the transition to heroin and injection drug use among youth as well as improve access to a variety of drug treatment options for people who use substances in Turkey.