Ncomeka Sineke

Drug-resistant tuberculosis (DR-TB) remains a major challenge to tuberculosis control, particularly in high-burden settings such as South Africa. Routine molecular diagnostic assays generate quantitative cycle threshold (Ct)-derived measurements from established resistance-associated targets; however, the extent to which these routinely generated data can be computationally characterized to distinguish molecular resistance profiles remains insufficiently explored. This study evaluated Ct-derived molecular markers associated with first- and second-line anti-tuberculosis drug-resistance classifications among Mycobacterium tuberculosis isolates from rural Eastern Cape, South Africa, using interpretable statistical modelling and machine-learning approaches. A retrospective analytical study was conducted using 2430 culture-confirmed M. tuberculosis specimens obtained from the National Health Laboratory Service. Ct-derived measurements for katG, inhA, gyrA, and rrs were analyzed using logistic regression, decision tree, random forest, extra trees, gradient boosting, support vector machine, extreme gradient boosting (XGBoost), and a multilayer perceptron neural network. Binary classification models were evaluated for isoniazid (INH), ethionamide (ETH), fluoroquinolones (FLQ), amikacin (AMK), capreomycin (CAP), and kanamycin (KAN), as well as multiclass resistance-profile classification. Among evaluable isolates, molecular resistance classifications were observed in 72.2% for INH, 31.1% for ETH, 13.2% for FLQ, and 6.9% each for AMK, CAP, and KAN. Random forest achieved the highest performance for five of the six binary classifications, whereas extra trees performed best for CAP. ROC-AUC values ranged from 0.994 to 1.000, with classification accuracies of 98.0–99.7%. Random forest also achieved the highest multiclass accuracy (98.4%), weighted F1-score (98.3%), and weighted ROC-AUC (0.999). Permutation feature importance predominantly identified katG-derived Ct measurements for INH, inhA for ETH, gyrA for FLQ, and rrs for AMK, CAP, and KAN, consistent with established molecular resistance relationships. The very high discriminatory performance should be interpreted in the context of the shared molecular-diagnostic origin of the Ct-derived predictors and resistance classifications, and therefore does not establish independent prediction of phenotypic resistance or the discovery of novel genetic determinants. Rather, these findings demonstrate that interpretable computational approaches can characterise and rank resistance-associated patterns embedded within routinely generated molecular diagnostic data, with potential utility for secondary analysis and molecular surveillance. Independent validation using phenotypic drug-susceptibility testing and/or genomic reference data is required to determine whether these approaches provide incremental predictive or clinical value beyond the underlying molecular assay.

Background: Drug-resistant tuberculosis (DR-TB) remains a major challenge in high-burden rural settings where biological, health-system, and social determinants interact. This study integrated whole-genome sequencing (WGS), mathematical modelling and Community-Based Participatory Research (CBPR)-informed principles to examine DR-TB within a conceptual biosocial framework. Methods: WGS data from 32 Mycobacterium tuberculosis isolates were analysed for resistance-associated mutations, phylogenetic lineages and genomic relatedness. A deterministic two-strain model explored drug-susceptible (DS) and DR-TB transmission. Hypothetical CBPR-informed scenarios modified transmission (β), treatment initiation (γ) and resistance amplification (α) parameters over 10 years. Results: Twenty-seven isolates (84.4%) showed resistance to at least one anti-TB drug, and 15 (46.9%) met MDR-TB criteria. Lineages 2 and 4 predominated. Under predefined assumptions, community-informed scenarios produced approximately 40–60% reductions in modelled DS-TB transmission and 20–35% reductions in DR-TB transmission relative to baseline. Conclusions: Integrating genomic evidence with transmission modelling and community-informed assumptions provides a useful hypothesis-generating framework for examining DR-TB. However, the modelled scenarios are exploratory and do not demonstrate the effectiveness of the CBPR intervention. Prospective studies using larger genomic datasets, longitudinal epidemiological data, and empirically measured community-level parameters are required.

Background: Early identification of individuals at heightened risk for drug-resistant tuberculosis (DR-TB) is essential for prioritising rapid diagnostic testing and targeted preventive care, particularly in resource-constrained rural settings. Although routine tuberculosis (TB) surveillance and household contact investigation records provide an accessible data source, their predictive capacity and utility for machine learning-based risk stratification remain under-evaluated. This study evaluated the predictive utility of routine demographic and household contact investigation data for programme-defined risk classification in rural Eastern Cape, South Africa, and identified critical data gaps relevant to the future development of clinically actionable DR-TB prediction models. Methods: A retrospective secondary analysis was conducted using anonymised, routinely collected household contact investigation data from 437 contacts of smear-positive pulmonary TB index cases in King Sabata Dalindyebo Municipality, Eastern Cape. The primary outcome was programme-assigned TB vulnerability classification (high versus low risk), rather than microbiologically confirmed TB or DR-TB. Bivariate analyses and multivariable logistic regression were used to assess factors associated with high-risk classification. Three supervised machine learning algorithms—logistic regression, random forest, and gradient boosting—were evaluated using an 80:20 stratified training–test split and five-fold cross-validation. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), area under the precision–recall curve (AUPRC), calibration measures, and decision curve analysis. A structured variable-gap assessment identified additional clinical, microbiological, and molecular predictors required for future DR-TB prediction. Results: Of 437 household contacts, 290 (66.4%) were classified as high risk and 147 (33.6%) as low risk. Age was significantly associated with programme-defined risk classification (χ² = 16.61, p = 0.005), whereas sex was not. Children aged 0–4 years had the highest proportion classified as high risk (35/41; 85.4%). In multivariable logistic regression, younger age (adjusted OR [aOR] = 2.41; 95% CI: 1.48–3.94) and larger household size (aOR = 1.12 per additional household member; 95% CI: 1.01–1.25) were independently associated with high-risk classification. Gradient boosting demonstrated the strongest cross-validated performance (AUROC = 0.657, SD = 0.052; AUPRC = 0.785, SD = 0.042), although overall discrimination remained modest. Household-level analyses demonstrated clustering of programme-defined vulnerability. The variable-gap assessment identified important unavailable predictors, including previous TB treatment, HIV status, diabetes, microbiological measures, molecular diagnostic results, and drug-susceptibility information. Conclusions: Routine household contact investigation data provide measurable but limited predictive information for programme-defined TB vulnerability, with age and household size emerging as important baseline factors. However, these data should not be interpreted as directly predicting microbiologically confirmed DR-TB. Development of robust and clinically meaningful DR-TB prediction models in rural African settings will require integration of routine programme data with clinical characteristics, molecular diagnostic results, and drug-susceptibility testing, followed by appropriate internal and external validation.

Background Drug-resistant tuberculosis remains a major public health challenge, particularly in high-burden settings where rapid characterization of resistance patterns is important for surveillance. Molecular diagnostics routinely detect resistance-associated targets, but these are commonly evaluated individually. This study investigated whether cycle threshold-derived molecular marker-detection profiles form reproducible resistance-associated networks among culture-positive Mycobacterium tuberculosis specimens from the Eastern Cape, South Africa. Methods A retrospective laboratory-based molecular epidemiology study analysed 2,430 unique culture-positive M. tuberculosis specimens tested between January 2021 and December 2024. Binary detection indicators were generated for inhA , katG , gyrA1 , gyrA2 , gyrA3 , and rrs. Marker co-detection and resistance relationships were examined using phi correlation, association-rule mining, resistance co-occurrence analysis, and integrated network modelling with centrality metrics. Results Simultaneous detection of all six markers occurred in 95.8% (2,327/2,430) of specimens. Strong correlations were observed between gyrA1–gyrA2 (φ = 0.84), gyrA1–gyrA3 (φ = 0.84), gyrA2–gyrA3 (φ = 0.81), and inhA–rrs (φ = 1.00). The integrated network comprised 17 nodes and 88 statistically supported edges, with a density of 0.647. Second-line resistance had the highest degree and betweenness centrality, while amikacin, capreomycin, kanamycin, and composite injectable resistance formed a tightly interconnected module. Association-rule analysis identified recurrent multidimensional marker combinations associated with injectable resistance. Conclusions Cycle threshold-derived molecular markers and resistance outcomes formed interconnected co-detection and co-occurrence architectures rather than isolated patterns. Network analysis identified coordinated isoniazid/ethionamide, fluoroquinolone, injectable, and second-line resistance modules. These findings demonstrate the potential utility of network-based approaches for interpreting routinely generated molecular tuberculosis surveillance data, though validation with sequence-confirmed and longitudinal data is required.

Tuberculosis (TB) treatment outcomes may differ among individuals with similar routinely recorded demographic and clinical characteristics, suggesting limitations in the explanatory scope of programme surveillance data. This study evaluated the capacity of routinely collected demographic, clinical, socioeconomic, and behavioural variables to explain TB treatment outcomes and used identified evidence gaps to define priorities for complementary molecular research. We conducted a retrospective secondary analysis of anonymised routine TB programme data from 422 adults treated in public healthcare facilities in rural OR Tambo District Municipality, Eastern Cape, South Africa, between January 2018 and December 2022. Treatment success was defined as cure or treatment completion, whereas unsuccessful outcomes included death, treatment failure, or loss to follow-up. The explanatory and predictive performance of logistic regression, Poisson regression, and Random Forest models was evaluated, alongside a structured evidence-gap analysis comparing variables available in routine surveillance with molecular determinants of treatment response and drug resistance identified in the literature. Previous TB treatment was associated with lower odds of treatment success in univariable analysis (OR 0.48, 95% CI 0.29–0.78; p=0.003), but the association attenuated after adjustment (OR 0.64, 95% CI 0.38–1.09; p=0.10). HIV status was not independently associated with treatment success in logistic (OR 0.80, 95% CI 0.49–1.29; p=0.36) or Poisson regression (aRR 0.95, 95% CI 0.66–1.37; p=0.79). Random Forest achieved a higher average precision than logistic regression (0.907 versus 0.810), demonstrating improved prediction from routinely available variables; however, molecular mechanisms could not be evaluated because genomic, transcriptomic, immunological, and pharmacological measures were absent from the dataset. These findings define an important boundary between programme-level prediction and mechanistic explanation. We propose a program-to-molecular translational framework in which routine surveillance identifies clinically relevant patterns and knowledge gaps that can subsequently be investigated using genomic, transcriptomic, and functional molecular approaches. This integration may strengthen mechanistic understanding of heterogeneous TB treatment responses and inform future precision-oriented TB research.

Background: Tuberculosis (TB) remains a major public health challenge in rural and socioeconomically disadvantaged communities, where limited access to health information, stigma, healthcare barriers, and social inequalities may influence knowledge and engagement with TB services. This study examined TB knowledge, personal and family experiences of TB, and perceived barriers to healthcare engagement among community members in rural Eastern Cape, South Africa. The findings were further interpreted through the complementary concepts of knowledge democracy and reparative engagement to explore how community and experiential knowledge could inform culturally responsive and participatory approaches to TB education. Methods: A secondary analysis was conducted using data from a cross-sectional Knowledge, Attitudes and Practices (KAP) survey involving 131 adults from Ntabankulu, Eastern Cape, South Africa. Descriptive statistics were used to summarise participant characteristics and TB knowledge levels. Associations between participant characteristics and TB knowledge were assessed using Fisher’s exact test and the Kruskal–Wallis test, as appropriate, with statistical significance set at p < 0.05. Findings relating to personal and family experience of TB, perceived barriers to healthcare engagement, and preferred approaches to TB education were interpreted using knowledge democracy and reparative engagement as complementary conceptual lenses. These frameworks were applied retrospectively and were not measured as participant-level constructs. Results: Most participants (64.9%) demonstrated moderate TB knowledge, while 23.7% demonstrated high knowledge. Educational attainment was significantly associated with TB knowledge (p < 0.001), with higher educational attainment corresponding to greater knowledge. Participants with personal or family experience of TB also demonstrated greater TB knowledge, suggesting that experiential exposure may contribute to community understanding of the disease. Perceived barriers to TB testing and healthcare engagement included stigma, fear of diagnosis, misinformation, and constraints in accessing healthcare services, reflecting individual, interpersonal, and structural influences. Participants identified survivor storytelling, peer and community education, community health worker involvement, culturally and linguistically appropriate communication, and participatory community dialogue as preferred approaches to strengthening TB education. Conclusions: TB knowledge and healthcare engagement in rural Eastern Cape communities are shaped by the interplay of educational attainment, experiential knowledge, and broader social and structural conditions. Interpreted through the lens of knowledge democracy and reparative engagement, the findings suggest that conventional biomedical health education could be strengthened by recognizing communities not only as recipients of health information but also as contributors to knowledge production and locally responsive solutions. Community-informed approaches incorporating lived experience, peer education, community health workers, culturally appropriate communication, and participatory dialogue warrant prospective evaluation. Future co-designed and implementation research should determine whether these approaches improve TB knowledge, reduce stigma, strengthen trust, and enhance engagement with TB testing, prevention, treatment, and care.

Background Tuberculosis (TB) treatment outcomes remain suboptimal in many high-burden settings because biomedical care alone does not adequately address the socioeconomic, community, and health system barriers influencing treatment adherence and completion. This study investigated socioeconomic and clinical determinants of TB treatment outcomes and treatment duration and modelled the potential impact of integrated health economics, clinical governance, and community engagement interventions in rural Eastern Cape, South Africa. Methods A retrospective observational study analysed routinely collected TB programme data from 538 patients in O.R. Tambo District. Multivariable logistic regression identified predictors of treatment success, while multiple linear regression and Cox proportional hazards models assessed determinants of treatment duration. Logistic Regression, Random Forest, and Extreme Gradient Boosting (XGBoost) models were developed to predict treatment outcomes. Scenario-based health systems modelling estimated the potential impact of integrated interventions under model assumptions. Results Overall, 411 patients (76.4%) achieved successful treatment outcomes, while 127 (23.6%) had unsuccessful outcomes. Socioeconomic vulnerability was substantial, with 77.5% reporting no regular income and 66.4% being unemployed. After adjustment, long treatment regimen remained the only independent predictor of treatment success, reducing the odds of success by 58% compared with short-course therapy (aOR 0.42; 95% CI 0.20–0.89; p = 0.024). Absence of income, unemployment, HIV co-infection, and long treatment regimen were associated with prolonged treatment duration and delayed completion. XGBoost demonstrated the strongest predictive performance (AUC = 0.89; accuracy = 85%). Scenario-based modelling projected that the integrated intervention framework could increase treatment success from 76.4% to 95.3% and reduce unsuccessful outcomes from 23.6% to 4.7%. Conclusions TB treatment outcomes in rural South Africa are shaped by the interaction of socioeconomic, clinical, community, and health system factors. Although prospective validation is required, integrating socioeconomic support with strengthened clinical governance and community engagement may improve treatment outcomes, reduce programme inefficiencies, and advance patient-centred TB care in resource-constrained settings.

Background: Drug-resistant tuberculosis (DR-TB) remains a major threat to global tuberculosis control, particularly in high-burden rural settings where transmission is driven by both biological and socio-structural determinants. Although genomic surveillance and mathematical transmission modelling have improved understanding of resistance evolution and transmission dynamics, these approaches often insufficiently incorporate community-level behavioral and social drivers of disease spread. This study integrated Community-Based Participatory Research (CBPR) within a genomic–epidemiological modelling framework to develop a biosocial understanding of DR-TB transmission dynamics in a rural South African setting. Methods: Whole-genome sequencing (WGS) was performed on 32 Mycobacterium tuberculosis isolates to identify resistance-associated mutations, phylogenetic lineages, and potential transmission clusters. A deterministic two-strain transmission model distinguishing drug-sensitive (DS) and drug-resistant (DR) tuberculosis was developed to simulate transmission dynamics. CBPR-informed mechanisms were incorporated into the model through modifications of key epidemiological parameters, including transmission rate (β), treatment initiation rate (γ), and resistance amplification rate (α), reflecting the influence of community engagement, treatment adherence, and health-seeking behavior. Scenario-based simulations evaluating varying levels of community-engaged interventions were conducted over a 10-year period. Results: A substantial burden of drug resistance was observed, with 84.4% of isolates resistant to at least one anti-tuberculosis drug. Multidrug-resistant tuberculosis (MDR-TB) accounted for 46.9% of isolates, while recurrent combinations of resistance-associated mutations suggested ongoing transmission of resistant strain lineages. Lineage 2 (Beijing genotype) and Lineage 4 predominated, with advanced resistance patterns occurring mainly within Lineage 2 isolates. Model simulations indicated that CBPR-informed interventions could reduce DS-TB transmission by approximately 40–60% and DR-TB transmission by 20–35%. Scenario-based estimates also indicated a higher transmission potential for DR-TB (R₀ ≈ 2.04) than for DS-TB (R₀ ≈ 1.29). Community-engaged interventions reduced transmission by improving treatment adherence, earlier diagnosis, and enhanced infection-prevention behaviors. Conclusions: Integrating CBPR into genomic and transmission modelling frameworks provides a novel biosocial approach for understanding tuberculosis dynamics in high-burden settings. The findings suggest that community-engaged interventions can substantially influence key epidemiological drivers of transmission and resistance amplification. Embedding community participation within TB surveillance and control strategies may strengthen efforts to reduce transmission, improve treatment continuity, and address the social determinants underpinning DR-TB persistence in rural settings.

Background: The success rates for treating drug-resistant tuberculosis (DR-TB) in programmatic settings have been unsatisfactory. By identifying the factors that predict treatment outcomes, we can implement effective corrective measures that will significantly enhance patient management and improve results for those with DR-TB. Aim: This study aimed to investigate predictive factors influencing treatment outcomes among DR-TB patients, focusing on the combined effects of body mass index (BMI), human immunodeficiency virus (HIV) status, comorbidities, socioeconomic factors, substance use and DR-TB type. Setting: The study was conducted in rural Eastern Cape, South Africa. Methods: This retrospective cohort study was designed to utilise logistic regression models on data from 200 patient medical records. We examined variables including BMI, HIV co-infection, comorbidities (e.g. diabetes, hypertension), income, substance use and DR-TB classifications (multidrug-resistant, rifampicin-resistant, pre-extensively drug-resistant, extensively drug-resistant). Results: Key findings indicate a weak association between lower BMI and reduced treatment success (odds ratio [OR]: 0.92, 95% confidence interval [CI]: 0.81–1.05). HIV-positive status was marginally associated with lower treatment success (OR: 0.89, 95% CI: 0.75–1.12), while income level and substance use emerged as stronger predictors (e.g. substance use OR: 0.72, 95% CI: 0.60–0.88). Among DR-TB types, extensively drug-resistant tuberculosis patients exhibited the poorest outcomes (OR: 0.55, 95% CI: 0.40–0.75). The multivariate model achieved an accuracy of 63.1%, suggesting limited predictive power of BMI and HIV alone and highlighting the significant influence of comorbidities, socioeconomic status and behavioural factors. Conclusion: These findings underscore the importance of a multidimensional approach in improving DR-TB treatment outcomes through tailored clinical and social interventions. Contribution: The study noted limited connections between DR-TB and various comorbidities. It highlights the necessity of managing coexisting conditions in DR-TB patients because of their significant impact on treatment outcomes. Customised interventions are essential for those with severe or complex comorbidities.

Background: Influenza commonly known as flu is a highly contagious viral infection that affects millions of people globally each year, causing major morbidity, mortality, and economic consequences. Despite vaccine availability, seasonal flu outbreaks continue, exacerbated by virus mutations and varied transmission patterns. The Eastern Cape region of South Africa faces specific obstacles when it comes to flu management, such as limited healthcare access and dependence on traditional treatments This study investigated the relationships among flu severity, symptom duration, and management strategies using regression models, statistical testing, and decision trees. Data included key metrics; severity scale, symptom duration, and management types (home remedies, medical treatment, prescription medication, and over-the-counter treatments). Regression analysis showed that the severity scale is influenced by symptom duration (r = 0.88, p < 0.001), with longer durations correlating with higher perceived severity. However, management strategies did not significantly affect severity (ANOVA F = 1.415, p = 0.291). Logistic regression demonstrated that severity scale positively predicted seeking medical attention (pseudo R² = 0.624, p = 0.038). Decision tree analysis identified prescription medication and symptom duration as key factors predicting high or low severity. Findings indicate that symptom duration strongly correlates with severity, while management strategies, including home remedies and medical treatments, show mixed effects. No significant interaction effects between management types were observed (p > 0.05). These results emphasize the need for timely interventions to reduce symptom duration and perceived severity. Limitations include a small sample size and potential biases in self-reported data. Future research should explore larger datasets and test intervention efficacy to refine symptom management strategies.