Prediction of precancerous cervical cancer lesions among women living with HIV on antiretroviral therapy in Uganda: a comparison of supervised machine learning algorithms.

BACKGROUND: Cervical cancer (CC) is among the most prevalent cancer types among women with the highest prevalence in low- and middle-income countries (LMICs). It is a curable disease if detected early. Machine learning (ML) techniques can aid in early detection and prediction thus reducing screening and treatment costs. This study focused on women living with HIV (WLHIV) in Uganda. Its aim was to identify the best predictors of CC and the supervised ML model that best predicts CC among WLHIV. METHODS: Secondary data that included 3025 women from three health facilities in central Uganda was used. A multivariate binary logistic regression and recursive feature elimination with random forest (RFERF) were used to identify the best predictors. Five models; logistic regression (LR), random forest (RF), K-Nearest neighbor (KNN), support vector machine (SVM), and multi-layer perceptron (MLP) were applied to identify the out-performer. The confusion matrix and the area under the receiver operating characteristic curve (AUC/ROC) were used to evaluate the models. RESULTS: The results revealed that duration on antiretroviral therapy (ART), WHO clinical stage, TPT status, Viral load status, and family planning were commonly selected by the two techniques and thus highly significant in CC prediction. The RF from the RFERF-selected features outperformed other models with the highest scores of 90% accuracy and 0.901 AUC. CONCLUSION: Early identification of CC and knowledge of the risk factors could help control the disease. The RF outperformed other models applied regardless of the selection technique used. Future research can be expanded to include ART-naïve women in predicting CC.

Association of impaired kidney function with mortality in rural Uganda: results of a general population cohort study.

OBJECTIVE: To determine the association between baseline kidney function and subsequent all-cause mortality. DESIGN AND SETTING: A general population-based cohort study from rural Uganda. PARTICIPANTS: People aged 18 years and above with measured baseline estimated glomerular filtration rate (eGFR), recruited from survey rounds in 2011-2012 or 2014-2015 and followed up to March 2019. OUTCOME MEASURE: The primary outcome was all-cause mortality, identified through reports from community health workers and verified by verbal autopsy. The association between baseline eGFR category and mortality was determined using multivariable Cox regression. RESULTS: Of 5812 participants in both rounds, we included 5678 (97.7%) participants with kidney function and mortality data; the median age was 36 years (IQR 24-50), 60.7% were female, 10.3% were hypertensive, 9.8% were HIV-positive and 1.5% were diabetic. During a median follow-up of 5.0 years (IQR 3.7-6.0) there were 140 deaths. In age-adjusted and sex-adjusted analyses, eGFR <45 mL/min/1.73 m2 at baseline was associated with a 5.97 (95% CI 2.55 to 13.98) increased risk of mortality compared with those with baseline eGFR >90 mL/min/1.73 m2. After inclusion of additional confounders (HIV, body mass index, diabetes, hypertension, alcohol and smoking status) into the model, eGFR <45 mL/min/1.73 m2 at baseline remained strongly associated with mortality (HR 6.12, 95% CI 2.27 to 16.45), although the sample size fell to 3102. Test for trend showed strong evidence (p<0.001) that the rate of mortality increased progressively as the category of baseline kidney function decreased. When very high eGFR was included as a separate category in age-adjusted and sex-adjusted analyses, baseline eGFR ≥120 mL/min/1.73 m2 was associated with increased risk of mortality (HR 2.68, 95% CI 1.47 to 4.87) compared with the reference category of 90-119 mL/min/1.73 m2. CONCLUSION: In a prospective cohort in rural Uganda we found that impaired baseline kidney function was associated with subsequently increased total mortality. Improved understanding of the determinants of kidney disease and its progression is needed in order to inform interventions for prevention and treatment.

Effectiveness of low-dose theophylline for the management of biomass-associated COPD (LODOT-BCOPD): study protocol for a randomized controlled trial.

BACKGROUND: COPD is a leading cause of death globally, with the majority of morbidity and mortality occurring in low- and middle-income country (LMIC) settings. While tobacco-smoke exposure is the most important risk factor for COPD in high-income settings, household air pollution from biomass smoke combustion is a leading risk factor for COPD in LMICs. Despite the high burden of biomass smoke-related COPD, few studies have evaluated the efficacy of pharmacotherapy in this context. Currently recommended inhaler-based therapy for COPD is neither available nor affordable in most resource-limited settings. Low-dose theophylline is an oral, once-a-day therapy, long used in high-income countries (HICs), which has been proposed for the management of COPD in LMICs in the absence of inhaled steroids and/or bronchodilators. The Low-dose Theophylline for the Management of Biomass-Associated COPD (LODOT-BCOPD) trial investigates the clinical efficacy and cost-effectiveness of low-dose theophylline for the management of biomass-related COPD in a low-income setting. METHODS: LODOT-BCOPD is a randomized, double-blind, placebo-controlled trial to test the efficacy of low-dose theophylline in improving respiratory symptoms in 110 participants with moderate to severe COPD in Central Uganda. The inclusion criteria are as follows: (1) age 40 to 80 years, (2) full-time resident of the study area, (3) daily biomass exposure, (4) post-bronchodilator FEV1/FVC below the 5th percentile of the Global Lung Initiative mixed ethnic reference population, and (5) GOLD Grade B-D COPD. Participants will be randomly assigned to receive once daily low-dose theophylline (200 mg ER, Unicontin-E) or placebo for 52 weeks. All participants will receive education about self-management of COPD and rescue salbutamol inhalers. We will measure health status using the St. George's Respiratory Questionnaire (SGRQ) and quality of life using the EuroQol-5D (EQ-5D) at baseline and every 6 months. In addition, we will assess household air pollution levels, serum inflammatory biomarkers (fibrinogen, hs-CRP), and theophylline levels at baseline, 1 month, and 6 months. The primary outcome is change in SGRQ score at 12 months. Lastly, we will assess the cost-effectiveness of the intervention by calculating quality-adjusted life years (QALYs) from the EQ-5D. TRIAL REGISTRATION: ClinicalTrials.gov  NCT03984188 . Registered on June 12, 2019 TRIAL ACRONYM: Low-dose Theophylline for the Management of Biomass-Associated COPD (LODOT-BCOPD).