Use of machine learning techniques to identify HIV predictors for screening in sub-Saharan Africa.
Ontology highlight
ABSTRACT: Aim
HIV prevention measures in sub-Saharan Africa are still short of attaining the UNAIDS 90-90-90 fast track targets set in 2014. Identifying predictors for HIV status may facilitate targeted screening interventions that improve health care. We aimed at identifying HIV predictors as well as predicting persons at high risk of the infection.Method
We applied machine learning approaches for building models using population-based HIV Impact Assessment (PHIA) data for 41,939 male and 45,105 female respondents with 30 and 40 variables respectively from four countries in sub-Saharan countries. We trained and validated the algorithms on 80% of the data and tested on the remaining 20% where we rotated around the left-out country. An algorithm with the best mean f1 score was retained and trained on the most predictive variables. We used the model to identify people living with HIV and individuals with a higher likelihood of contracting the disease.Results
Application of XGBoost algorithm appeared to significantly improve identification of HIV positivity over the other five algorithms by f1 scoring mean of 90% and 92% for males and females respectively. Amongst the eight most predictor features in both sexes were: age, relationship with family head, the highest level of education, highest grade at that school level, work for payment, avoiding pregnancy, age at the first experience of sex, and wealth quintile. Model performance using these variables increased significantly compared to having all the variables included. We identified five males and 19 females individuals that would require testing to find one HIV positive individual. We also predicted that 4·14% of males and 10.81% of females are at high risk of infection.Conclusion
Our findings provide a potential use of the XGBoost algorithm with socio-behavioural-driven data at substantially identifying HIV predictors and predicting individuals at high risk of infection for targeted screening.
SUBMITTER: Mutai CK
PROVIDER: S-EPMC8325403 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:IntroductionSignificant regional variations in the HIV epidemic hurt effective common interventions in sub-Saharan Africa. It is crucial to analyze HIV positivity distributions within clusters and assess the homogeneity of countries. We aim at identifying clusters of countries based on socio-behavioural predictors of HIV for screening.MethodWe used an agglomerative hierarchical, unsupervised machine learning, approach for clustering to analyse data for 146,733 male and 155,622 female respondents from 13 sub-Saharan African countries with 20 and 26 features, respectively, using Population-based HIV Impact Assessment (PHIA) data from the survey years 2015-2019. We employed agglomerative hierarchical clustering and optimal silhouette index criterion to identify clusters of countries based on the similarity of socio-behavioural characteristics. We analyse the distribution of HIV positivity with socio-behavioural predictors of HIV within each cluster.ResultsTwo principal components were obtained, with the first describing 62.3% and 70.1% and the second explaining 18.3% and 20.6% variance of the total socio-behavioural variation in females and males, respectively. Two clusters per sex were identified, and the most predictor features in both sexes were: relationship with family head, enrolled in school, circumcision status for males, delayed pregnancy, work for payment in last 12 months, Urban area indicator, known HIV status and delayed pregnancy. The HIV positivity distribution with these variables was significant within each cluster.Conclusions /findingsThe findings provide a potential use of unsupervised machine learning approaches for substantially identifying clustered countries based on the underlying socio-behavioural characteristics.
| S-EPMC10354889 | biostudies-literature
Project description:Sub-Saharan Africa represents 69% of the total number of individuals living with HIV infection worldwide and 72% of AIDS deaths globally. Pulmonary infection is a common and frequently fatal complication, though little is known regarding the lower airway microbiome composition of this population. Our objectives were to characterize the lower airway microbiome of Ugandan HIV-infected patients with pneumonia, to determine relationships with demographic, clinical, immunological, and microbiological variables and to compare the composition and predicted metagenome of these communities to a comparable cohort of patients in the US (San Francisco). Bronchoalveolar lavage samples from a cohort of 60 Ugandan HIV-infected patients with acute pneumonia were collected. Amplified 16S ribosomal RNA was profiled and aforementioned relationships examined. Ugandan airway microbiome composition and predicted metagenomic function were compared to US HIV-infected pneumonia patients. Among the most common bacterial pulmonary pathogens, Pseudomonas aeruginosa was most prevalent in the Ugandan cohort. Patients with a richer and more diverse airway microbiome exhibited lower bacterial burden, enrichment of members of the Lachnospiraceae and sulfur-reducing bacteria and reduced expression of TNF-alpha and matrix metalloproteinase-9. Compared to San Franciscan patients, Ugandan airway microbiome were significantly richer, and compositionally distinct with predicted metagenomes that encoded a multitude of distinct pathogenic pathways e.g secretion systems. Ugandan pneumonia-associated airway microbiome is compositionally and functionally distinct from those detected in comparable patients in developed countries, a feature which may contribute to adverse outcomes in this population. Please note that the data from the comparable cohort of patients in the USUS data was published as supplemental material of PMID: 22760045 but not submitted to GEO The 'patient_info.txt' contains 12 clinical, 7 immunological and 3 microbiological variables for each patient.
2014-02-20 | GSE52791 | GEO
Project description:Sub-Saharan Africa represents 69% of the total number of individuals living with HIV infection worldwide and 72% of AIDS deaths globally. Pulmonary infection is a common and frequently fatal complication, though little is known regarding the lower airway microbiome composition of this population. Our objectives were to characterize the lower airway microbiome of Ugandan HIV-infected patients with pneumonia, to determine relationships with demographic, clinical, immunological, and microbiological variables and to compare the composition and predicted metagenome of these communities to a comparable cohort of patients in the US (San Francisco). Bronchoalveolar lavage samples from a cohort of 60 Ugandan HIV-infected patients with acute pneumonia were collected. Amplified 16S ribosomal RNA was profiled and aforementioned relationships examined. Ugandan airway microbiome composition and predicted metagenomic function were compared to US HIV-infected pneumonia patients. Among the most common bacterial pulmonary pathogens, Pseudomonas aeruginosa was most prevalent in the Ugandan cohort. Patients with a richer and more diverse airway microbiome exhibited lower bacterial burden, enrichment of members of the Lachnospiraceae and sulfur-reducing bacteria and reduced expression of TNF-alpha and matrix metalloproteinase-9. Compared to San Franciscan patients, Ugandan airway microbiome were significantly richer, and compositionally distinct with predicted metagenomes that encoded a multitude of distinct pathogenic pathways e.g secretion systems. Ugandan pneumonia-associated airway microbiome is compositionally and functionally distinct from those detected in comparable patients in developed countries, a feature which may contribute to adverse outcomes in this population. Please note that the data from the comparable cohort of patients in the USUS data was published as supplemental material of PMID: 22760045 but not submitted to GEO The 'patient_info.txt' contains 12 clinical, 7 immunological and 3 microbiological variables for each patient. The G2 PhyloChip microarray platform (commercially available from Second Genome, Inc.) was used to profile bacteria in lower airway samples from 60 subjects
2014-02-20 | E-GEOD-52791 | biostudies-arrayexpress
Project description:BackgroundBabies born early and/or small for gestational age in Low and Middle-income countries (LMICs) contribute substantially to global neonatal and infant mortality. Tracking this metric is critical at a population level for informed policy, advocacy, resources allocation and program evaluation and at an individual level for targeted care. Early prenatal ultrasound examination is not available in these settings, gestational age (GA) is estimated using new-born assessment, last menstrual period (LMP) recalls and birth weight, which are unreliable. Algorithms in developed settings, using metabolic screen data, provided GA estimates within 1-2 weeks of ultrasonography-based GA. We sought to leverage machine learning algorithms to improve accuracy and applicability of this approach to LMICs settings.MethodsThis study uses data from AMANHI-ACT, a prospective pregnancy cohorts in Asia and Africa where early pregnancy ultrasonography estimated GA and birth weight are available and metabolite screening data in a subset of 1318 new-borns were also available. We utilized this opportunity to develop machine learning (ML) algorithms. Random Forest Regressor was used where data was randomly split into model-building and model-testing dataset. Mean absolute error (MAE) and root mean square error (RMSE) were used to evaluate performance. Bootstrap procedures were used to estimate confidence intervals (CI) for RMSE and MAE. For pre-term birth identification ROC analysis with bootstrap and exact estimation of CI for area under curve (AUC) were performed.ResultsOverall model estimated GA had MAE of 5.2 days (95% CI 4.6-6.8), which was similar to performance in SGA, MAE 5.3 days (95% CI 4.6-6.2). GA was correctly estimated to within 1 week for 85.21% (95% CI 72.31-94.65). For preterm birth classification, AUC in ROC analysis was 98.1% (95% CI 96.0-99.0; p < 0.001). This model performed better than Iowa regression, AUC Difference 14.4% (95% CI 5-23.7; p = 0.002).ConclusionsMachine learning algorithms and models applied to metabolomic gestational age dating offer a ladder of opportunity for providing accurate population-level gestational age estimates in LMICs settings. These findings also point to an opportunity for investigation of region-specific models, more focused feasible analyte models, and broad untargeted metabolome investigation.
| S-EPMC8424940 | biostudies-literature
Project description:To summarize recent developments for HIV-associated malignancies (HIVAM) in low-income and middle-income countries (LMIC) with particular focus on sub-Saharan Africa (SSA).Antiretroviral therapy (ART) scale-up is leading to epidemiologic transitions in LMIC similar to high-income countries, with aging and growth of HIV-infected populations, declining infectious deaths, increasing cancer deaths, and transitions from AIDS-defining cancers to non-AIDS defining cancers. Despite ART scale-up, the HIVAM burden remains high including an enormous AIDS-defining cancers burden in SSA. For Kaposi sarcoma, patients treated with ART and chemotherapy can experience good outcomes even in rural SSA, but Kaposi sarcoma heterogeneity remains insufficiently understood including virologic, immunologic, and inflammatory features that may be unique to LMIC. For cervical cancer, scale-up of prevention efforts including vaccination and screening is underway, with benefits already apparent despite continuing high disease burden. For non-Hodgkin lymphoma, curative treatment is possible in the ART era even in SSA, and multifaceted approaches can improve outcomes further. For many other prevalent HIVAM, care and research efforts are being established to guide treatment and prevention specifically in LMIC.Sustained investment for HIVAM in LMIC can help catalyze a cancer care and research agenda that benefits HIV-positive and HIV-negative patients worldwide.
| S-EPMC5241291 | biostudies-other
Project description:BACKGROUND:Patients with acute HIV-1 infection (AHI) have elevated infectivity, but cannot be diagnosed using antibody-based testing. Approaches to screen patients for AHI are urgently needed to enable counselling and treatment to reduce onward transmission. METHODS:We pooled data from four African studies of high-risk adults that evaluated symptoms and signs compatible with acute retroviral syndrome and tested for HIV-1 at each visit. AHI was defined as detectable plasma viral load or p24 antigen in an HIV-1-antibody-negative patient who subsequently seroconverted. Using generalized estimating equation, we identified symptoms, signs, and demographic factors predictive of AHI, adjusting for study site. We assigned a predictor score to each statistically significant predictor based on its beta coefficient, summing predictor scores to calculate a risk score for each participant. We evaluated the performance of this algorithm overall and at each site. RESULTS:We compared 122 AHI visits with 45?961 visits by uninfected patients. Younger age (18-29 years), fever, fatigue, body pains, diarrhoea, sore throat, and genital ulcer disease were independent predictors of AHI. The overall area under the receiver operating characteristics curve (AUC) for the algorithm was 0.78, with site-specific AUCs ranging from 0.61 to 0.89. A risk score of at least 2 would indicate AHI testing for 5-50% of participants, substantially decreasing the number needing testing. CONCLUSION:Our targeted risk score algorithm based on seven characteristics reduced the number of patients needing AHI testing and had good performance overall. We recommend this risk score algorithm for use by HIV programs in sub-Saharan Africa with capacity to test high-risk patients for AHI.
| S-EPMC4714928 | biostudies-literature
Project description:Konzo, a disease characterized by sudden, irreversible spastic paraparesis, affecting up to 10% of the population in some regions of Sub-Saharan Africa during outbreaks and is strongly associated with dietary exposure to cyanogenic bitter cassava. The molecular mechanisms underlying the development of konzo, remain largely unknown. Here, through an analysis of 16 individuals with konzo and matched healthy controls from the same outbreak zones, we identified 117 differentially methylated loci involved in numerous biological processes that may identify cyanogenic- sensitive regions of the genome, providing the first study of epigenomic alterations associated with sub-lethal cyanide exposure and a clinical phenotype.
2022-12-31 | GSE180119 | GEO
Project description:In sub-Saharan Africa (SSA), antiretroviral therapy (ART) can prolong life for HIV-infected patients. However, patients initiating ART, especially in routine treatment programs, commonly dropout from care either due to death or loss to follow-up.In a cohort of HIV-infected patients initiating ART at a public sector clinic in Uganda, we assessed predictors of dropout from care (a composite outcome combining death and loss to follow-up). From a large set of socio-demographic, clinical, and laboratory variables routinely collected at ART initiation, we selected those predicting dropout at P <0.1 in unadjusted analyses for inclusion into a multivariable proportional hazards regression model. We then used a stepwise backward selection procedure to identify variables which independently predicted dropout at P <0.05.Data from 5,057 patients were analyzed. The median age was 33 years (IQR 28 to 40) and 27.4% had CD4+ T-cell counts <100 cells/μL at ART initiation. The median duration of follow-up was 24 months (IQR = 14 to 42, maximum follow-up = 64 months). Overall dropout was 26.9% (established cumulative mortality = 2.3%, loss to follow-up = 24.6%), 5.6% were transferred to other service providers, and 67.5% were retained in care. A diagnosis of Kaposi's sarcoma (hazard ratio (HR) = 3.3, 95% CI 2.5 to 4.5); HIV-associated dementia (HR = 2.6, 95% CI 1.5 to 4.6); history of cryptococcosis (HR = 2.2, 95% CI 1.4 to 3.3); and reduced hemoglobin concentration (<11 g/dl versus ≥13.8 g/dl (HR = 1.9, 95% CI 1.6 to 2.2) were strong predictors of dropout. Other independent predictors of dropout were: year of ART initiation; weight loss ≥10%; reduced total lymphocyte count; chronic diarrhea; male sex; young age (≤28 years); and marital status.Among HIV-infected patients initiating ART at a public sector clinic in SSA, biological factors that usually predict death were especially predictive of dropout. As most of the dropouts were lost to follow-up, this observation suggests that many losses to follow-up may have died. Future studies are needed to identify appropriate interventions that may improve both individual-level patient outcomes and outcome ascertainment among HIV-infected ART initiators in this setting.
| S-EPMC4736127 | biostudies-other
Project description:ObjectiveTo better understand diverse experts' views about the ethical implications of ongoing research funded by the National Institutes of Health that uses machine learning to predict HIV/AIDS risk in sub-Saharan Africa (SSA) based on publicly available Demographic and Health Surveys data.DesignThree rounds of semi-structured surveys in an online expert panel using a modified Delphi approach.ParticipantsExperts in informatics, African public health and HIV/AIDS and bioethics were invited to participate.MeasuresPerceived importance of or agreement about relevance of ethical issues on 5-point unipolar Likert scales. Qualitative data analysis identified emergent themes related to ethical issues and development of an ethical framework and recommendations for open-ended questions.ResultsOf the 35 invited experts, 22 participated in the online expert panel (63%). Emergent themes were the inclusion of African researchers in all aspects of study design, analysis and dissemination to identify and address local contextual issues, as well as engagement of communities. Experts focused on engagement with health and science professionals to address risks, benefits and communication of findings. Respondents prioritised the mitigation of stigma to research participants but recognised trade-offs between privacy and the need to disseminate findings to realise public health benefits. Strategies for responsible communication of results were suggested, including careful word choice in presentation of results and limited dissemination to need-to-know stakeholders such as public health planners.ConclusionExperts identified ethical issues specific to the African context and to research on sensitive, publicly available data and strategies for addressing these issues. These findings can be used to inform an ethical implementation framework with research stage-specific recommendations on how to use publicly available data for machine learning-based predictive analytics to predict HIV/AIDS risk in SSA.
| S-EPMC8320245 | biostudies-literature
Project description:BackgroundIn 2015 around 15 million people living with HIV were receiving antiretroviral treatment (ART) in sub-Saharan Africa. Sustained provision of ART, though both prudent and necessary, creates substantial long-term fiscal obligations for countries affected by HIV/AIDS. As donor assistance for health remains constrained, novel financing mechanisms are needed to augment funding domestic sources. We explore how Innovative Financing has been used to co-finance domestic HIV/AIDS responses. Based on analysis of non-health sectors, we identify innovative financing instruments that could be used in the HIV response.MethodsWe undertook a systematic review to identify innovative financing instruments used for (1) domestic HIV/AIDS financing in sub-Saharan Africa (2) international health financing and (3) financing in non-health sectors. We analyzed peer-reviewed and grey literature published between 2002 and 2014. We examined the nature and volume of funds mobilized with innovative financing, then in consultation with leading experts, identified instruments that held potential for financing the HIV response.ResultsOur analysis revealed three innovative financing instruments in use: Zimbabwe's AIDS Trust Fund (a tax/levy-based instrument), Botswana's National HIV/AIDS Prevention Support (BNAPS) International Bank for Reconstruction and Development (IBRD) Buy-Down (a debt conversion instrument), and Côte d'Ivoire's Debt2Health Debt Swap Agreement (a debt conversion instrument). Zimbabwe's AIDS Trust Fund generated US$ 52.7 million between 2008 and 2011, Botswana's IBRD Buy-Down generated US$ 20 million, and Côte d'Ivoire's Debt2Health Debt Swap Agreement generated US$ 27 million, at least half of which was to be invested in HIV/AIDS programs. Four additional categories of innovative financing instruments met our criteria for future use: (1) remittances and diaspora bonds (2) social and development impact bonds (3) sovereign wealth funds (4) risk and credit guarantees.ConclusionA limited number of innovative financing instruments contributed a very modest share of funding toward domestic HIV/AIDS programs. Several innovative financing instruments successfully applied in other sectors could be used to augment domestic financing toward HIV/AIDS programmes.
| S-EPMC4871060 | biostudies-literature