Skip to content
Frontiers
Search

Review

Frontiers in Tropical Diseases ·

Volume 7 - 2026 | https://doi.org/10.3389/fitd.2026.1894145

Explainable machine learning for predicting treatment outcomes in female genital schistosomiasis and HIV co-morbidity: a narrative review and translational framework

  • David Chinaecherem Innocent1*corresponding author
  • Rejoicing Chijindum Innocent1
  • Increase Praise Innocent1
  1. Centicini Research Lab, Centicini Team Ltd, Abuja, Nigeria
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Female genital schistosomiasis (FGS) remains a neglected tropical disease with profound implications for women’s health, particularly in sub-Saharan Africa where it is highly co-endemic with HIV. The World Health Organization estimates that over 56 million women are at risk of FGS, with prevalence reaching up to 75% in high-burden lake regions. Epidemiological evidence demonstrates that FGS increases susceptibility to HIV acquisition through genital lesions, mucosal immune changes, and chronic inflammation, thereby amplifying the dual burden of morbidity. Despite large-scale praziquantel mass drug administration, reinfection is common, and integration with HIV prevention and care remains limited. This review explores the potential of explainable machine learning (ML) approaches to predict treatment outcomes and reinfection risk in women living with FGS–HIV co-morbidity, offering a framework for precision public health interventions. A narrative literature synthesis was undertaken across databases including PubMed, Scopus, and Web of Science, focusing on FGS epidemiology, treatment outcomes, HIV risk, and applications of ML in infectious disease modeling. Evidence was critically appraised and used to develop a conceptual ML framework. Current evidence highlights significant gaps in the integration of FGS and HIV care, limited application of ML to FGS, and persistent reinfection despite preventive chemotherapy. However, emerging ML studies in malaria, tuberculosis, trachoma, and HIV prediction illustrate the feasibility of interpretable models for infectious diseases. Explainable ML has the potential to enable patient-level risk prediction, strengthen clinical decision-making, and inform integrated FGS–HIV strategies, ultimately advancing precision interventions in high-burden settings.

Keywords

Introduction

Schistosomiasis, also known as snail fever or bilharzia, is a parasitic disease primarily caused by Schistosoma haematobium and Schistosoma mansoni in sub-Saharan Africa (SSA) regions characterized by inadequate sanitation and limited access to clean water (1). Female genital schistosomiasis (FGS) is a neglected manifestation of urogenital schistosomiasis, defined by the presence of S. haematobium eggs embedded in the female genital tract, including the vagina, cervix, uterus, fallopian tubes, or vulva (2). FGS affects an estimated 40 to 56 million women and girls in SSA, many of whom remain undiagnosed and untreated due to lack of awareness and diagnostic tools (3, 4). Region-specific studies have revealed prevalences of 57.5% in Zambia (2016) and 79.5% in Ghana (2013) among young females aged 11–20, detected via sensitive PCR methods, which outperform conventional diagnostics by up to 15-fold (5). Globally, schistosomiasis contributes to over 200,000 deaths per year in SSA and drives chronic morbidity including anemia, impaired growth in children, and diminished educational attainment, highlighting the enormous, yet underrecognized, public health burden (1).

Evidence suggests that female genital schistosomiasis increases susceptibility to HIV, a sexually transmitted infection characterized by chronic immune depletion through multiple biological pathways (2). FGS leads to mucosal lesions and “sandy patches” in the cervix and vagina, disrupts the integrity of the genital epithelial barrier, and triggers localized inflammation, all of which facilitate viral entry and replication (6). In some settings, women with FGS are three to four times more likely to be HIV-positive than those without FGS (7). The parasite also induces upregulation of the CCR5 receptor, a key HIV entry point on T cells and recruits vulnerable immune cells to genital tissues, intensifying the risk of HIV acquisition (8). Thus, the convergence of FGS and HIV represents a syndemic whereby the overlap of two diseases creates worse outcomes than either alone, demanding integrated intervention approaches (5, 7).

Despite the availability of praziquantel, a safe, inexpensive anthelmintic administered via mass drug administration (MDA), treatment, challenges persist due to high reinfection rates and fragmented health services (2). Praziquantel effectively kills adult worms, yet without addressing environmental exposure and reinfection risk, many women experience persistent or recurrent FGS, sometimes with irreversible reproductive tract damage (2, 9). Moreover, integration between schistosomiasis control and HIV care remains poor; FGS is seldom included in sexual and reproductive health programs, leading to missed opportunities for diagnosis and co-management (10). There is therefore an urgent need for innovative strategies to identify at-risk individuals, predict treatment outcomes, and deliver tailored interventions.

Recent advances in ML offer powerful tools for predictive modeling, risk stratification, and precision public health (11–13). ML can integrate multifactorial data—clinical, and behavioral, environmental—to forecast outcomes such as treatment success or reinfection in FGS–HIV co-morbid populations (14). However, evidence from the literature shows that many ML models function as opaque “black boxes”, limiting their acceptance in clinical and policy contexts. Explainable ML techniques such as SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-agnostic Explanations), and counterfactual reasoning lend interpretability by quantifying contributions of individual predictors and enabling transparent decision-making (15–20). These methods can help identify modifiable risk drivers, build clinician trust, and support ethical, actionable insights in resource-limited settings. This review aimed to provide insights based on existing evidence regarding FGS, HIV interaction, and ML applications to infectious diseases, and proposes an explainable ML framework for predicting treatment outcomes in FGS–HIV co-morbidity.

Methods

This study was conducted as a structured narrative review. A comprehensive literature search detailed in Appendix 1 was performed in PubMed, Scopus, Web of Science, and WHO IRIS to identify relevant studies published up to August 2025. Search terms included combinations of “female genital schistosomiasis”, “HIV”, “machine learning”, “artificial intelligence”, “explainable artificial intelligence”, “SHAP”, “LIME”, and “neglected tropical diseases” using Boolean operators (AND/OR). Eligible studies included original research, systematic reviews, clinical studies, and relevant policy documents addressing FGS, HIV co-morbidity, machine learning applications, and explainable AI in healthcare. Titles, abstracts, and full texts were screened for relevance, and key information on study characteristics, machine learning approaches, outcomes, and interpretability methods was extracted. Evidence was synthesized narratively to develop the proposed translational framework Appendix 2 shows the search flow chart for literature.

Epidemiology of female genital schistosomiasis and HIV co-morbidity

Report by WHO (1) define schistosomiasis as a parasitic disease caused by Schistosoma spp., with S. haematobium driving urogenital disease in sub-Saharan Africa (SSA), where almost 240 million people are infected and over 700 million live at risk; within this, another report (2) estimated ~56 million girls and women are affected by female genital schistosomiasis (FGS), a manifestation characterized by schistosome eggs lodged in the cervix, vagina, and vulva, producing “sandy patches,” contact bleeding, and chronic inflammation; it is further revealed that FGS is under-diagnosed and under-reported (1, 4). Tetteh (21) reported that ~11% of women of reproductive age in Ghana have FGS, another evidence shows PCR-based estimates reaching 57.5% in Zambia and 79.5% Ghana, highlighting how sensitive diagnostics reveal hidden burden (5); it is noted that >90% of people needing preventive chemotherapy for schistosomiasis reside in Africa (1). Figure 1 below illustrates a heat map of overlapping national FGS and HIV prevalence to visualize high-risk geographies.

Heat map of overlapping FGS and HIV prevalence by region **(A)** Prevalence of *S. haematobium* infection in school-aged children in Sub-Saharan Africa from 2000 onwards; **(B)** HIV prevalence in African adults (15-49); **(C)** African countries with published reports of FGS, FGS cases have been published by countries shaded in red; **(D)** Secondary infertility in Sub-Saharan Africa (adapted from 23).
Figure 1. Heat map of overlapping FGS and HIV prevalence by region (A) Prevalence of S. haematobium infection in school-aged children in Sub-Saharan Africa from 2000 onwards; (B) HIV prevalence in African adults (15-49); (C) African countries with published reports of FGS, FGS cases have been published by countries shaded in red; (D) Secondary infertility in Sub-Saharan Africa (adapted from 23).

FGS biologically amplifies HIV susceptibility via epithelial barrier disruption, mucosal lesions, and inflammatory cell recruitment (22). Sturt et al. (23) detail pathways including CCR5-expressing target cell influx that plausibly increases acquisition risk. Report by UNAIDS (2) synthesize epidemiology showing elevated HIV odds among women with FGS, and (Patel et al., 2020) conclude women with schistosomiasis are at higher HIV risk, consistent with Bochner et al. (24) who report sex-specific effects; critically, the Association of Schistosomiasis and HIV in Tanzania, quantify ~4-fold higher odds of HIV among women with female urogenital schistosomiasis in rural Tanzania, underscoring programmatic urgency (25). Also, the WHO (26) emphasize women of reproductive age as a priority risk group in endemic areas, justifying integrated FGS–HIV surveillance and care.

Current treatment modalities and outcomes in FGS

A report explained that praziquantel (PZQ) remains the cornerstone anthelminthic for schistosomiasis control, with preventive chemotherapy reaching only 75.3 million of the 251.4 million people who required treatment in 2021 evidence of substantial unmet need and residual morbidity in endemic settings (27, 28); clinically, PZQ is effective against adult worms but not eggs or immature stages, necessitating follow-up where exposure persists (1, 29). Fukushige et al. (30) report consistently high egg-reduction rates (ERR > 80%) even when cure rates (CR) fall below 50%, underscoring that “response” and “cure” are distinct endpoints in urogenital disease; importantly for FGS lesions, Kabengele et al. (31) show 60% lesion-severity reduction and 23% complete resolution 6–12 months after PZQ among Zambian women, while another evidence detect residual schistosome DNA after single-dose therapy, suggesting persistent tissue involvement that may sustain symptoms (25). Arenholt et al. (32), further find in a phase-2 randomized trial that repeated-dose PZQ did not outperform a single dose for cervicovaginal lesion resolution, whereas earlier schistosomiasis literature indicates repeat dosing can raise CR for urogenital infection generally, highlighting a disease-site–specific evidence gap in FGS (33). Aboagye et al. (34), quantify urinary-schistosomiasis reinfection between 8.1–39.6% within 8–28 weeks post-treatment, and Zacharia et al. (35) estimate a mean reinfection rate of 36.1%, arguing that chemotherapy alone without water, sanitation and hygiene (WASH) improvements cannot sustain gains.

Literature contends that programmatic fragmentation compounds clinical challenges: FGS care is rarely integrated into sexual and reproductive health (SRH) or HIV services, despite overlapping syndromes and care pathways; notably, PZQ supply and coverage shortfalls have historically hindered scale-up (1, 2, 36). Umbelino-Walker et al. (37) identify integration of FGS screening and treatment into HIV/SRH platforms as a “missed opportunity, “ also it outlines practical linkages with HIV and cervical-cancer prevention (e.g., co-located screening, shared counseling, syndromic algorithms augmented to include FGS). It has been posited that embedding PC/MDA within broader health-system strategies and updating guidelines to strengthen coverage, monitoring, and cross-program linkages (1, 38). Research also add that syndromic STI management often misclassifies FGS-related discharge or contact bleeding as STI, delaying definitive PZQ and perpetuating morbidity an argument for integrated algorithms and provider training (39, 40). Table 1 summarizes FGS treatment trials and reinfection outcomes to clarify effect sizes and residual risks.

Table 1. Summary of treatment trials in FGS and reinfection outcomes.

Study (authors, year)Population & settingTreatment regimenSample sizeOutcomes & results
Kabengele et al. (31)Adult women with FGS, Lusaka/Ndola, ZambiaSingle dose praziquantel (40 mg/kg)4360% reported lesion severity reduction; 23% achieved complete FGS lesion resolution after 6–12 months
Arenholt et al. (32), (Phase-2 RCT)Women with FGSSingle vs repeated dosing—Repeated-dose PZQ did not improve cervicovaginal lesion resolution compared to single dose
Nemungadi et al. (41), (Modeling)Girls in KwaZulu-Natal schools, South AfricaMDA with praziquantel>200, 000Regular PZQ MDA could prevent FGS in ~200, 000 young women and avert ~5, 000 HIV infections
Senghor et al. (42)School-age children, Niakhar, SenegalSingle dose praziquantel329Prevalence dropped from 73.2% to 4.6% post-treatment; reinfection at 2–3 months was 12.6%
Dolo et al. (43), (Systematic review)School-age children across SSASingle and double-dose PZQ—Double dose demonstrated higher efficacy and lower reinfection risk than single dose in high-transmission areas

Risk factors for poor outcomes in co-morbid populations

FGS which is an entrapment of Schistosoma haematobium eggs in the cervix, vagina or vulva that produces “sandy patches,” mucosal friability and contact bleeding; in co-morbid populations these lesion-severity features constitute biological risk factors for poor outcomes because they sustain genital inflammation and tissue damage that complicate HIV care and recovery (2). Sturt et al. (23) synthesize evidence that FGS drives local immune activation recruiting HIV-target cells and disrupting the epithelial barrier, while Hotez (22) emphasizes that these mucosal mechanisms plausibly heighten susceptibility and may perpetuate higher genital viral shedding, thereby worsening clinical trajectories in women living with HIV. Patel et al. (6) report higher odds of HIV among people with schistosomiasis, including an ecological estimate of a 2.9% relative rise in national HIV prevalence for each additional 100 individuals infected with Schistosoma, underlining that infection pressure and parasite exposure are population-level risk multipliers for adverse outcomes. Kleppa et al. (44), show no significant association between CD4 count and urogenital schistosomiasis egg intensity, indicating that systemic immunosuppression alone may not explain poor outcomes and that local mucosal pathology and parasite load in genital tissues are the proximate drivers of risk.

UNAIDS (2) note that sexual activity without barrier protection in high-prevalence settings increases HIV exposure and that FGS lesions (bleeding, ulceration) amplify acquisition risk during intercourse, reinforcing the need for combined prevention in co-morbid populations. Tchakoute et al. (45) find that each 10% increase in antiretroviral therapy (ART) adherence raises the odds of viral suppression by 37% and increases CD4 by 8.5 cells/µL at six months, implying that sub-optimal adherence and resulting viremia exacerbates FGS-related mucosal vulnerability and clinical failure. Furthermore, it was demonstrated that stigma, misdiagnosis of FGS as STIs, and limited provider knowledge delay definitive praziquantel therapy and integrated SRH/HIV services, entrenching morbidity in co-affected women (46, 47). Evidence document 8.1–39.6% urinary-schistosomiasis reinfection within 8–28 weeks post-treatment (34) and UNAIDS (2) link daily freshwater contact for chores to ongoing exposure, showing how structural water/sanitation barriers sustain poor outcomes despite therapy.

Machine learning applications in neglected tropical diseases and STIs

Recent evidence shows that ML defined as data-driven algorithms that learn patterns to make predictions has achieved strong diagnostic and prognostic performance across high-burden infections such as tuberculosis (TB), malaria, trachoma and HIV, with deep-learning image models and tabular learners increasingly used in low-resource settings (48–51). For TB, pooled meta-analytic estimates from chest-x-ray studies report mean AUCs around 0.92 with pooled sensitivity 0.986 and specificity 0.981, indicating screening-grade performance that could triage at scale when radiologists are scarce (52). For malaria, supervised learners using climate and entomological features predict incidence at sub-national level with high discrimination, while newer work integrates explainable AI (XAI) to expose feature importance e.g., XGBoost + SHAP supporting actionable surveillance and interpretation (53). In trachoma, computer-vision pipelines trained on 56, 725 eyelid images from 11, 358 Ethiopian children demonstrate robust TF grading and cost-efficient survey support, and transfer-learned, XAI-assisted classifiers have further improved classification stability for programmatic use (54, 55). In HIV, risk-prediction models for incident infection and linkage-to-care routinely achieve high discrimination (pooled AUC ≈ 0.91) and identify modifiable determinants to target prevention resources among adolescents and key populations in sub-Saharan Africa (56, 57).

Despite this momentum, application to FGS remains nascent: while digital microscopy and robotized image analysis have been prototyped for urogenital schistosomiasis diagnostics, there are few or no validated ML studies that (a) detect FGS lesions from clinical images, (b) predict FGS-HIV co-morbidity risk, or (c) forecast post-praziquantel outcomes and reinfection at the individual level (58, 59). This gap is consequential because explainable risk models could integrate clinical (lesion severity), behavioral (sexual exposures, ART adherence), and environmental (water contact) features to prioritize women for integrated FGS–SRH–HIV services, mirroring gains already realized in TB radiology and malaria early-warning systems (52). Accordingly, Table 2 illustrates the catalogue exemplar ML applications across NTDs/STIs (task, dataset, algorithm, performance e.g., AUC, sensitivity, specificity and make explicit the paucity of FGS-specific models, thereby motivating development of interpretable frameworks tailored to FGS–HIV co-morbidity.

Table 2. ML applications in NTDs and STIs.

Disease/taskDataset & settingAlgorithmPerformance (AUC/Acc/Sens/Spec)Source
TB detection (chest X-ray)Chest radiographs in low-resource settingsDeep learning modelsAUC ~0.92; Sensitivity ~0.986; Specificity ~0.981Hansun et al. (52)
Malaria incidence forecastingMulti-country climate dataSpatial–temporal climate modelsLead time 4 months; ROC metrics not reportedArmando et al. (60)
Malaria outbreak prediction (Gambia)District-level climate + epidemiologic dataXGBoost, decision tree, random forestAccuracy: XGBoost 93.3%; Random Forest 91.5%Khan et al. (61)
Malaria status in under-fives (Nigeria)2021 Nigeria Malaria Indicator SurveyRandom Forest, SVM, etc.RF best overall metrics not specifiedMthethwa & Melesse (62)
Explainable malaria diagnosis (Nigeria)Clinical data (337 patients)RF, XGBoost, CatBoost + XAI (SHAP/LIME)AUC: RF 0.869; CatBoost 0.787; XGBoost 0.770Awe et al. (63)
HIV risk prediction (adolescents/sub-Saharan Africa)Programmatic datasetsMultiple ML modelsPooled AUC ~0.91; range 0.67–1.00Mutai et al. (64)
HIV status prediction (Thailand, MSM)EMR, MSM 15–24 yearsXGB + SMOTEAccuracy 0.72; Sens 0.73; Spec 0.72; AUC 0.72Soha et al. (57)
LTFU in HIV care (Ethiopia)EMR of ART initiatesRandom ForestAccuracy 84.2%; Sens 82.4%; Spec 85.7%; AUC 89.5Endebu et al. (65)
VL suppression & visit attendance (SA)HIV program data (South Africa)Logistic regression, RF, AdaBoostPerformance metrics: sensitivity, specificity, AUC (specific values not given)Maskew et al. (66)
HIV outbreak forecasting (SSA)Demographic, behavioral, epidemiologic dataVarious ML modelsNot specified; models identify high-risk geographiesVidona (67)

Explainable machine learning approaches

Explainable machine learning (XAI) refers to methods that make model behavior understandable at the level of features or individual predictions (68, 69); core approaches include SHAP an additive, Shapley-values framework that unifies several attribution methods and is now widely adopted with >39, 000 citations and LIME, which fits a simple surrogate model locally around a prediction to reveal influential inputs (70, 71). In high-stakes domains, interpretable-by-design models such as decision trees and generalized additive models with pairwise interactions (GA2Ms) are argued to be preferable to opaque “black-box” models, with seminal healthcare case studies showing GA2Ms can match or exceed black-box accuracy while exposing clinically sensible risk factors for pneumonia and readmission (72, 73). For image-based tasks typical of field diagnostics, Grad-CAM produces class-discriminative heat maps that localize the regions driving a CNN’s decision, and human studies demonstrate these visual explanations help users assess model trustworthiness an essential property for clinical adoption in resource-limited settings (74). Accordingly, Figure 2 illustrates the model families (e.g., tree ensembles + SHAP; CNNs + Grad-CAM), datasets and performance metrics (AUC, sensitivity, specificity) from NTD/STI applications, underscoring where XAI already supports screening-grade performance (e.g., TB CXR models with AUC ≈ 0.92, sensitivity ≈ 0.99) yet remains under-used for FGS (52).

Workflow of explainable ML in predicting outcomes [Source: Sadeghi et al. (77)].
Figure 2. Workflow of explainable ML in predicting outcomes [Source: Sadeghi et al. (77)].

Causal machine learning targets patient-level effect heterogeneity, estimating how predicted outcomes change under alternative actions a critical step for turning explanations into actionable policy, triage, or treatment decisions (75, 76). In the FGS–HIV context, an explainable pipeline could: train an interpretable tree-ensemble to predict post-praziquantel outcomes and reinfection; compute SHAP values to quantify driver importance (e.g., lesion severity, water-contact proxies, ART adherence, partner concurrency); and overlay Grad-CAM for any cervicovaginal imaging to verify lesion-focused attention thereby exposing modifiable risk factors to clinicians and programs (70, 74). Because black-box models can mask spurious correlations, pairing transparent models (GA2Ms/trees) with local explanations (SHAP/LIME) and, where appropriate, causal forests to profile treatment benefit by subgroup strengthens validity, clinician trust, and equity in deployment (72, 73, 75). Figure 2 shows this workflow from data ingestion to model training, explanation, and policy targeting highlighting checkpoints (calibration, fairness screens) that are necessary for safe use in endemic FGS–HIV programs (77).

Proposed ML framework for predicting treatment outcomes in FGS–HIV

A pragmatic, explainable ML framework for FGS and HIV should draw on multi-source data that reflect both infection dynamics and care pathways, combining longitudinal clinical cohorts, national health-facility records, WHO/ESPEN surveillance, and population surveys such as the Demographic and Health Surveys (DHS) to ensure statistical power, external validity and subnational resolution (78–80). To justify prediction of clinically meaningful endpoints, the framework targets outcomes including praziquantel treatment success, reinfection and HIV progression, noting that pooled reinfection after therapy remains substantial overall 33.2% and 19.4% for urogenital schistosomiasis indicating the need to model post-treatment risk in exposed women (81). Predictors should be specified a priori across four domains demographic (age, education, wealth), clinical (lesion severity, prior FGS, CD4/viral load), behavioral (water contact, sexual exposure, ART adherence) and environmental (surface-water proximity, sanitation access) because FGS pathology, freshwater exposure and co-occurring HIV vulnerabilities jointly drive adverse outcomes in endemic African settings (1, 2, 38). Table 3 demonstrates these variables and their data provenance to support transparent model building and reproducibility (79, 80).

Table 3. Proposed variables for ML model (by data domain).

DomainVariable (proposed)Data source
DemographicAge, sex, educational attainment, wealth indexDHS (DHS Program)
Urban/rural residence, socioeconomic statusDHS; WHO/ESPEN surveillance
ClinicalLesion severity score, prior FGS history, CD4 count, viral loadClinical cohorts, hospital records
Praziquantel treatment timing, HIV treatment regimenHospital data, cohort studies
BehavioralWater contact frequency (e.g., daily river use), sexual behavior (e.g., condom use)DHS – water/sanitation indicators, behavioral modules; NTD studies
ART adherence, healthcare-seeking behaviorClinic data (possibly cohort surveys)
EnvironmentalDistance to water bodies, WASH infrastructure availabilityDHS geospatial data linked to cluster coordinates; WHO/ESPEN maps
Climate variables (rainfall, temperature)Gridded environmental data used in malaria risk mapping

Model selection should prioritize interpretable learners that retain state-of-the-art discrimination, such as gradient-boosting trees with SHAP attribution for global and patient-level feature importance, and intelligible generalized additive models with pairwise interactions (GA2Ms) for calibrated risk scores and policy rules (70, 72). Where image data are available (e.g., cervicovaginal photography), explainable neural pipelines can incorporate post-hoc visualizations alongside structured-data models to cross-validate lesion-centric signals, while XAI governance checks calibration, bias and stability should be embedded for safe deployment in high-stakes clinical decisions (77). Operationally, country programs can operationalize this framework by merging routine facility data with ESPEN surveillance and DHS covariates, training site-specific models that output risk tiers for targeted retreatment and integrated FGS–SRH–HIV services, with periodic re-training to address drift as MDA coverage and water-sanitation conditions change (1, 79, 80).

Discussion

Based on this review, current evidence establishes that FGS is a highly prevalent, under-diagnosed inflammatory condition affecting up to 56 million girls and women in sub-Saharan Africa and characterized by ova-driven genital lesions that plausibly increase susceptibility to HIV through epithelial disruption and mucosal immune activation (2, 22, 23). Empirical syntheses show a significant association between schistosomiasis and HIV, with a pooled odds ratio of approximately 2.3 for HIV among people with schistosomiasis and an ecological estimate of a 2.9% relative rise in national HIV prevalence per additional 100 S. haematobium infections, underscoring population-level risk amplification in co-endemic settings (6). Treatment-response data specific to FGS indicate partial lesion resolution after praziquantel 60% reduced severity and 23% complete resolution at 6–12 months in Zambian adults while a phase-2 trial found no superiority of repeated over single dosing for cervicovaginal lesion outcomes, suggesting residual pathology and persistent exposure drive ongoing risk despite chemotherapy (31, 32). Meta-analytic estimates of reinfection after treatment remain substantial, reinforcing the need to integrate FGS services with HIV and sexual-reproductive health platforms and environmental prevention, rather than relying on mass drug administration alone (2, 81). Against this clinical and epidemiologic backdrop, explainable ML can bridge evidence gaps by modeling individualized risk trajectories and triaging women for combined FGS–HIV care along the lines detailed in Table 2.

In addition, comparative experience across infectious diseases shows that ML already delivers screening-grade performance and operational insights when high-quality data are available, as illustrated by deep-learning chest-x-ray tools for tuberculosis that reach high discrimination and require context-specific thresholds for different screening scenarios (82–85). In malaria, gradient-boosted models augmented with explainable AI have predicted incidence and mortality with strong accuracy while revealing the relative importance of climatic and environmental drivers, supporting spatially targeted interventions (53). In trachoma, large image datasets have enabled robust computer-vision grading pipelines that both scale surveys and provide visual localization for clinician review an approach translatable to cervicovaginal imaging of FGS lesions where available (54, 55). In HIV, supervised learners have identified high-risk subgroups and predicted loss-to-follow-up or unsuppressed viral load, demonstrating value for patient-level decision-support in resource-limited programs (57, 65). Collectively, these results argue that FGS–HIV programs can benefit from interpretable models that ingest routine clinical records, surveillance and survey covariates to prioritize praziquantel retreatment, intensify adherence support, and direct WASH or risk-reduction resources to the highest-yield places and people (79, 80). Yet the literature remains sparse for FGS-specific ML, creating an opportunity to adapt proven designs from TB, malaria and trachoma to genital schistosomiasis while coupling explanations to programmatic actions (23).

An explainable framework should combine interpretable gradient boosting with SHAP attribution, intrinsically intelligible generalized additive models with pairwise interactions, and causal forests to estimate heterogeneous treatment effects, ensuring both prediction and actionability for endpoints such as lesion resolution, reinfection and HIV progression (70, 72, 75). This approach aligns with the theory of precision public health placing the right intervention in the right population at the right time—and with the One Health lens that links environmental water exposure, human behavior and pathogen dynamics in schistosomiasis-HIV co-endemic ecologies (1, 86). Critical limitations include fragmented data systems, variable digital infrastructure and algorithmic bias, which together can erode generalizability unless models are transparently validated, calibrated and audited across settings using open reporting and threshold optimization adapted to local prevalence (73, 82, 87–91). Future priorities include building regional ML repositories that federate ESPEN and health-facility data, formal partnerships with Ministries of Health to co-design use-cases and governance, external validation across countries, and incorporation into national NTD–HIV strategies with retraining schedules that track shifts in mass drug administration coverage and water-sanitation conditions (1, 79, 80).

Implications

The implications of this review are multifaceted, spanning policy, practice, and research, and highlight the transformative potential of explainable ML in addressing the dual burden of FGS and HIV (11). From a policy perspective, the integration of FGS screening within existing HIV programs is critical, particularly in sub-Saharan Africa where co-endemicity is high; in line with wide literature, leveraging ML-driven risk stratification can ensure that high-risk women are identified earlier and interventions are better targeted (92–95). In clinical practice, the use of explainable ML dashboards offers frontline health workers patient-level interpretability, thereby reducing diagnostic uncertainty and enabling timely, informed decisions even in resource-constrained settings (96–101). Such approaches also empower non-specialist providers to recognize modifiable risk factors and deliver more person-centered care. From a research standpoint, this review underscores the urgent need for multi-country datasets that reflect diverse populations, as well as federated learning frameworks that allow data pooling across borders while safeguarding patient privacy.

Conclusions

In conclusion, this review has illustrated the substantial burden of FGS and its strong epidemiological and biological links with heightened HIV acquisition risk in sub-Saharan Africa. Persistent reinfection, limited integration with sexual and reproductive health services, and weak surveillance systems, despite ongoing mass drug administration with praziquantel, continue to hinder progress. Against this backdrop, machine learning (ML) offers a transformative pathway for risk stratification, outcome prediction, and the design of integrated care models that respond to both infections concurrently. Unlike conventional statistical tools, explainable ML approaches such as SHAP, LIME, and causal models have the potential to provide patient-level interpretability, strengthen clinical decision-making, and enhance trust among healthcare workers in resource-limited contexts. By drawing on diverse data streams, including clinical, behavioral, demographic, and environmental indicators, these models can help identify modifiable risk factors and optimize treatment strategies. The novelty of this review lies in proposing an explainable ML framework tailored for FGS–HIV co-morbidity, highlighting both opportunities and barriers to implementation. We urgently need regional data-sharing consortia, federated learning initiatives, and pilot implementation projects embedded within national NTD-HIV strategies to fully realize this potential. Such steps would foster precision public health approaches, bridge gaps in care, and advance global goals for eliminating NTDs while reducing HIV transmission.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Roch Christian Johnson, CIFRED UAC, Benin

Reviewed by: Dina S. Nasr, Faculty of Medicine Cairo University, Egypt

Ncomeka Sineke, Walter Sisulu University, South Africa

Footnotes

AI, Artificial Intelligence; ART, Antiretroviral Therapy; AUC, Area Under the Curve; CCR5, C-C Chemokine Receptor Type 5; CD4, Cluster of Differentiation 4; CNN, Convolutional Neural Network; CR, Cure Rate; DHS, Demographic and Health Surveys; EMR, Electronic Medical Records; ERR, Egg Reduction Rate; ESPEN, Expanded Special Project for Elimination of Neglected Tropical Diseases; FGS, Female Genital Schistosomiasis; GA2M, Generalized Additive Model with Pairwise Interactions; Grad-CAM, Gradient-weighted Class Activation Mapping; HIV, Human Immunodeficiency Virus; LIME, Local Interpretable Model-Agnostic Explanations; LMICs, Low- and Middle-Income Countries; LTFU, Loss to Follow-Up; MDA, Mass Drug Administration; ML, Machine Learning; MSM, Men Who Have Sex with Men; NTDs, Neglected Tropical Diseases; PCR, Polymerase Chain Reaction; PZQ, Praziquantel; RF, Random Forest; ROC, Receiver Operating Characteristic; SHAP, SHapley Additive exPlanations; SMOTE, Synthetic Minority Over-sampling Technique; SRH, Sexual and Reproductive Health; SSA, Sub-Saharan Africa; STI, Sexually Transmitted Infection; SVM, Support Vector Machine; TB, Tuberculosis; VL, Viral Load; WASH, Water, Sanitation, and Hygiene; WHO, World Health Organization; XAI, Explainable Artificial Intelligence; XGBoost (XGB), Extreme Gradient Boosting.

Statements

Conflict of interest

Authors DI, RI and II were employed by Centicini Team Ltd.

Funding

The author(s) declared that financial support was not received for this work and/or its publication.

Author contributions

DI: Investigation, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. RI: Investigation, Methodology, Project administration, Supervision, Writing – review & editing. II: Data curation, Formal analysis, Writing – review & editing.

Acknowledgements

We want to specially acknowledge the entire team at Centicini Research Lab for their resilient efforts in making research communication an engaging and productive one. Also, we acknowledge the efforts of Prof U.M Chukwuocha and Prof I.N.S Dozie for their academic mentorship to this research lab.

References

  1. 1 World Health Organization. Schistosomiasis: fact sheet (2023). Geneva: WHO. Available online at: https://www.who.int/news-room/fact-sheets/detail/schistosomiasis (Accessed August 19, 2025).
  2. 2 UNAIDS. No more neglect — Female genital schistosomiasis and HIV (2019). Geneva: UNAIDS. Available online at: https://www.unaids.org/sites/default/files/media_asset/female_genital_schistosomiasis_and_hiv_en.pdf (Accessed August 19, 2025).
  3. 3 Mberu M, Zongo K, Leaning E, Makau-Barasa L, Kamara K, Jacobson J. Female genital schistosomiasis: addressing diagnostic and programmatic gaps to advance elimination efforts. Int J Infect Dis. (2025) 152: 107798. doi: doi:10.1016/j.ijid.2025.107798
  4. 4 Ursini T, Scarso S, Mugassa S, Othman JB, Yussuph AJ, Ndaboine E. Assessing the prevalence of Female Genital Schistosomiasis and comparing the acceptability and performance of health worker-collected and self-collected cervical-vaginal swabs using PCR testing among women in North-Western Tanzania: The ShWAB study. PLoS NeglTrop Dis. (2023) 17: e0011465. doi: doi:10.1371/journal.pntd.0011465
  5. 5 Kaur N, Buss L, Zorn L, Mwansa J, Mutengo MM, Lodh N. Female Genital Schistosomiasis (FGS) prevalence and burden across endemic countries, timelines, and age groups: a retrospective study. Parasitologia. (2024) 4: 382–9. doi: doi:10.3390/parasitologia4040034
  6. 6 Furch BD, Koethe JR, Kayamba V, Heimburger DC, Kelly P. Interactions of Schistosoma and HIV in Sub-Saharan Africa: a systematic review. Am J Trop Med Hyg. (2020) 102 (4): 711–8. doi: doi:10.4269/ajtmh.19-0494
  7. 7 UNAIDS/WHO. No more neglect—Female genital schistosomiasis and HIV (2019). Available online at: https://www.unaids.org/sites/default/files/media_asset/female_genital_schistosomiasis_and_hiv_en.pdf. FGS care integration call (Accessed August 19, 2025).
  8. 8 Saha A, Chakraborty T, Rahimikollu J, Xiao H, de Oliveira LB, Hand TW. Deep humoral profiling coupled to interpretable machine learning unveils diagnostic markers and pathophysiology of schistosomiasis. Sci Transl Med. (2024) 16: eadk7832. doi: doi:10.1126/scitranslmed.adk7832
  9. 9 World Health organization. Control of neglected tropical diseases. Available online at: https://www.who.int/teams/control-of-neglected-tropical-diseases/overview (Accessed August 19, 2025).
  10. 10 Franc KA, Krecek RC, Häsler BN, Arenas-Gamboa AM. Brucellosis remains a neglected disease in the developing world: a call for interdisciplinary action. BMC Public Health. (2018) 18: 125. doi: doi:10.1186/s12889-017-5016-y
  11. 11 Dhanda SS, Panwar D, Lin CC, Sharma TK, Rastogi D, Bindewari S. Advancement in public health through machine learning: a narrative review of opportunities and ethical considerations. J Big Data. (2025) 12: 1 – 58. doi: doi:10.1186/s40537-025-01201-x
  12. 12 Olalekan Kehinde A. Leveraging machine learning for predictive models in healthcare to enhance patient outcome management. Int Res J Mod Eng Technol Sci. (2025) 7: 1465. doi: doi:10.56726/IRJMETS66198
  13. 13 Islam MA, Yeasmin S, Hosen A, Vanu N, Riipa MB, Tasnim AF. Harnessing predictive analytics: The role of machine learning in early disease detection and healthcare optimization. J Ecohumanism. (2025) 4: 312–21. doi: doi:10.2139/ssrn.5411042
  14. 14 Pennisi F, Pinto A, Ricciardi GE, Signorelli C, Gianfredi V. The role of artificial intelligence and machine learning models in antimicrobial stewardship in public health: a narrative review. Antibiotics. (2025) 14: 134. doi: doi:10.3390/antibiotics14020134
  15. 15 Kalasampath K, Spoorthi KN, Sajeev S, Kuppa SS, Ajay K, Maruthamuthu A. A literature review on applications of explainable artificial intelligence (XAI). IEEE Access. (2025) 13: 41111–40. doi: doi:10.1109/access.2025.3546681
  16. 16 Tiwari R. Explainable AI (XAI) and its applications in building trust and understanding in AI decision making. Int J Sci Res Eng Manag. (2023) 7: 1 – 3. doi: doi:10.55041/ijsrem17592
  17. 17 de Brito Duarte R, Correia F, Arriaga P, Paiva A. AI trust: Can explainable AI enhance warranted trust ? Hum Behav Emerging Technol. (2023) 2023: 4637678. doi: doi:10.1155/2023/4637678
  18. 18 Hassija V, Chamola V, Mahapatra A, Singal A, Goel D, Huang K. Interpreting black-box models: a review on explainable artificial intelligence. Cognit Comput. (2024) 16: 45 – 74. doi: doi:10.1007/s12559-023-10179-8
  19. 19 Rane N, Choudhary S, Rane J. Explainable artificial intelligence (XAI) in healthcare: interpretable models for clinical decision support. (2023). doi: doi:10.2139/ssrn.4637897
  20. 20 Houssein EH, Gamal AM, Younis EM, Mohamed E. Explainable artificial intelligence for medical imaging systems using deep learning: a comprehensive review. Cluster Comput. (2025) 28: 469. doi: doi:10.1007/s10586-025-05281-5
  21. 21 Tetteh CD, Manyeh AK, Ncayiyana JR, Ginindza TG. Female genital schistosomiasis in Ghana: an exploration of knowledge, attitudes, and practice among women of reproductive age. Public Health Pract (Oxf). (2025) 10: 100632. doi: doi:10.1016/j.puhip.2025.100632
  22. 22 Hotez PJ, Engels D, Fenwick A, Savioli L. Africa is desperate for praziquantel. Lancet. (2010) 376 (9740): 496–98. doi: doi:10.1016/S0140-6736(10)60879-3
  23. 23 Sturt AS, Webb EL, Phiri CR, Mudenda M, Mapani J, Kosloff B. Female genital schistosomiasis and HIV-1 incidence in Zambian women: a retrospective cohort study. Open Forum Infect Dis. (2021) 8 (7): ofab349. doi: doi:10.1093/ofid/ofab349
  24. 24 Bochner AF, Baeten JM, Secor WE, van Dam GJ, Szpiro AA, Njenga SM. Associations between schistosomiasis and HIV-1 acquisition risk in four prospective cohorts: a nested case–control analysis. J Int AIDS Soc. (2020) 23: e25534. doi: doi:10.1002/jia2.25534
  25. 25 Downs JA, van Dam GJ, Changalucha JM, Corstjens PL, Peck RN, de Dood CJ. Association of schistosomiasis and HIV infection in Tanzania. Am J Trop Med Hyg. (2012) 87: 868–73. doi: doi:10.4269/ajtmh.2012.12-0395
  26. 26 World Health Organization Regional Office for Africa (ESPEN). Female genital schistosomiasis: a silent threat to women’s health and gender equality (2024). Available online at: https://espen.afro.who.int/female-genital-schistosomiasis-silent-threat-womens-health-and-gender-equality (Accessed August 19, 2025).
  27. 27 Alibrahim AO, Elkholy WA, El‐Derbawy MM, Zahran NF, Alexiou A, Papadakis M. Schistosomiasis chemotherapy, chemoprevention, and vaccines: history, progress, and priorities. Immun Inflammation Dis. (2024) 12: e70054. doi: doi:10.1002/iid3.70054
  28. 28 Summers S, Bhattacharyya T, Allan F, Stothard JR, Edielu A, Webster BL. A review of the genetic determinants of praziquantel resistance in Schistosoma mansoni: Is praziquantel and intestinal schistosomiasis a perfect match ? FrontTrop Dis. (2022) 3: 933097. doi: doi:10.3389/fitd.2022.933097
  29. 29 Rossi B, Previtali L, Salvi M, Gerami R, Tomasoni LR, Quiros-Roldan E. Female genital schistosomiasis: a neglected among the neglected tropical diseases. Microorganisms. (2024) 12: 458. doi: doi:10.3390/microorganisms12030458
  30. 30 Fukushige M, Chase-Topping M, Woolhouse ME, Mutapi F. Efficacy of praziquantel has been maintained over four decades (from 1977 to 2018): A systematic review and meta-analysis of factors influence its efficacy. PLoS NeglTrop Dis. (2021) 15: e0009189. doi: doi:10.1371/journal.pntd.0009189
  31. 31 Kabengele C, Mwangelwa S, Kilembe W, Vwalika B, Inambao M, Moonga V. Female genital schistosomiasis lesion resolution post-treatment with praziquantel in Zambian adults. Am J Trop Med Hyg. (2024) 110 (2): 250–3. doi: doi:10.4269/ajtmh.23-0552
  32. 32 Arenholt LTS, Randrianasolo BS, Rabozakandraina TOO, Ramarokoto CE, Jøker K, Kæstel Aarøe K. Repeated versus single praziquantel dosing regimen in treatment of female genital schistosomiasis: a phase 2 randomised controlled trial showing no difference in efficacy. Front Trop Dis. (2024) 5: 1322652. doi: doi:10.3389/fitd.2024.1322652
  33. 33 King CH, Olbrych SK, Soon M, Singer ME, Carter J, Colley DG. Utility of repeated praziquantel dosing in the treatment of schistosomiasis in high-risk communities in Africa: a systematic review. PLoS Negl Trop Dis. (2011) 5 (9): e1321. doi: doi:10.1371/journal.pntd.0001321
  34. 34 Aboagye IF, Addison YAA. Praziquantel efficacy, urinary and intestinal schistosomiasis reinfection: a systematic review. Pathog Glob Health. (2023) 117 (7): 623–30. doi: doi:10.1080/20477724.2022.2145070
  35. 35 Zacharia A, Mushi V, Makene T. A systematic review and meta-analysis on the rate of human schistosomiasis reinfection. PLoS One. (2020) 15 (12): e0243224. doi: doi:10.1371/journal.pone.0243224
  36. 36 World Health Organization. Expanding preventive chemotherapy to all in need. Geneva: WHO. Available online at: https://www.who.int/activities/expanding-preventive-chemotherapy-to-all-in-need (Accessed August 19, 2025).
  37. 37 Umbelino-Walker I, Wong F, Cassolato M, Pantelias A, Jacobson J, Kalume C. Integration of female genital schistosomiasis into HIV/sexual and reproductive health and rights and neglected tropical diseases programmes and services: a scoping review. Sex Reprod Health Matters. (2023) 31 (1): 2262882. doi: doi:10.1080/26410397.2023.2262882
  38. 38 Lo NC, Bezerra FSM, Colley DG, Fleming FM, Homeida M, Kabatereine N. Review of 2022 WHO guidelines on the control and elimination of schistosomiasis. Lancet Infect Dis. (2022) 22 (11): e327–e335. doi: doi:10.1016/S1473-3099(22)00221-3
  39. 39 Søfteland S, Sebitloane M, Galappaththi-Arachchige HN, Kleppa E, Holmen S, Pillay P. Diagnosis of female genital schistosomiasis and other genital infections in young South African women: challenges in the syndromic approach. Front Trop Dis. (2024) 5: 1379526. doi: doi:10.3389/fitd.2024.1379526
  40. 40 Mazigo HD, Samson A, Lambert VJ, Kosia AL, Ngoma DD, Murphy R. Female genital schistosomiasis is a sexually transmitted disease”: Gaps in healthcare workers’ knowledge about female genital schistosomiasis in Tanzania. PLoS Glob Public Health. (2022) 2: e0000059. doi: doi:10.1371/journal.pgph.0000059
  41. 41 Nemungadi TG, Kleppa E, Galappaththi-Arachchige HN, Pillay P, Gundersen SG, Vennervald BJ. Predictors for participation in mass-treatment and female genital schistosomiasis re-investigation, and the effect of praziquantel treatment in South African adolescents. PLoS Negl Trop Dis. (2024) 18 (3): e0011798. doi: doi:10.1371/journal.pntd.0011798
  42. 42 Senghor B, Diaw OT, Doucoure S, Sylla SN, Seye M, Talla I. Efficacy of praziquantel against urinary schistosomiasis and reinfection in Senegalese school children where there is a single well-defined transmission period. Parasit Vectors. (2015) 8: 362. doi: doi:10.1186/s13071-015-0980-5
  43. 43 Dolo M, Maiga H, Coulibaly YI, Dolo H, Sangaré M, Maiga O. The efficacy of praziquantel in school children infected with urogenital schistosomiasis in sub-Saharan Africa: a systematic review from 2005 to 2020. Afr J Parasitol Mycol Entomol. (2024) 2 (1): 8. doi: doi:10.35995/ajpme02010008
  44. 44 Kleppa E, Klinge KF, Galaphaththi-Arachchige HN, Holmen SD, Lillebø K, Onsrud M. Schistosoma haematobium infection and CD4+ T-cell levels: a cross-sectional study of young South African women. PLoS One. (2015) 10 (3): e0119326. doi: doi:10.1371/journal.pone.0119326
  45. 45 Tchakoute CT, Rhee SY, Hare CB, Shafer RW, Sainani K. Adherence to contemporary antiretroviral treatment regimens and impact on immunological and virologic outcomes in a US healthcare system. PLoS One. (2022) 17 (2): e0263742. doi: doi:10.1371/journal.pone.0263742
  46. 46 Jacobson J, Pantelias A, Williamson M, Kjetland EF, Krentel A, Gyapong M. Addressing a silent and neglected scourge in sexual and reproductive health in Sub-Saharan Africa by development of training competencies to improve prevention, diagnosis, and treatment of female genital schistosomiasis for health workers. Reprod Health. (2022) 19 (1): 20. doi: doi:10.1186/s12978-021-01252-2
  47. 47 Sommerfelt I, Ndhlovu P, Taylor M, Naidoo S, Pillay P, Haaland H. Health professionals' knowledge about female genital schistosomiasis: a qualitative investigation in a schistosomiasis endemic area in South Africa. SSM Qual Res Health. (2023) 3: 100292. doi: doi:10.1016/j.ssmqr.2023.100292
  48. 48 Siyamayambo C, Phalane E, Phaswana-Mafuya RN. Diagnosis and management of sexually transmitted infections using artificial intelligence applications among key and general populations in Sub-Saharan Africa: A systematic review and meta-analysis. Algorithms. (2025) 18: 151. doi: doi:10.3390/a18030151
  49. 49 Zhu J, Noble JA, Gomes M. Artificial intelligence image-diagnosis for female genital schistosomiasis. Mayo Clinic Proceedings: Digital Health. (2025), 100245. doi: doi:10.1016/j.mcpdig.2025.100245
  50. 50 Wasson PS, Williamson MK, Hallas-Potts A, Lopez-Jimena B, Koscielniak M, Tait S. Female Genital Schistosomiasis: Translational Challenges and Opportunities”: Outputs and actions from a consultative, collaborative and translational workshop. In: Parasitology. p. 1 – 44. doi: doi:10.1017/s0031182025100346
  51. 51 Shija J. A machine learning model for early detection of sexually transmitted infections. Comput Ecol Software. (2025) 15: 30. doi: doi:10.58694/20.500.123456789/3395
  52. 52 Hansun S, Argha A, Liaw ST, Celler BG, Marks GB. Machine and deep learning for tuberculosis detection on chest X-rays: systematic literature review. J Med Internet Res. (2023) 25: e43154. doi: doi:10.2196/43154
  53. 53 Ndlovu N, Ndlovu B. Explainable artificial intelligence in multimodal malaria prediction: a systematic review and roadmap integrating climate change, parasite genomics, and public health decision support. J Appl Inform Comput. (2026) 10 (2): 1263–75. doi: doi:10.30871/jaic.v10i2.12347
  54. 54 Joye AS, Firlie MG, Wittberg DM, Aragie S, Nash SD, Tadesse Z. Computer vision identification of trachomatous inflammation-follicular using deep learning. Cornea. (2024) 44 (5): 613–8. doi: doi:10.1097/ICO.0000000000003701
  55. 55 Pan Y, Lan W, Xu B. Active trachoma: enhancing image classification using pretrained SOTA models and explainable AI. Front Bacteriol. (2024) 3: 1333641. doi: doi:10.3389/fbrio.2024.1333641
  56. 56 Latt PM, Soe NN, Fairley CK, Chow EPF, Johnson CC, Shah P. Machine learning for personalized risk assessment of HIV, syphilis, gonorrhoea and chlamydia: a systematic review and meta-analysis. Int J Infect Dis. (2025) 157: 107922. doi: doi:10.1016/j.ijid.2025.107922
  57. 57 Mutai CK, McSharry PE, Ngaruye I, Musabanganji E. Use of machine learning techniques to identify HIV predictors for screening in sub-Saharan Africa. BMC Med Res Methodol. (2021) 21 (1): 159. doi: doi:10.1186/s12874-021-01346-2
  58. 58 Rubio Maturana C, Dantas de Oliveira A, Zarzuela F, Ruiz E, Sulleiro Igual E, Mediavilla A. Development of an automated artificial intelligence-based system for urogenital schistosomiasis diagnosis using digital image analysis techniques and a robotized microscope. PLoS NeglTrop Dis. (2024) 18: e0012614. doi: doi:10.1371/journal.pntd.0012614
  59. 59 Zhu J, Noble JA, Gomes M. Artificial intelligence image-diagnosis for female genital schistosomiasis. Mayo Clin Proc Digit Health. (2025) 3 (3): 100245. doi: doi:10.1016/j.mcpdig.2025.100245
  60. 60 Armando CJ, Rocklöv J, Sidat M, Tozan Y, Mavume AF, Sewe MO. Spatiotemporal modelling and prediction of malaria incidence in Mozambique using climatic indicators from 2001 to 2018. Sci Rep. (2025) 15 (1): 11971. doi: doi:10.1038/s41598-025-97072-6
  61. 61 Khan O. Predicting malaria outbreak in The Gambia using machine learning techniques. PLoS One. (2024) 19: e0299386. Available online at: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0299386 (Accessed August 06, 2026).
  62. 62 Mthethwa SM, Melesse DN. Predicting malaria status in Nigerian children under five using 2021 Malaria Indicator Survey data: a machine learning comparison study. Open Public Health J. (2025) 18: 85 – 95. Available online at: https://openpublichealthjournal.com/VOLUME/18/ELOCATOR/e18749445396163/FULLTEXT/.
  63. 63 Awe OO, Mwangi PN, Goudoungou SK, Esho RV, Oyejide OS. Explainable AI for enhanced accuracy in malaria diagnosis using ensemble machine learning models. BMC Med Inform Decis Mak. (2025) 25: 162. Available online at: https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/s12911-025-02874-3 (Accessed August 06, 2026).
  64. 64 Mutai CK, McSharry PE, Ngaruye I, Musabanganji E. Use of machine learning techniques to identify HIV predictors for screening in sub-Saharan Africa. BMC Med Res Methodol. (2021) 21: 159. doi: doi:10.1186/s12874-021-01346-2
  65. 65 Endebu T, Taye G, Deressa W. Development of a machine learning prediction model for loss to follow-up in HIV care using routine electronic medical records in a low-resource setting. BMC Med Inform Decis Mak. (2025) 25 (1): 192. doi: doi:10.1186/s12911-025-03030-7
  66. 66 Maskew M, Sharpey-Schafer K, De Voux L, Crompton T, Bor J, Rennick M. Applying machine learning and predictive modeling to retention and viral suppression in South African HIV treatment cohorts. Sci Rep. (2022) 12: 12715. doi: doi:10.1038/s41598-022-16062-0
  67. 67 Vidona WB, Willy-Vidona C, David LK, Abia BM, Paul JN, Eze UA. Assessment of the demographic burden of epidemics and diseases within the African sub-region: A review of recent epidemiological data, responses, and resultant psychosocial risk tendencies. Hygiene. (2024) 4: 346–60. doi: doi:10.3390/hygiene4030027
  68. 68 Linardatos P, Papastefanopoulos V, Kotsiantis S. Explainable ai: A review of machine learning interpretability methods. Entropy. (2020) 23: 18. doi: doi:10.3390/e23010018
  69. 69 Minh D, Wang HX, Li YF, Nguyen TN. Explainable artificial intelligence: a comprehensive review. Artif Intell Rev. (2022) 55: 3503–68. doi: doi:10.1007/s10462-021-10088-y
  70. 70 Lundberg SM, Lee SI. A unified approach to interpreting model predictions. Adv Neural Inf Process Syst. (2017) 30: 4765–74. Available online at: https://arxiv.org/abs/1705.07874 (Accessed August 06, 2026).
  71. 71 Ribeiro MT, Singh S, Guestrin C. Why should I trust you?": Explaining the predictions of any classifier. arXiv. (2016). Available online at: https://arxiv.org/abs/1602.04938 (Accessed August 06, 2026).
  72. 72 Caruana R, Lou Y, Gehrke J, Koch P, Sturm M, Elhadad N. (2015). “ Intelligible models for healthcare: Predicting pneumonia risk and hospital 30-day readmission ”, in: Proc 21st ACM SIGKDD Int Conf Knowl Discov Data Min, 1721–30. doi: doi:10.1145/2783258.2788613
  73. 73 Rudin C. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat Mach Intell. (2019) 1: 206–15. doi: doi:10.1038/s42256-019-0048-x
  74. 74 Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. (2017). “ Grad-CAM: Visual explanations from deep networks via gradient-based localization ”, in: ICCV Proceedings. Available online at: https://arxiv.org/abs/1610.02391 (Accessed August 06, 2026).
  75. 75 Wager S, Athey S. Estimation and inference of heterogeneous treatment effects using random forests. J Am Stat Assoc. (2018) 113: 1228–42. Available online at: https://www.tandfonline.com/doi/abs/10.1080/01621459.2017.1319839 (Accessed August 06, 2026).
  76. 76 Athey S, Wager S. Estimating treatment effects with causal forests. arXiv. (2019). Available online at: https://arxiv.org/abs/1902.07409 (Accessed August 06, 2026).
  77. 77 Srivastav A, Singh A, Rivers B, Lillard J, Singh R. Revolutionizing oncology through AI: addressing cancer disparities by improving screening, treatment, and survival outcomes via integration of social determinants of health. Cancers (Basel). (2025) 17: 2866. doi: doi:10.3390/cancers17172866
  78. 78 WHO. Increasing awareness on genital manifestations of schistosomiasis. Geneva: World Health Organization. (2026). Available online at: https://www.who.int/activities/increasing-awareness-on-genital-manifestations-of-schistosomiasis (Accessed August 06, 2026).
  79. 79 ESPEN (Expanded Special Project for Elimination of NTDs). Female Genital Schistosomiasis: A Silent Threat to Women’s Health and Gender Equality (2024). WHO Regional Office for Africa. Available online at: https://espen.afro.who.int/female-genital-schistosomiasis-silent-threat-womens-health-and-gender-equality (Accessed August 19, 2025).
  80. 80 The DHS Program. Demographic and Health Surveys Datasets. Available online at: https://dhsprogram.com/?utm_source=chatgpt.com (Accessed August 19, 2025).
  81. 81 Zacharia A, Mushi V, Makene T. A systematic review and meta-analysis on the rate of human schistosomiasis reinfection. PLoS One. (2020) 15: e0243224. doi: doi:10.1371/journal.pone.0243224
  82. 82 Hansun S, Argha A, Liaw ST, Celler BG, Marks GB. Machine and deep learning for tuberculosis detection on chest X-rays: systematic literature review. J Med Internet Res. (2023) 25: e43154. doi: doi:10.2196/43154
  83. 83 Zhang Q, Chen P, Li J, Feng L, Liu S, Zhao H. Pediabench: A comprehensive chinese pediatric dataset for benchmarking large language models. arXiv preprint arXiv:2412.06287. (2024). doi: doi:10.48550/arXiv.2412.06287
  84. 84 Chiu HY, Hwang CK, Chen SY, Shih FY, Han HC, King CC. Machine learning for emerging infectious disease field responses. Sci Rep. (2022) 12: 328. doi: doi:10.1038/s41598-021-03687-w
  85. 85 Al Meslamani AZ, Sobrino I, de la Fuente J. Machine learning in infectious diseases: potential applications and limitations. Ann Med. (2024) 56: 2362869. doi: doi:10.1080/07853890.2024.2362869
  86. 86 Khoury MJ, Iademarco MF, Riley WT. Precision public health for the era of precision medicine. Am J Prev Med. (2016) 50 (3): 398–401. doi: doi:10.1016/j.amepre.2015.08.031
  87. 87 Narasimhan M, Yeh PT, Haberlen S, Warren CE, Kennedy CE. Integration of HIV testing services into family planning services: a systematic review. Reprod Health. (2019) 16: 61. doi: doi:10.1186/s12978-019-0714-9
  88. 88 Pillay LN, Umbelino-Walker I, Schlosser D, Kalume C, Karuga R. Minimum service package for the integration of female genital schistosomiasis into sexual and reproductive health and rights interventions. FrontTrop Dis. (2024) 5: 1321069. doi: doi:10.3389/fitd.2024.1321069
  89. 89 Church K, Mayhew SH. Integration of STI and HIV prevention, care, and treatment into family planning services: a review of the literature. Stud Family Plann. (2009) 40: 171–86. doi: doi:10.1111/j.1728-4465.2009.00201.x
  90. 90 Karuga RN, Ouma M, Mulupi S, Pillay L, Pensotti C, Gamba V. Acceptability and feasibility of integrating female genital schistosomiasis and sexual and reproductive health interventions in Kenya: a demonstration study. PLOS Glob Public Health. (2025) 5 (9): e0004938. doi: doi:10.1371/journal.pgph.0004938
  91. 91 Desai KT, Befano B, Xue Z, Kelly H, Campos NG, Egemen D. The development of “automated visual evaluation’ for cervical cancer screening: the promise and challenges in adapting deep learning for clinical testing: interdisciplinary principles of automated visual evaluation in cervical screening. Int J Cancer. (2022) 150: 741–52. doi: doi:10.1002/ijc.33879
  92. 92 Wasson PS, Williamson MK, Hallas-Potts A, Lopez-Jimena B, Koscielniak M, Tait S. Female Genital Schistosomiasis: Translational Challenges and Opportunities: Outputs and actions from a consultative, collaborative and translational workshop. Parasitology., 1 – 44. doi: doi:10.1017/s0031182025100346
  93. 93 Mazigo HD, Maganga JK, Downs JA. Female genital schistosomiasis is a neglected public health problem in Tanzania: evidence from a scoping review. PLoS Negl Trop Dis. (2024) 18 (3): e0011954. doi: doi:10.1371/journal.pntd.0011954
  94. 94 Monteiro KHC, da Silva Rocha É, Morais LA, Santos EG, Neto SRDS, Sampaio V, Endo PT. Integrating machine learning and spatial clustering for malaria case prediction in Brazil’s Legal Amazon. BMC Infect Dis. (2025) 25 (1): 802. doi: doi:10.1186/s12879-025-11193-x
  95. 95 Mwinzi PN. Priority knowledge gaps for schistosomiasis research and …. Infect Dis Poverty. (2025). doi: doi:10.1186/s40249-025-01285-w
  96. 96 Oyibo P. Deep learning-based automated detection and multiclass classification of helminth eggs (STH and Schistosoma mansoni) using field-of-view images. Sci Rep. (2025). Available online at: https://www.nature.com/articles/s41598-025-02755-9 (Accessed August 06, 2026).
  97. 97 Kabuyaya M, Chimbari MJ, Mukaratirwa S. Efficacy of praziquantel treatment regimens in pre-school and school aged children infected with schistosomiasis in sub-Saharan Africa: a systematic review. Infect Dis Poverty. (2018) 7 (1): 73. doi: doi:10.1186/s40249-018-0448-x
  98. 98 Abbas Q, Jeong W, Lee SW. Explainable AI in clinical decision support systems: a meta-analysis of methods, applications, and usability challenges. Healthcare (Basel). (2025) 13 (17): 2154. doi: doi:10.3390/healthcare13172154
  99. 99 Christinet V, Lazdins-Helds JK, Stothard JR, Reinhard-Rupp J. Female genital schistosomiasis (FGS): from case reports to a call for concerted action against this neglected gynaecological disease. Int J Parasitol. (2016) 46 (7): 395–404. doi: doi:10.1016/j.ijpara.2016.02.006
  100. 100 Cornell University News. Taking on a Tropical Parasite, With Women in Mind (2025).
  101. 101 Díaz de León Derby M, Delahunt CB, Spencer E, Coulibaly JT, Silué KD, Bogoch II. Multi-contrast machine learning improves schistosomiasis diagnostic performance. PLoS Negl Trop Dis. (2025) 19 (8): e0012879. doi: doi:10.1371/journal.pntd.0012879

Citation

Innocent, D. C., et al. (2026). Explainable machine learning for predicting treatment outcomes in female genital schistosomiasis and HIV co-morbidity: a narrative review and translational framework. Frontiers in Tropical Diseases, Volume 7. https://doi.org/
Received
Accepted
Published
Volume
7 - 2026

Copyright

Copyright © 2026 Innocent, Innocent and Innocent.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: David Chinaecherem Innocent

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Original ResearchCapacity strengthening for community health workers in trachoma prevention and referrals: a mixed-methods study in Al Sabra district, Ibb governorate, YemenLeen Daoud, Magid Al Gunaid, Aïssatou Diawara, Mahanna Elsheik, Alison Krentel, Sami A. Al-HaidariBrief Research ReportComparative genomic analysis of established and emerging anti-tubercular drug targets reveals evolutionary conservation and antimicrobial resistance patterns in Mycobacterium tuberculosis isolates from IndonesiaNi Made Puspawati, Iryanti Eka Suprihatin, Akshay Suthar, Ashaka Vansia, Rajesh PatelReviewThe global emergence of Trichophyton indotineae: epidemiology, resistance, and management challengesDipika Shaw, Adnan Omar, Michelle Singh, Merin Chandanathil, Chandana Bhargavi, Nagaraj Mysore, Dharam Bir, Srinidhi Ramasamy, Prasanna Honnavar