Enterprise AI Analysis
Geochemical and machine learning approaches to groundwater fluoride prediction in Karaga District, Northern Ghana
Authors: Emmanuel Daanoba Sunkari, Dickson Abdul-Wahab, Mélida Gutiérrez, Prasun Chakrabarti, Abayneh Ataro Ambushe
Publication: Scientific Reports | Date: March 30, 2026
Executive Impact
Revolutionizing Fluoride Risk Management in West Africa
This analysis reveals a groundbreaking integrated framework to tackle severe groundwater fluoride contamination, ensuring safe drinking water for communities and driving sustainable public health outcomes.
The Opportunity
Traditional approaches to fluoride prediction often fall short in complex hydrogeological settings. Our integrated geochemical modeling and machine learning framework provides superior accuracy and interpretability, offering a scalable solution for similar regions globally. This translates to cost-effective community screening and targeted interventions, moving beyond reactive measures to proactive risk management.
0
Samples Exceeding WHO Guideline
0
ML Model Accuracy (MLP)
0
Mobility Index Discrimination
0
Dominant Na-HCO₃ Water Type
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
The study characterized the hydrogeochemical controls on fluoride mobilization, highlighting the dominance of Na-HCO₃ waters and the role of evaporite dissolution.
17.6%
of groundwater samples in Karaga District exceed WHO guideline (1.5 mg/L) for fluoride.
Karaga District - A Case Study in Fluoride Mobilization
The Karaga District in Ghana's Northern Region serves as a critical hotspot for fluoride contamination, with 4 out of 10 children potentially exposed to concentrations exceeding 1.5 mg/L. Our integrated approach revealed that evaporite dissolution drives extreme contamination, with Na-Cl waters exhibiting the highest fluoride levels (mean 3.75 mg/L). Machine learning identified total dissolved solids (TDS) and pH as primary predictors, demonstrating the non-linear fluoride behavior. The developed Mobility Index achieved an AUROC of 0.94 for WHO guideline exceedance, enabling cost-effective community screening and supporting targeted intervention strategies. This highlights the importance of localized, mechanistically grounded frameworks for addressing geogenic fluoride risk.
An integrated framework combining geochemical modeling, compositional data analysis, and machine learning was used to predict fluoride concentrations and elucidate mobilization mechanisms.
| Model | R² | MAE | RMSE |
|---|---|---|---|
| Multilayer Perceptron | 0.668 | 0.654 | 1.196 |
| Histogram-based Gradient Boosting | 0.648 | 0.747 | 1.436 |
| Random Forest | 0.586 | 0.762 | 1.534 |
| XGBoost | 0.350 | 0.894 | 1.876 |
| Ridge Regression | -6.064 | 1.421 | 3.608 |
| Lasso Regression | -0.035 | 1.455 | 2.323 |
Enterprise Process Flow
Field Chemistry & Hydrogeological Context
→
PHREEQC Thermodynamic Modelling
→
Compositional Data Analysis (CoDA)
→
ML Prediction & Interpretability
→
Mechanistic Aggregation: Mobility Index (MI)
→
WHO Exceedance Screening
The developed Mobility Index offers a field-deployable and cost-effective solution for community-level fluoride surveillance and targeted intervention strategies.
AUROC 0.94
for Mobility Index in identifying WHO exceedance.
The Mobility Index provides a cost-effective, field-operationalized early warning framework for community-level fluoride surveillance. It leverages basic measurements (EC, pH, major ions) to predict fluoride risk without requiring direct fluoride analysis, making it ideal for resource-constrained regions. The framework is designed for robust spatial transferability, enabling broad deployment across similar sedimentary aquifer contexts.
Calculate Your Impact
Estimate Your Potential ROI with AI-Powered Water Quality Analysis
Quantify the potential savings and efficiency gains for your organization by implementing AI-driven solutions for environmental monitoring and risk management.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Your Journey to AI-Powered Water Quality
Seamlessly Integrate Advanced Prediction into Your Operations
Our structured roadmap ensures a smooth transition to an AI-driven water quality monitoring and risk assessment system, tailored to your organization's needs.
Phase 1: Discovery & Data Integration
We begin by understanding your specific hydrogeological context and integrating existing water quality datasets. This involves configuring PHREEQC for your local geochemistry and setting up compositional data analysis pipelines.
Phase 2: Model Customization & Calibration
Our machine learning experts customize and calibrate the predictive models using your historical data, ensuring optimal performance for your region. The Mobility Index is tuned for local conditions and WHO guideline exceedance thresholds.
Phase 3: Deployment & Training
The AI-powered system is deployed, and your team receives comprehensive training on using the Mobility Index for real-time risk assessment and interpreting model insights. We ensure seamless integration with your existing monitoring workflows.
Phase 4: Ongoing Optimization & Support
We provide continuous support and iterative model optimization based on new data and evolving environmental conditions. This phase includes periodic performance reviews and advanced feature enhancements.
Next Steps
Ready to Transform Your Water Quality Monitoring?
Book a personalized consultation with our AI specialists to explore how these advanced geochemical and machine learning approaches can benefit your organization.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat