Locate Lead Service Lines with ground truth accuracy
BlueConduit uses rigorous, academically-based and market-tested AI and machine learning algorithms, supported by best-in-class people + process, to provide the best predictions available.
Why BlueConduit?
Benefits of BlueConduit’s Lead Service Line predictive analysis solution
Market-tested accuracy
Early BlueConduit clients report a >90% ground truth accuracy rate for lead likelihood predictions.
Rigorous machine learning algorithms
The BlueConduit Data Science Team applies a wealth of academic and industry experience to arrive at deep analytical insights tailored to specific communities.
Predictions for multiple materials
LSL Predictions provides predictive analysis for a variety of hazardous materials, including lead, GRR, and more.
Proven process
BlueConduit’s predictive process, supported by expert Data Scientists, has built lead likelihood predictive models for hundreds of communities.
Transparency at every step
BlueConduit customers stay informed at every step of the process with easy-to-understand written reports and communications available to share with internal and external stakeholders, funding partners, and anyone who needs to be in the know, regardless of their level of data science expertise.
Esri integration
Predictions and inspection lists are directly into Esri, alongside easy to use dashboards and metrics, as well as other easy-to-use formats.
How it Works
BlueConduit conducts a comprehensive review of your existing service line inventory data.
Based on the data review and conversations with water system stakeholders, BlueConduit’s Data Science Team will recommend a phased list of representative field investigations needed to ensure statistically sound, compliant predictions.
Based on the representative sample combined with existing data, LSL Predictions, supported by expert Data Scientists, makes likelihood of lead predictions for every service line segment of unknown material. For water systems with no known or inspected lead, a No Lead Validation report can be generated to statistically validate the low likelihood of lead across the full water system.
Guided by our lead service line predictions, operations to dig and replace pipes can begin. As work is carried out, the predictive model applies newly discovered information to improve its predictions going forward.
BlueConduit conducts a comprehensive review of your existing service line inventory data.
Based on the data review and conversations with water system stakeholders, BlueConduit’s Data Science Team will recommend a phased list of representative field investigations needed to ensure statistically sound, compliant predictions.
Based on the representative sample combined with existing data, LSL Predictions, supported by expert Data Scientists, makes likelihood of lead predictions for every service line segment of unknown material. For water systems with no known or inspected lead, a No Lead Validation report can be generated to statistically validate the low likelihood of lead across the full water system.
Guided by our lead service line predictions, operations to dig and replace pipes can begin. As work is carried out, the predictive model applies newly discovered information to improve its predictions going forward.
Metrics
BlueConduit’s Predictions
80-90%+
Recall & precision
BlueConduit predictive models routinely identify 80-90%+ of lead total lines (recall) and these positive predictions are 80-90%+ accurate (precision).
Up to 80%
Increase in AUC ROC
BlueConduit predictive models demonstrate up to an 80% additional gain in performance compared to baseline, simple heuristic models (ex, age of home).
>90%
Hit rate
For early customers who’ve already started replacements, lead likelihood was predicted correctly with a >90% ground truth success rate.
Ready to get the lead out?
Connect with BlueConduit today to set up a personalized consultation.
