28 July 2026 · Research team

From Western Kenya data to discontinuation risk annotation

78,000+ anonymised programme records inform an advisory model; full methodology is in our technical report.

  • Methods

Our models learn from anonymised family planning programme records from Siaya and Busia counties via the DASSA platform under the HASH Innovation Challenge. We report holdout performance and calibration in the technical report and share reproducibility materials with research partners under agreement.

Local data matters: discontinuation drivers in Western Kenya are not interchangeable with global averages. We document bias audits and geographic splits so partners can trust what the numbers mean — and what they do not mean.

All research notes

Open Source

Repository coming soon

Reproducibility Package

Reproduce our ML results