A new model for delivering medicine in Liberia

Researchers developed a way to manage scarce medication supplies for community health workers.

A new model helps manage scarce medications in Liberia's community health program. This model helps decide how to distribute medicine from a central warehouse to health facilities and Community Health Assistants (CHAs).

The model balances the cost of patient referrals with the need for fairness across locations. A referral occurs when a CHA runs out of medicine and patients must travel to a facility. Keeping medicine at the central warehouse provides flexibility to respond to uncertain demand.

The researchers used a finite-horizon stochastic control problem to study this allocation. They developed a continuous-time approximation involving the drift rate of a reflected Brownian motion. They also used the deep splitting method with deep neural networks to find a dynamic allocation policy.

The model was tested using malaria case data from children in the Fangonda region of Lofa County, Liberia. The new policy reduced operating costs by about 6% compared to a fairness-only policy. It also kept approximately 98% of the fairness utility.

Higher referral costs shift inventory toward CHAs and reduce local unmet demand. Greater demand variability increases the performance of this policy over the fairness-only benchmark.

Key facts

  • The policy reduces operating costs by about 6% compared to the fairness-only policy.
  • The model retains approximately 98% of the fairness utility.
  • The study used malaria data from children in the Fangonda region of Lofa County, Liberia.

All sections

Edition