Optimizing Donor Outreach for Blood Collection Sessions: A Scalable Decision Support Framework
Summary: arXiv:2603.29643v1 Announce Type: new
Abstract
Blood donation centers are often challenged with balancing the supply and demand of blood while also managing the availability of donors. Targeted outreach is critical in this context; however, it can lead to donor fatigue through over-solicitation. To effectively recruit donors, it is essential to target the right individuals at the appropriate times, all while considering constraints such as donor convenience and eligibility.
Despite significant research on blood supply chain optimization and an increasing interest in algorithmic donor recruitment, the operational challenge of assigning donors to sessions in a multi-site network remains largely unaddressed. This task involves factoring in donor eligibility, capacity limitations, blood-type demand targets, geographic convenience, and donor safety.
Proposed Framework
To address this gap, we present an optimization framework designed for scheduling donor invitations. This framework incorporates various critical factors including:
- Donor eligibility
- Travel convenience
- Blood-type demand targets
- Penalties for not meeting targets
We evaluate two primary strategies within this framework:
- Binary Integer Linear Programming (BILP) formulation
- Efficient greedy heuristic
Evaluation and Results
The evaluation utilizes data from the Instituto Português do Sangue e da Transplantação (IPST) to plan invitations in the Lisbon operational region over four-month windows. A prospective pipeline is integrated, which consists of:
- Organic attendance forecasting
- Quantile-based demand targets
- Residual capacity estimation for forward-looking invitation plans
Our findings indicate that this framework plays a crucial role in bridging the supply-demand gap within the Lisbon operational region. A controlled comparison reveals that the greedy heuristic achieves performance results comparable to the BILP, with the following notable advantages:
- 188x less peak memory usage
- 115x faster runtime
However, there are trade-offs associated with this approach, including:
- 3.9 percentage points lower demand fulfillment (86.1% compared to 90.0%)
- Larger distances between donors and sessions
- Higher exposure to adverse reactions for donors
- Increased invitation burden for non-high-frequency donors
Conclusion
Our experiments further demonstrate how constraint-aware scheduling can help mobilize eligible inactive or lapsing donors. This scalable decision support framework not only enhances the efficiency of donor outreach but also aims to optimize the overall operation of blood donation centers, ultimately contributing to better management of blood resources.