EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Predicting Blood Donor Retention Using the Random Forest Classification Algorithm: A Machine Learning Approach

Marchel A Acilador, Ann P Opiña

Acta Medica Philippina · 04.06.2026

DOI: 10.47895/amp.vi0.13962 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Acta Medica Philippina, online erschienen am 04.06.2026. Verfasst von Marchel A Acilador und 1 weiteren Autor:innen. Beteiligte Einrichtung: Saint Louis University, Baguio City, Philippines. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Maintaining a stable and safe blood supply remains a persistent challenge for blood collection agencies, particularly in settings where donor participation is influenced by behavioral and psychological factors. Identifying reliable predictors of blood donor retention is essential for improving donor management and sustaining blood services.

Objective

To develop and evaluate a Random Forest–based machine learning model for predicting blood donor retention using demographic, behavioral, and psychological data.

Methods

A retrospective cohort with a cross-sectional exploratory component was conducted among blood donors from the Department of Health–Regional Blood Center of Cagayan Valley. Data were obtained from two sources: retrospective records of 612 donors (2023–2025), which included demographic characteristics, blood type, and donation history, and a survey administered to 100 prospective donors to assess psychological factors such as motivation, attitudes, satisfaction, and perceived barriers. Behavioral features—donation frequency, recency, tenure, and total donation count—were derived from donation records. Psychological variables were analyzed using Principal Component Analysis and K-means clustering. Model performance was evaluated using accuracy, F1 score, and area under the receiver operating characteristic curve (ROC–AUC).

Results

Demographic and biological characteristics, including age, sex, and blood type, described the donor population but were not key predictors of donor retention. Behavioral indicators—particularly donation frequency, recency, tenure, and cumulative donation count—were the strongest predictors of continued donation. Psychological analysis identified distinct donor profiles, with most respondents exhibiting high motivation and satisfaction, and smaller clusters reporting lower satisfaction or greater perceived barriers. The Random Forest model demonstrated strong predictive performance, with an accuracy of 0.989, F1 score of 0.994, and ROC–AUC of 0.994.

Conclusion

Blood donor retention is driven primarily by behavioral engagement and psychological factors rather than demographic or biological characteristics. The Random Forest model effectively identifies donors likely to return, supporting targeted donor engagement strategies and improved resource allocation in blood service facilities.

Schlagwörter

Stichwörter der Autor:innen

Random Forestbehavioral predictorsdonor retentionmachine learningpsychological factors

Autor:innen und Institutionen

  • Saint Louis University, Baguio City, Philippines.

Zitieren

APA

Acilador, M. A., & Opiña, A. P. (2026). Predicting Blood Donor Retention Using the Random Forest Classification Algorithm: A Machine Learning Approach. Acta Medica Philippina. https://doi.org/10.47895/amp.vi0.13962

Vancouver

Acilador MA, Opiña AP. Predicting Blood Donor Retention Using the Random Forest Classification Algorithm: A Machine Learning Approach. Acta Medica Philippina. 2026. doi:10.47895/amp.vi0.13962

BibTeX
@article{Acilador2026,
  author = {Acilador, Marchel A and Opiña, Ann P},
  title = {Predicting Blood Donor Retention Using the Random Forest Classification Algorithm: A Machine Learning Approach},
  journal = {Acta Medica Philippina},
  year = {2026},
  doi = {10.47895/amp.vi0.13962}
}
Kostenlos starten →