A carregar...
Publicação
Data-driven decision-making: Drivers of Year-over-Year Patient Return in a Healthcare Clinic
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.92 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Machine-Learning (ML) is transforming data-driven decision-making across industries.
However, Small and Medium Enterprises (SMEs), which are dominant both in Portuguese and
European economies, are still struggling to adopt the tools necessary to drive managerial
decisions and extract value from data. In the healthcare sector in particular, medical
practitioners are already using data and predictive analytics to improve disease diagnosis and
prediction; however, management teams often rely too heavily on intuition when making
decisions. To address the challenges of rapid technological advancements, an aging
population, increased patient demands, and scarcer human resources, healthcare
organizations must effectively leverage data as a strategic asset. Data-driven insights can
enhance business performance, with patient retention strategies serving as a critical example.
This study examines a real-world private clinic in Portugal, demonstrating how predictive
analytics can optimize strategic decision-making by integrating transactional data with ML
techniques. Using the Cross-Industry Standard Process for Data Mining (CRISP-DM)
methodology, we identified the variables most strongly associated with year-on-year patient
recurrence. We then developed and evaluated binary classification models to assess how
effectively these features could predict patient loyalty, both overall and within two key
medical specialties. The shortlisted features informed actionable recommendations to
support data-driven management and long-term organizational sustainability. Although the
primary dataset analyzed was moderately imbalanced, the balancing strategies used were
unable to improve predictive performance. The top-performing model, based on a Random
Forest classifier, achieved ROC scores exceeding 0.75. The 10 most predictive features
excluded categorical variables and comprised Recency, Frequency, Monetary, and
Clumpiness. The proposed methodology used in this empirical study offers a replicable
framework for other SMEs and sectors seeking to enhance customer loyalty through datadriven decision-making.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Business Intelligence
Palavras-chave
Machine-Learning Healthcare Management Decision-making SDG 3 - Good health and well-being SDG 8 - Decent work and economic growth SDG 9 - Industry, innovation and infrastructure SDG 11 - Sustainable cities and communities