JMIR Publications

ABSTRACT

Background:

Machine learning has been demonstrated to enhance healthcare cost prediction by handling high-dimensional data and identifying complex patterns. However, current risk adjustment models rarely incorporate structured nursing information derived from the nursing process. This information captures care needs and human responses to health problems. C

Objective:

urrent study aimed to evaluate the impact of integrating nursing process data into machine learning-based predictive models of individual healthcare costs, including cost component analyses, compared with models based solely on sociodemographic, clinical, and morbidity-related variables.

Methods:

A retrospective observational study was conducted using a population-based cohort of 1,691,075 individuals aged ≥ 15 years who were registered with the Canary Islands Health Service. Predictors were derived from data available up to 2017 and included sociodemographic and clinical variables, morbidity adjustment (AMG), healthcare utilization, and structured nursing records (functional health patterns, NANDA, NOC and NIC). Predictive models were developed using feedforward neural networks and XGBoost; predictions were combined using an ensemble approach. An autoencoder was applied as a dimensionality-reduction technique for the nursing variables. Model performance with and without nursing variables was compared on total cost and individual cost components, and the coefficient of determination (R²) was used on the test set.

Results:

Including the nursing methodology yielded modest but consistent improvements in predictive performance. With respect to total cost, the best-performing model improved from R²=0.5023 to R²=0.5058 when the nursing variables were added. In the component-level analyses, performance gains were observed in hospital care (R²=0.3829) and pharmaceutical costs (R²=0.6631). Reducing 789 nursing variables to 16 latent dimensions using an autoencoder substantially simplified the feature space while retaining comparable predictive performance.

Conclusions:

Integrating structured information from the nursing process adds incremental value to machine learning-based predictive models and complements commonly used sociodemographic, clinical, and morbidity variables. The systematic incorporation of nursing data into predictive tools may contribute to more accurate healthcare cost prediction and support more holistic, person-centered approaches. Clinical Trial: Not Applicable