Development and usability testing of a health recommender system for symptom management in women with breast cancer who were receiving or had recently received chemotherapy.

OBJECTIVE

Both patients with breast cancer and health care providers face ongoing challenges in managing physical and psychological symptoms. This study aimed to develop a health recommender system (HRS) for symptom management of women with breast cancer who were receiving or had recently received chemotherapy and evaluate its usability.

METHODS

Conceptually informed by the Digital Intelligent Precise Nursing Framework, the Breast Health Recommender App (BHRA) was developed for breast cancer symptom management. This development involved qualitative interviews with patients with breast cancer and health care providers to establish a comprehensive knowledge base. Recommender rules were formulated based on insights from a multicenter, cross-sectional study using deep neural network (DNN) analyses. A multidisciplinary research team employed a user-centered design methodology to develop the app. A preliminary usability assessment was then conducted to assess the app's preliminary acceptability and perceived usefulness.

RESULTS

The BHRA comprises three user-facing modules-user profile management, symptom assessment, and individualized symptom management-and one back-end DNN classifier module. Its knowledge base is divided into a common module and a personalized recommendation module. Following iterative optimization and hyperparameter tuning, the final classifier employed four hidden layers with an initial learning rate of 0.10 and demonstrated robust multiclass classification performance, with excellent discrimination and overall predictive accuracy across the three symptom profile groups. Usability testing with 10 patients generated predominantly positive feedback, and participants suggested enhancements in the app's information presentation, instructions, structure, and interactive feedback capabilities.

CONCLUSIONS

The BHRA demonstrated preliminary promise as a supportive tool for individualized symptom self-management among women with breast cancer who were receiving or had recently received chemotherapy. Further controlled or longitudinal studies are needed to evaluate its clinical effectiveness and real-world implementation.