Using longitudinal data and deep learning models to enhance resource allocation in home-based medical care

Area of Science:

• Healthcare Informatics
• Artificial Intelligence in Medicine
• Geriatric Care

Background:

• Aging population increases healthcare demand and costs.
• Effective home healthcare programs are essential.
• Accurate patient assessment is key for resource allocation and tailored services.

Purpose of the Study:

• To explore artificial intelligence (AI) applications in predicting home medical care stages.
• To enhance home care delivery through predictive modeling.
• To evaluate the performance of deep learning models in this context.

Main Methods:

• Retrospective study using electronic health records from Taipei City Hospital (2015-2021).
• Compared deep learning (Transformer, LSTM, GRU) and machine learning models.
• Trained models on 3, 5, and 10 consecutive patient visits for classification tasks.

Main Results:

• LSTM model achieved highest AUC (0.908) distinguishing home care presence/absence using 5 visits.
• Transformer model achieved best AUC (0.86) for classifying 4 home care stages (S0-S3) with 5 visits.
• Models trained on 5 visits provided sufficient accuracy for predictions.

Conclusions:

• AI deep learning models demonstrate significant potential for predicting home medical care stages.
• The best-performing models can aid healthcare professionals in optimizing resource allocation.
• Accurate prediction of care stages can lead to more efficient and effective home healthcare delivery.