Time-Aware Missing Healthcare Data Prediction Based on ARIMA Model.

Area of Science:

• Biomedical Informatics
• Data Science in Healthcare
• Health Informatics

Background:

• Modern healthcare generates vast datasets from devices like wearables and glucose meters.
• Missing healthcare data, caused by system failures, leads to inaccurate diagnoses and treatment anomalies.
• Accurate imputation of missing data is crucial for reliable patient management, medical research, and healthcare services.

Purpose of the Study:

• To propose a time-aware approach for predicting and imputing missing healthcare data.
• To address the limitations of traditional methods that ignore the temporal aspect of data.
• To enhance the accuracy and efficiency of healthcare data completion.

Main Methods:

• Developed a novel approach, MHDP _{SVD_ARIMA}, combining Singular Value Decomposition (SVD) with the Autoregressive Integrated Moving Average (ARIMA) model.
• Utilized truncated SVD to reduce data redundancy and noise, improving ARIMA model efficiency.
• Incorporated a time-aware component to capture temporal patterns in healthcare data.

Main Results:

• The MHDP _{SVD_ARIMA} approach demonstrated effectiveness and efficiency in predicting missing healthcare data.
• Experimental results on the WISDM dataset validated the proposed method's performance.
• The time-aware model accurately captured underlying patterns of healthcare data changes over time.

Conclusions:

• The proposed MHDP _{SVD_ARIMA} method offers a significant advancement in handling missing healthcare data.
• Accurate imputation of missing data using time-aware models is essential for robust healthcare analytics and decision-making.
• This approach provides a reliable foundation for patients, clinicians, and researchers by ensuring data completeness.