DNN based reliability evaluation for telemedicine data.

Area of Science:

• Biomedical Engineering
• Health Informatics
• Artificial Intelligence in Medicine

Background:

• Telemedicine data quality is often compromised by untrained patient measurements.
• Existing deep learning methods lack a robust basis for judging data reliability.
• Accurate assessment of remote patient data is crucial for effective diagnosis and treatment.

Purpose of the Study:

• To develop and validate a deep neural network filter-based system for evaluating telemedicine data reliability.
• To establish an accurate basis for judging the quality of patient-measured data.
• To assess the system's performance using photoplethysmography signals and blood pressure data.

Main Methods:

• Implementation of a deep neural network filter for reliability evaluation.
• Clinical trials involving photoplethysmography (oxygen saturation) and diastolic blood pressure measurements.
• Analysis of data deviation under different judgment criteria for normal and abnormal data.
• Evaluation of system performance with single versus multiple judgment conditions.

Main Results:

• Low deviation (0.3%-0.82%) for normal oxygen saturation judgments compared to abnormal (3.86%).
• Diastolic blood pressure deviation reduced by ~4% for normal judgments versus abnormal.
• The system demonstrated superior discrimination of abnormal data when multiple criteria were met.
• The proposed system provides a reliable basis for judging telemedicine data quality.

Conclusions:

• The deep neural network filter system effectively enhances the reliability of telemedicine data.
• The system provides an accurate judgment basis, crucial for improving remote patient monitoring.
• This approach can significantly improve the quality and trustworthiness of data collected via telemedicine platforms.