Combining Machine Learning and Blind Estimation for Central Aortic Blood Pressure Reconstruction.

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Data Science

Background:

• Central blood pressure is crucial for assessing cardiovascular disease risk.
• Invasive measurement methods are challenging, while non-invasive methods lack accuracy.
• Existing estimation techniques often miss key features of pressure waveforms.

Purpose of the Study:

• To develop a novel, data-driven approach for reconstructing central aortic blood pressure waveforms.
• To improve the accuracy of non-invasive blood pressure estimation using machine learning and blind estimation.
• To validate the proposed method against existing techniques using in-silico data.

Main Methods:

• A hybrid approach combining machine learning and cross-relation-based blind estimation.
• Utilizing in-silico virtual pulse wave databases for model training due to limited real-world data.
• Comparing the hybrid model's performance against pure machine learning and pure cross-relation methods.

Main Results:

• The hybrid approach demonstrated superior performance in reconstructing central blood pressure waveforms.
• Root-mean-squared error was reduced by 25% compared to pure machine learning methods.
• A 40% reduction in root-mean-squared error was observed compared to cross-relation-based blind estimation.

Conclusions:

• The proposed hybrid method offers a more accurate and non-invasive way to estimate central blood pressure.
• This approach holds promise for improved cardiovascular disease risk factor monitoring.
• In-silico databases are effective for training machine learning models in scenarios with limited real-world data.