Planned tutorial topics

[peterhcharlton]

• Data Exploration: Would be good to extend this to illustrate the structure of the WFDB files (e.g. intermediate and record directories, as described here).
• Data Extraction: This currently uses data from the BIDMC dataset ('bidmc'), rather than MIMIC ('mimic3wdb'). This is because using MIMIC data currently results in an error when using the wfdb.rdrecord function in Section 1.2.
• Data Visualisation: Also currently uses data from BIDMC for the same reason as above.
• Signal filtering: filtering PPG signals to eliminate baseline wander and high frequency noise - although perhaps we should omit this as it isn't really needed for the MIMIC database, since the signals are already filtered by the monitoring equipment.
• Beat Detection: detecting beats in PPG and BP signals from a specified record.
• Signal Quality Assessment: (optional) assessing PPG and BP signal quality.
• Signal segmentation: segmenting signals into short durations (e.g. 10s) or individual pulse waves.
• Differentiation: differentiating PPG signals to obtain the first and second derivatives for analysis.
• Pulse wave analysis: identifying fiducial points in PPG pulse waves, and extracting features from them (such as the time delay between two fiducial points).
• Data modelling: creating a linear regression model to estimate BP from PPG pulse wave features.
• Data analysis: plotting results (e.g. scatter plot, or Bland-Altman plot); calculating statistics (such as bias and limits of agreement, or mean absolute error)
• Data interpretation: comparing performance to that required by standards (e.g. AAMI standard).