Methodology to automatically detect abnormal values of vital parameters in anesthesia time-series: Proposal for an...

Area of Science:

• Anesthesiology
• Medical Informatics
• Health Data Analysis

Background:

• Abnormal vital signs during anesthesia, like hypotension and tachycardia, can indicate risks.
• Linking these abnormal vital signs to postoperative morbidity and mortality is crucial for patient safety.
• Current methods for automatic detection of abnormal vital signs are not well-documented, hindering reproducibility.

Purpose of the Study:

• To propose a novel methodology for the automatic detection of abnormal vital parameter values.
• To develop an algorithm capable of handling configurable thresholds and missing data for vital signs.
• To apply and validate the proposed methodology on a large dataset of anesthetic records.

Main Methods:

• Developed an algorithm for automatic detection of abnormal vital parameters.
• Algorithm allows configuration of thresholds for any vital sign.
• Algorithm incorporates management of missing data points.
• Applied the algorithm to heart rate, SpO2, and mean arterial pressure data.

Main Results:

• Successfully applied the algorithm to detect abnormal vital signs.
• Demonstrated the algorithm's ability to manage configurable thresholds and missing data.
• Validated the methodology on 2014 anesthetic records from the institution.

Conclusions:

• The proposed methodology offers a robust approach for automatic detection of abnormal vital signs during anesthesia.
• This method facilitates consistent and reproducible identification of critical events.
• Automated detection can improve patient safety by flagging events linked to adverse outcomes.