NCT05024838

Prediction of Block Height of Spinal Anesthesia

Prediction of Block Height of Spinal Anesthesia Via Machine Learning Approach

• Status: Unknown
• Phase: —
• Study type: Observational
• Enrollment: 3,000 (estimated)
• Sponsor: Taipei Veterans General Hospital, Taiwan · Other Government
• Sex: All
• Age: 18 Years
• Healthy volunteers: Accepted

Summary

Spinal anesthesia is one of the most used techniques for surgery. Anesthesiologists usually check the block height (dermatome) of spinal anesthesia before surgery start. More than 20 factors have been postulated to alter spinal anesthetic block height. We would like to use machine learning to comprehensively consider various factors such as physiological parameters and different drug characteristics to establish a predictive model to evaluate the sensory blockade of spinal anesthesia.

Detailed description

This is an observational study of the retrospective collection of patient data. The investigators retrospectively collected the electronic medical record of patients receiving spinal anesthesia from July 1, 2018, to Dec 31, 2018. Anesthesia-related factors such as anesthesiologist's expertise, injection site, patient position, the dosage of local anesthetics, needle size, the direction of needle bevel, and basic demographic information of the patients were used for data analysis. Patients less than 18 years old were excluded from this study. Twenty percent of the dataset was used as a testing dataset, and the remaining were used for model training. The investigators will utilize four machine learning algorithms as XGBoost (Extreme Gradient Boosting), AdaBoost (Adaptive Boosting), Random Forest (RF), and support vector machine (SVM). Model performances were evaluated visually with a confusion matrix.

Conditions

• Anesthesia; Reaction

Interventions

Type	Name	Description
OTHER	Machine learning methods	This is an observational study of the retrospective collection of patient data. Anesthesia-related factors such as anesthesiologist's expertise, injection site, patient position, the dosage of local anesthetics, needle size, the direction of needle bevel, and basic demographic information of the patients were used for data analysis. Patients less than 18 years old were excluded from this study. Twenty percent of the dataset was used as a testing dataset, and the remaining were used for model training. The investigators will utilize four machine learning algorithms as XGBoost (Extreme Gradient Boosting), AdaBoost (Adaptive Boosting), Random Forest (RF), and support vector machine (SVM). Model performances were evaluated visually with a confusion matrix.

Timeline

• Start date: 2020-10-01
• Primary completion: 2022-07-01
• Completion: 2022-07-01
• First posted: 2021-08-27
• Last updated: 2021-08-27

Locations

1 site across 1 country: Taiwan