Machine Learning-Based Risk Profile Classification of Patients Undergoing Elective Heart Valve Surgery

Completed

• Conditions: Surgery--Complications; Heart Valve Diseases

• Registration Number: NCT03724123
• Lead Sponsor: Kepler University Hospital

Brief Summary

Machine learning methods potentially provide a highly accurate and detailed assessment of expected individual patient risk before elective cardiac surgery. Correct anticipation of this risk allows for improved counseling of patients and avoidance of possible complications. The investigators therefore investigate the benefit of modern machine learning methods in personalized risk prediction in patients undergoing elective heart valve surgery.

Detailed Description

The investigators performe a monocentric retrospective study in patients who underwent elective heart valve surgery between January 1, 2008, and December 31, 2014 at our center. The investigators use random forests, artificial neural networks, and support vector machines to predict the 30-days mortality from a subset of demographic and preoperative parameters. Exclusion criteria were re-operation of the same patient, patients that needed anterograde cerebral perfusion due to aortic arch surgery, and patients with grown up congenital heart disease.

Study Timeline

• Study Start: 2008-01-01
• Primary Completion: 2014-12-31
• Study Completion: 2014-12-31
• Status Verified: 2018-10
• Study First Submitted: 2018-10-25
• Study First Submitted QC: 2018-10-26
• Last Update Submitted: 2018-10-26
• Study First Posted: 2018-10-30
• Last Update Posted: 2018-10-30

Recruitment & Eligibility

• Status: COMPLETED
• Sex: All
• Target Recruitment: 2229

Inclusion Criteria

* Patients who underwent heart valve surgery of any kind between 2008-01-01 and 2014-12-31 were included.

Exclusion Criteria

• re-operation of the same patient
• patients that needed anterograde cerebral perfusion due to aortic arch surgery
• patients with grown-up congenital heart disease

Study & Design

• Study Type: OBSERVATIONAL
• Study Design: Not specified

Primary Outcome Measures

Name	Time	Method
Area under the curve for different prediction models	Patients will included from 01.01.2008 - 31.12.2014	Three different predictions models will be used.