Dr. Monish Raut throws light on a very novel SEMMELWEIS-CRT score which significantly outperforms other conventional risk scores to predict mortality among patients undergoing cardiac resynchronization therapy (CRT) implantation.

Cardiac resynchronization therapy (CRT) has been effectively useful in managing patients with symptomatic systolic heart failure and wide QRS complex. Many studies have demonstated the advantages of improved clinical symptoms, functional capacity and quality of life in these patients after CRT implantation. However, it is essential to know that not every patient is benefitted equally. High mortality rate in CRT recipients can not be ignored.

Risk stratification of pre-CRT implantation patients can potentially identify high risk groups. This has prompted the development of presently available risk scores such as (CRT-score, Seattle Heart Failure Model, ScREEN, VALID-CRT and EAARN), but none of these were found to be reliable and effective in predicting the outcomes for an individual patient.

A researcher from the Heart and Vascular Center of Semmelweis University (Budapest, Hungary) has developed a novel SEMMELWEIS-CRT score based on machine learning (ML) and artificial intelligence (AI). Machine learning recognises the patterns from multidimensional datasets to study high-dimensional and non-linear relationships of clinical profiles of the patients without using any prior assumptions. This risk stratification system based on machine learning can predict mortality more accurately for individual patients undergoing CRT implantation.

The SEMMELWEIS-CRT score analyses 33 routinely assessed clinical variables of these patients available from electronic medical records. Researcher uses the two independent cohorts in retrospective and the prospective databases extracted from electronic medical records at the Heart and Vascular Center of Semmelweis University (Budapest, Hungary).

The retrospective database of 2282 patients with successful CRT implantation comprises preimplant clinical parameters such as medical history, physical vitals, present medical therapy, laboratory investigations, ECG and echocardiography. Similarly, prospective database of patients undergoing CRT implantation was also used additionaly. Machine learning algorithms use these cohorts as training and test cohort. All-cause mortality was the primary endpoint of the study. Six classes of possible outcomes were generated using the follow up data.

• Class 1 - Death during the 1st year after CRT implantation
• Class 2 - Death during the 2nd year after CRT implantation
• Class 3 - Death during the 3rd year after CRT implantation
• Class 4 - Death during the 4th year after CRT implantation
• Class 5 - Death during the 5th year after CRT implantation
• Class 6 - No death during the first 5 years following the implantation

Other preexisting risk scores such as CRT-score, Seattle Heart Failure Model, ScREEN, VALID-CRT and EAARN were also applied to every patient of the test cohort. Their prediction results were compared with SEMMELWEIS-CRT score.

An online calculator (available at semmelweiscrtscore.com) can be conveniently used to calculate the predicted mortality in patients. Optimal patient selection for CRT implantation and appropriate prognostication can be facilitated using this new machine learning based risk stratification system.

Reference: Márton Tokodi, Walter Richard Schwertner, Attila Kovács, Zoltán Tősér, Levente Staub, András Sárkány, Bálint Károly Lakatos, Anett Behon, András Mihály Boros, Péter Perge, Valentina Kutyifa, Gábor Széplaki, László Gellér, Béla Merkely, Annamária Kosztin, Machine learning-based mortality prediction of patients undergoing cardiac resynchronization therapy: the SEMMELWEIS-CRT score, European Heart Journal, , ehz902, https://doi.org/10.1093/eurheartj/ehz902

Disclaimer- The views and opinions expressed in this article are those of the author's and do not necessarily reflect the official policy or position of M3 India.

The author, Dr. Monish S Raut is a Consultant in Cardiothoracic Vascular Anaesthesiology. His area of expertise is perioperative management and echocardiography with numerous publications in various national and international indexed journals.