Although chemotherapy drugs improve the survival rate of cancer patients, they also inevitably bring a series of cardiovascular toxicities such as arrhythmia and heart failure. Professor Ryuichiro Yagi of Brigham and Women's Hospital, USA shared at the ESC 2022 conference. This issue specially invited him to introduce an artificial intelligence (AI) application study carried out by his team in this part of patients.
International Circulation: Would you please introduce the influencing factors of chemotherapy-induced cardiotoxicity?
Ryuichiro Yagi: Several factors such as age, sex, cancer type, and treatment regimen are known to be associated with the occurrence of chemotherapy-induced cardiotoxicity (CIC). However, it has long been challenging to accurately predict the occurrence of cardiotoxicity before starting chemotherapy. Therefore, we aim to tackle this issue using a machine learning approach analyzing ECG.
International Circulation: Would you please share us the main results of a related study you announced at this ESC conference?
Ryuichiro Yagi: Basically, the Mayo team published an AI model detecting low LVEF from a single recording of ECG data. We generated the model from scratch using our master clinical datasets and applied this model for our purposes - the screening of high-risk patients for chemotherapy-induced cardiotoxicity prior to the initiation of chemotherapy. We found that the model accurately predicted the occurrence of cardiotoxicity using only the baseline ECG taken before starting chemotherapy. I believe this is a very impressive result for this area of research.
International Circulation: How effective is this artificial intelligence model in clinical applications, and how will it be improved in the future?
Ryuichiro Yagi: Currently, the guidelines recommend repeated evaluation of cardiac function using echocardiograms before, during and after cardiotoxic chemotherapy. However, this is not routinely done in many hospitals, which means that patients with cardiotoxicity are under-diagnosed. We showed the potential for the ECG as a screening tool enabling thedetection of high-risk patients for chemotherapy-induced cardiotoxicity using the power of AI in a very easy way. That could be a game changer for predicting and preventing cardiotoxicity.
We performed this study across two institutions, but this study should be further validated using data from outside our hospitals to show the robustness of the model prediction, and to show the clinical utility of the model’s prediction. This is the work that should be done next.
International Circulation: With several guidelines and new research published at this year’s conference, which research are you most interested in?
Ryuichiro Yagi: I was interested in the EchoNet study in which AI analyzed echocardiograms to show the clinical utility. This is a very interesting result. I feel like the reality of the clinical application of AI is getting closer.
There are other centers doing machine learning research, but I was surprised to see so many presentations at this conference related to this topic. There are many obstacles for AI to be deployed in the clinical setting until now, but I think the future is bright for future applications of artificial intelligence and for use by clinicians for the benefit of their patients.
Comments
Post a Comment