PHRT

SWISSHEART Failure Network (SHFN) – PHRT

Project

SWISSHEART Failure Network (SHFN)

Short Summary

Heart failure (HF) is a frequently occurring syndrome whose multiple triggers and drivers remain poorly understood. Currently, most data are collected in a non-standardized fashion. Our national consortium will create a standardized data infrastructure for a SwissHeart Failure Registry. Automated patient data analyses using machine learning is expected to improve HF patient management.

Goals

The SPHN participants will build a Swiss-wide, standardized data infrastructure focusing on typical cardiovascular patients. The SWISSHEART Failure Registry will collect clinical, laboratory, electrocardiogram and imaging data of patients at risk for HF (patients with heart attack) and patients hospitalized for acute HF. The PHRT participants will integrate multi-dimensional features from these patient data into machine learning-based diagnostic and risk scores. A magnetic resonance imaging-based disease simulator reduced to a 4D digital heart model will be personalized based on cardiac ultrasound and electrocardiogram data.

Significance

Building this national collaborative SWISSHEART Failure Network will set the stage for standardized nationwide cardiovascular research in Switzerland and strengthen synergies between cardiology experts at Universities and computer science experts at ETH. We anticipate that this project will improve prediction and prevention of HF and decrease HF progression in specific patients.

Background

Heart failure (HF) affects one in five adults over 40 years of age and confers a dismal prognosis. In half of all HF patients, HF is caused by a heart attack. Yet, the triggers of acute HF, the prediction of patients at risk for HF and the drivers of HF progression remain poorly understood. Moreover, the combined analysis of risk factors, electrocardiograms, imaging and laboratory data is complex, multidimensional and not standardized in Switzerland. Machine Learning (ML) can deal with such complexity and offers an automated way to discover key patterns in big data sets of clinical and diagnostic nature. ML methods bear a great potential to improve diagnostic and prognostic accuracy, thereby allowing a more tailored management of patients with HF.

Driver Project

Prof. Dr. Joachim M. Buhmann

ETH Zurich

Co-Investigators

  • Christian M. Matter (USZ)

Consortium

Status
In Progress

Funded by