The nationwide competence network for congenital heart defects (KNAHF) is starting its AI research in what could mean new chances for heart medicine in Germany.
KNAHF is a German association that has been conducting research in the field of heart disease since 2003 in order to improve medical diagnostics, therapy and prevention. Its founding was initiated by the Federal Ministry of Education and Research (BMBF), and more than 30 universities, clinics, specialists and research institutions are involved in its international research.
Under the direction of cardiologists at the University Hospital Münster (UKM), it now intends to use AI to improve prevention, diagnostics and therapy for congenital heart defects in adults, and it has founded a platform drawing on one of the world’s largest patient registers of cardiac data.
The source is the National Register for Congenital Heart Defects, a database of patients worldwide used for research purposes, maintained by the competence network and several other scientific societies.
The registry currently collects around 55,000 anonymous patient datasets and samples, and the high volume enables researchers to determine disease patterns and give more reliable statements on prevention and chances of recovery by means of genetic data, the type of heart defect and existing reports.
Using international AI studies as a basis
Two international studies published in January 2019 in collaboration with the Royal Brompton Hospital in London paved the way for the new AI research platform in Germany, after the results indicated that deep learning algorithms may derive benefit when used in the diagnosis of heart defects by evaluating patient image data .
“The precise diagnosis and the differentiation of the findings of the heart chambers using AI exceeded our expectations,” said Gerhard-Paul Diller, senior physician at the Department of Cardiology at UKM, who will be heading the new platform along with an international team of scientists.
According to the first AI study, the diagnosis of congenital heart defects in adults becomes more precise with the help of deep learning algorithms. Compared with the conventional diagnosis by specialists, it achieved an accuracy of 98 percent.
Therapy-relevant: precise diagnosis helps with follow-up decisions
The deep learning algorithms also proved to be beneficial in terms of therapeutic decisions for healthcare professionals, as the second AI study showed.
“Used judiciously, algorithmically learning systems can significantly increase the life expectancy of patients,” predicted Diller.
The platform has been funded in a first instance by the EMAH Foundation Karla Völlm. In the future, the researchers hope, it will be supported by the German Federal Government and relevant stakeholders of the German economy.
Anna Engberg is a Wiesbaden-based freelance journalist specialising in health and technology.