New Technologies for Cancer Registry Reporting in North Rhine-Westphalia
The KIMONA project uses modern artificial intelligence (AI) methods to automatically analyze medical data and prepare it for cancer registration. The goal is to reduce the workload on medical staff, transmit reports more quickly, and simultaneously improve data quality. The project is part of a collaboration among the major university cancer centers in North Rhine-Westphalia, which are organized within the "Network of Excellence in Cancer Medicine". These include, among others, the Center for Integrated Oncology (CIO) at the University Hospitals of Aachen, Bonn, Cologne, and Düsseldorf, as well as the West German Tumor Center (WTZ) at the University Hospitals of Essen and Münster. Together with the North Rhine-Westphalia State Cancer Registry and the Essen-based startup TamedAI, they aim to develop innovative digital solutions that will both improve patient care and strengthen North Rhine-Westphalia as a center for research.
The comprehensive and timely collection of cancer data is a cornerstone of research and care; for this reason, Germany has a legal reporting requirement, which is regulated in North Rhine-Westphalia by the State Cancer Registry Act. Through the systematic documentation of initial diagnoses, relapses, and disease progression, the data is intended to inform conclusions about the effectiveness of therapies and to help develop new approaches to cancer treatment. “Reliable cancer registration is therefore indispensable. With KIMONA, we aim to use modern AI methods in a way that significantly streamlines documentation processes while simultaneously providing high-quality data for research and patient care,” says project coordinator Lars-Thorben Moos from the IT team at CIO Cologne.
Currently, cancer cases in Germany are documented using a standardized “basic oncology dataset.” This data is consolidated from various clinical systems - such as hospital information systems, pathology, or radiology systems - and reported to the state cancer registry. However, much of this information must currently be compiled manually by medical documenters. Analyses show that currently around 62 percent of data fields must be entered manually. Automated interfaces to primary clinical systems currently exist for only about 27 percent of the data fields.
Another problem is that much of the important information is not available in a structured format, but is “hidden” in doctors’ notes or other free-text documents. Analyzing these documents is often a time-consuming and labor-intensive process. This is precisely where the KIMONA project comes in: using modern methods, such as language models, the aim is to automatically analyze medical texts and extract relevant information in the future. “For example, information on diagnoses, treatments, or disease progression can be identified in doctors’ letters and made available in a structured format. That would be a major step toward better data processing,” says Prof. Dr. Michael Hallek, Director of Clinic I for Internal Medicine and the CIO at the University Hospital of Cologne, as well as Executive Board Member and Principal Investigator at the Center for Molecular Medicine Cologne (CMMC). According to estimates, up to 50 percent of the data fields could be extracted from unstructured documents in this way and—after human review—transferred to the cancer registry.
“Comprehensive, high-quality cancer registration is a cornerstone of research and quality assurance in oncology. Projects like KIMONA can help ensure that reports are recorded more quickly, completely, and consistently,” explains Prof. Dr. Andreas Stang, Medical Director of the North Rhine-Westphalia State Cancer Registry. In KIMONA, the relevant medical documents are first analyzed and technical infrastructures are created, such as so-called “data lake systems,” in which clinical data from various sources is consolidated. Building on this, AI models are used to automatically extract the information and prepare it for the cancer registry.
In the long term, the solutions developed in the project are intended not only to be used for cancer registration, but other clinical registries or research databases are also expected to benefit from them.
The KIMONA project is funded with 5.96 million euros as part of the Gesünder.IN.NRW innovation competition by the State of North Rhine-Westphalia / Ministry of Economic Affairs, Industry, Climate Protection, and Energy with funds from the European Union. It has a duration of three years. Coordination is handled by the Center for Integrated Oncology at the University Hospital of Cologne and the Medical Faculty of the University of Cologne. In addition to the North Rhine-Westphalia State Cancer Registry, the project partners include the cancer centers of the six university hospitals in North Rhine-Westphalia, which are organized under the “Network of Excellence in Cancer Medicine”: the Center for Integrated Oncology Aachen-Bonn-Cologne-Düsseldorf, the West German Cancer Center with locations in Essen and Münster, and the Essen-based startup TamedAI.
This message has been translated and modified by the CMMC (K. Heber & D. Grosskopf-Kroiher) and is based on the text by the press and communications teams of the University Hospoital Cologne (original German version here).