A research team from Marburg wants to better assign people with COPD to specific groups and thus enable them to receive customised personalised treatment. This is to be achieved with the help of methods based on machine learning.
More than 65 million people worldwide suffer from COPD and unfortunately there is no way to completely stop or even reverse the course of the disease so far. Another hurdle is that the course of the disease often varies greatly. Concomitant diseases such as diabetes or osteoporosis often play a major role. Depending on the concomitant disease and the patient's state of health, different therapeutic approaches are often necessary.
With the help of a machine learning method, a team from the Philipps University of Marburg would like to develop software that supports doctors in their everyday work when making diagnosis and therapy decisions. The goal is for the programme to automatically classify patients based on their health data and assign them to a group as early as possible to enable optimal treatment.
For the development of the system, the research team uses data from several thousand people with COPD, collected over several years in large patient cohorts. In addition to clinical data on disease progression, it also collects more specific data from laboratory tests to learn even more about the disease.
The software's algorithm learns to recognise matches in different patients and to form groups with similar symptoms and courses based on countless combinations of laboratory data and disease courses. In particular, they want to find out which parameters and data are relevant to the course of the disease and which are not.
In everyday clinical practice, the software is to be installed as part of the normal hospital IT so that it can access the existing patient data. In this way, it could give treating physicians suggestions for diagnosis and therapy right on the spot. For example, whether a concomitant disease should be treated first in order to improve the patient's overall condition.
The researchers want to use the software for the first tests in the clinic as early as next year. In addition to supporting therapy decisions, the researchers hope that the subdivision into patient groups could also help in the development of new drugs. After all, an active substance does not always help all patients equally well.
References:
German Federal Ministry of Education and Research: artificial intelligence improves treatment of COPD. March 2021 Newsletter 101 (Original Title: Bundesministerium für Bildung und Forschung: Künstliche Intelligenz verbessert Behandlung von COPD).