KIDOHE - AI-supported documentation for midwives

ZeTeM Logo ohne Text

Researchers: Pascal Fernsel, Hannes Albers
Project funding: Europäische Union
Project sponsor: Bremer Aufbaubank (BAB), funding no FUE0631B
Partners: Karen Güttler, atacama blooms GmbH & Co. KG, Bremen
Duration: 01.05.2020 - 31.01.2022


No other event brings people and the healthcare system together in such a gratifying way as pregnancy and birth. However, obstetrics as part of the health care system has found itself under severe pressure in Germany in recent years. On the one hand, there is a shortage of midwives, and on the other, there are too few delivery rooms and birthing centers - while at the same time the number of births continues to rise (from 678,000 births in 2010 to 785,000 births in 2017). By law, the presence of a midwife or maternity nurse is required for every birth in Germany, regardless of where and how it takes place.

Overall, the number of midwives has increased slightly in the last ten years, but more than half of them work part-time. The reasons for this are overwork, low remuneration and a high risk of recourse. The latter has led to a sharp rise in liability premiums in recent years, with the result that fewer and fewer midwives are performing obstetric work on a freelance basis. Midwives employed in hospitals are relieved of liability premiums, but complain about a considerable workload.

KIDOHE aims to improve the burden and recourse situation of midwives by means of an innovative, intelligent, decision-support system. The atacama blooms GmbH & Co. KG plans to develop such a system in cooperation with the University of Bremen. The system will represent both scientifically based expert knowledge and the midwives' empirical knowledge in networks (e.g. semantic networks, Bayesian networks or neural networks). The technical work goals are divided into two work areas: (1) Representing expert knowledge in networks and (2) Linking continuous data (CTG) with Boolean and categorical data (patient record, knowledge base) using neural networks (deep learning).