re:publica: AI is already providing support in hospitals and the care sector

Contents

AI in care: The influence of artificial intelligence is growing, particularly in the hospital sector. The development has not yet penetrated the outpatient sector as much. The critical points in care are well known: There are very few staff and little money, but different facilities and ever new requirements. In addition, data that has recently been lost or suddenly appeared in the public domain has caused scandals; this also makes digitalization and the use of AI more difficult.

At the re:publica session"Artificial intelligence in nursing - what are the challenges and special features?", nurse and nursing scientist Kathrin Seibert, Professor of Machine Learning at the Berlin University of Applied Sciences Felix Biessmann and moderator Lea Bergmann from the Association for Digitization in the Social Economy (vediso) discussed the topic.

Good data basis, more difficult for care services

A good three years ago, the Federal Ministry of Education and Research (BMBF) published a guideline for funding research and development in the field of "Making repositories and AI systems usable in everyday care". According to the guideline, the focus is on "innovative artificial intelligence solutions for the field of care applications. Primarily software-based innovative applications of artificial intelligence are promoted in two modules with the aim of supporting caregivers and caregiving relatives as well as improving the self-determination and quality of life of people in need of care."

It became clear during the session that this is not so easy: the funded projects deal with the question of how to build up a good database in the first place – certainly easier in hospitals, but how is this supposed to work with the large number of different care services? And in families: around four million people in Germany have a care degree and are cared for at home, says Kathrin Seibert.

Digital technologies in use on numerous levels

Others are making provisions and reorganizing their homes. "We can also consider care technologies for this area," says Seibert, "but we are not currently doing so. We are currently concentrating on nursing homes, day care, care services, hospitals and care provided by relatives."

Nevertheless, digital technologies are already being used at all levels: In collaboration between professional groups, she lists, "be it in a clinic, be it between departments, be it between different service providers in the large area of control and administration, or in the area of training, further education and training". And as far as the area of "direct interaction" is concerned, this is "not widespread in Germany, but it is widespread"; the most prominent example is systems for detecting falls, such as mats with sensors that recognize - and report - whether someone has just tripped, whether there is a water stain or whether a person is lying down.

"AI remains with doctors in the medical context"

Voice assistants are increasingly being used for care documentation, especially in outpatient care, she explains. "I speak documentation information into the mobile device and don't have to worry about what happens to it: The system sorts it to the right person. And I can also access information from the documentation by voice." Intelligent duty and route planning systems are now also being used. The next item on her list is likely to be more controversial: social robots, such as the artificial harp seal Paaro.

Felix Biessmann confirms that AI is already being used, especially in hospitals, but that it also remains with doctors in a medical context. "Care is not yet being taken into account as much," hence the BMBF program. Language assistance systems have improved. But we also know "that these models all have problems with hallucinations or generally poor predictions - and I think that's a problem." Solvable? "You just need AI researchers to work on it in a transdisciplinary way together with care scientists. However, this is made more difficult by the fact that the infrastructure and model development processes are still very slow," he complains. "There is simply a lack of infrastructure in small care facilities or even in outpatient settings to try things out." And then there is the issue of data security.

There is a tension between security and usefulness: many technical solutions are already available, but to put them into practice, you need established processes that, unlike in hospitals, do not yet exist in the care sector. "I don't think this has been thought through to the end."

This session was part of the "Health" focus at re:publica 24, which was supported by the pharmaceutical company Pfizer Germany.

(olb)