How can artificial intelligence (AI) be meaningfully integrated in hospitals?
A Bertelsmann Foundation's study on digital health shows that the German healthcare system lags far behind in international comparison in AI integration. Read more to learn how AICURA Platform is solving this problem.
Self-learning algorithms have long since arrived in our everyday lives. Algorithms based on artificial intelligence support us when writing emails, shopping on the internet or searching for information. Nevertheless, studies such as the Bertelsmann Foundation's study on digital health show that the German healthcare system lags far behind in international comparison. Even today, a relevant part of doctors' work consists of collecting patient information from analogue sources and documenting it at great expense. At the same time, medical staff in hospitals are confronted with a large amount of medical information due to new diagnostic procedures and a high workload. This leads not only to an enormous additional workload, but also to ineffective use of the available information. It is estimated that only three per cent of the accumulating medical data is used. The best possible evaluation of the annually accumulating patient data of up to 50 petabytes per hospital can hardly be done by the staff. As in our everyday lives, AI-based applications - AI apps for short - could also provide great relief here. However, the systematic use of AI in hospitals is often hampered by high integration costs and the lack of experience on the part of hospitals in dealing with AI applications. Furthermore, the strict data protection requirements in Germany prevent the comprehensive use of medical data for the development of medical AI apps.
The Berlin-based start-up AICURA Medical aims to solve this problem. Via the AICURA platform, artificial intelligence is brought into clinics and medical data is made usable for (research) and development of new AI apps in compliance with data protection laws. AICURA's vision is to use AI in everyday clinical practice to relieve the burden on medical staff and to enable the best possible care for patients.
To achieve this, the AICURA experts from the fields of informatics, data science, medicine and medical product development have developed a software platform. This consists of two parts: Firstly, the AICURA Operating System (AICURA OS), which is installed directly in the hospital and enables the use of various AI apps through a single integration. And secondly, the AICURA App Store, through which hospitals can select, download, and directly use AI apps. In addition to these functions, which are particularly relevant for hospitals, the AICURA development environment enables researchers and app developers to train, validate and ultimately distribute their AI apps.
The AICURA platform is characterised by its innovative technology. One challenge in the use of AI apps is that enormously large amounts of data are needed to train these apps. To solve this challenge, AICURA uses a technology called Federated Learning. Federated learning enables decentralised learning on distributed data. The big advantage here is that the decentralised learning process is data-protection compliant, as the data does not have to be merged centrally.
The AICURA App Store for hospitals offers AI apps in the areas of business intelligence, medical research, and decision support. For example, together with KMS AG, AICURA has developed an AI app for predicting the individual inpatient length of stay of patients. The so-called LoS app not only makes it possible to predict the length of stay, but also to identify patient characteristics that have an influence on the individual length of stay. The AICURA platform is also in great demand in medical research. In the INALO project, for example, AICURA is currently developing an AI app for optimising false positive alarms in intensive care units in cooperation with a university hospital and various AI researchers. The INALO app will not only relieve nursing staff enormously in the future, but above all enable more targeted intervention in the event of alarms and thus increase patient safety. Other projects include the areas of intelligent optimisation of post-operative follow-up care for cardiac surgery patients as well as AI-based evaluation of cardiological and radiological data.