Dario Antweiler, Data Scientist at Fraunhofer IAIS and expert in Healthcare Analytics, explains how Artificial Intelligence is already changing patient care and medical diagnosis, the opportunities and risks AI technologies hold in medicine, and what caregivers should consider when dealing with Artificial Intelligence.
The influence of Artificial Intelligence in medicine and care
Artificial Intelligence (AI) can assist medical professionals in their daily work and support patients during treatment. AI systems already help detect dangers early, such as the risk of pressure ulcers in bedridden patients. They also simplify care documentation, for example, through intelligent wound documentation via photo apps and speech recognition during medical history intake. In one of our research projects, we are developing intelligent protocol creation through AI speech recognition for trauma rooms, where severely injured patients are treated (“TraumAInterface”). In the “SmartHospital.NRW” project, several AI applications for hospitals are being developed, including an automated doctor’s letter generator. Medical diagnoses are also now supported by AI, where AI systems automatically highlight image areas with significant findings in radiological images and provide diagnosis suggestions to the medical staff. Additionally, AI can help in many non-medical processes in healthcare, such as optimizing staff scheduling or supply chain analysis in logistics. As a result, many areas of diagnostics, care, and therapy for patients are changing. On the one hand, AI technologies can reduce the workload by offering good suggestions and useful hints, as well as automating organizational tasks (e.g., documentation), preventing misunderstandings and errors, and on the other hand, enabling new personalized therapies. Furthermore, the digitization will change the tasks of staff, and entirely new job profiles with technical skills will emerge.
Opportunities and risks of AI use in medicine
In the medical field, society can initially benefit from Artificial Intelligence by ensuring high-quality and personalized therapy and care for all patients. Due to demographic changes and increasing life expectancy, patient numbers are steadily rising, and there is a shortage of personnel. A meaningful integration of technologies into the daily work of medical and caregiving staff can open up more time for direct treatment and care, as well as save costs in the healthcare system. An important aspect of integrating AI into medical processes is the early involvement of users in the development process. Furthermore, systematic biases in the underlying data must be recognized and avoided. For example, the data should be balanced by gender, as it is becoming clear how diseases and therapies can differ between men and women. Trained AI models can be specifically tested for fairness for this purpose. Finally, blind trust in technologies (automation bias) must be avoided through transparent explanations and interaction with AI systems. Otherwise, this could lead us to trust an automatic recommendation more than is appropriate.
What should be considered for the individuals receiving care?
Patients should be encouraged to actively participate in the development of AI applications. AI systems require intensive feedback from the people who will ultimately interact with the applications. This actively reduces abstract fears that still exist in society regarding terms like “algorithm” and “Artificial Intelligence.” Successful deployment requires informed trust between humans and machines. This is ensured through transparent education by medical personnel and a basic understanding of digital technologies. Therefore, broad societal offerings on so-called “digital health education” should be introduced, where people are informed about the pros and cons and enabled to make independent assessments. Additionally, medical staff must be intensively trained and prepared to answer questions from patients.
