Azernews.Az

by Alimat Aliyeva

Swedish hospitals are increasingly relying on artificial intelligence to automatically transcribe patient admission data, but many doctors warn that the system has serious shortcomings. According to medical staff, the AI struggles to recognize regional dialects—especially the Skåne dialect—leading to frequent errors ranging from simple spelling mistakes to completely nonsensical sentences, Azernews reports.

The automated systems were introduced as a response to staff shortages and were intended to replace manual transcription of medical dictations. However, doctors report that instead of saving time, they are often forced to carefully review and correct the transcripts. This extra workload distracts them from direct patient care and, in some cases, delays clinical decision-making.

Authorities in the Östergötland region acknowledge both the benefits and the limitations of the new technology, emphasizing that digital tools are a necessary part of a broader modernization of the healthcare system. At the same time, they admit that returning to the previous model with dedicated transcription staff is unrealistic due to financial constraints.

Some hospitals are now experimenting with hybrid solutions, where AI handles the initial transcription while specialized staff review complex cases or dialect-heavy recordings. Doctors hope that with better training data and closer collaboration between developers and healthcare workers, the technology will eventually become a real asset rather than an added burden.