Scholarlink

Background <p>Artificial intelligence (AI)-driven discharge-letter tools aim to reduce documentation burden for clinicians.</p> Objective <p>Map the German-speaking market, compare feature sets and assess real-world feasibility, data security and costs.</p> Methods <p>Structured market scan (Jan–Mar 2025); 14&#xa0;vendors contacted, 4&#xa0;questionnaires, and 4&#xa0;live demos analysed. A&#xa0;26-item feature matrix and directed qualitative content analysis were applied.</p> Results <p>Response rate 28%. Full-stack systems (Atacama, Fraunhofer) covered 65% and 38% of features, producing almost complete letters yet requiring manual approval. Basic transcription tools (Grundig) generated raw text only. No product offered automatic ICD/OPS coding or therapy suggestions. Pricing ranged from € 49 per user/month to six-figure enterprise packages.</p> Conclusion <p>AI-generated letters are technically viable but limited by interoperability gaps, missing reimbursement and strict EU-AI-Act requirements. Hospitals should launch KPI-driven pilots with privacy-by-design to build evidence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

KI-basierte Arztbriefe – Zukunft oder Gefahr?

  • 期刊论文

  • 发表日期：

    2025年12月23日

查看全文
  • Christoph Vogel,
  • David Baur,
  • Moritz Künzel,
  • Stefano Pagano,
  • Yasmin Youssef,
  • Georg Osterhoff

摘要

Background

Artificial intelligence (AI)-driven discharge-letter tools aim to reduce documentation burden for clinicians.

Objective

Map the German-speaking market, compare feature sets and assess real-world feasibility, data security and costs.

Methods

Structured market scan (Jan–Mar 2025); 14 vendors contacted, 4 questionnaires, and 4 live demos analysed. A 26-item feature matrix and directed qualitative content analysis were applied.

Results

Response rate 28%. Full-stack systems (Atacama, Fraunhofer) covered 65% and 38% of features, producing almost complete letters yet requiring manual approval. Basic transcription tools (Grundig) generated raw text only. No product offered automatic ICD/OPS coding or therapy suggestions. Pricing ranged from € 49 per user/month to six-figure enterprise packages.

Conclusion

AI-generated letters are technically viable but limited by interoperability gaps, missing reimbursement and strict EU-AI-Act requirements. Hospitals should launch KPI-driven pilots with privacy-by-design to build evidence.