JMIR Publications

ABSTRACT

Background:

Effective doctor–patient communication is pivotal for quality health care and patient outcomes. However, many patients struggle to understand complex medical language in documents like referral letters or discharge summaries. In Germany, over half of adults have limited health literacy, impeding their comprehension of health information. Short consultation times (often under 10 minutes on average) further constrain physicians’ ability to explain details. Consequently, patients often feel insecure, resort to internet searches, or fail to follow medical advice, potentially harming health outcomes. Advances in artificial intelligence (AI) now enable automated translation of medical jargon into layperson-friendly language. This offers a novel approach to bridging communication gaps.

Objective:

The AI-INFOCARE and AI-MEDTALK trials will evaluate whether providing patients with an AI-generated, easy-to-understand translation of their medical documents (referral letters, discharge summaries) before clinical consultations improves doctor–patient communication. Key outcomes include patients’ perceived quality of communication, autonomy support, understanding of information, physicians’ perception of encounter difficulty, and consultation length.

Methods:

We designed two single-center, parallel-group randomized controlled trials at a university neurosurgery clinic. AI-INFOCARE will enroll 300 adult outpatients before a planned treatment or procedure, and AI-MEDTALK will enroll 300 adult inpatients at discharge after surgery. Patients are randomized 1:1 to receive either (a) an AI-generated German-language summary of their medical report in lay terms (intervention) or (b) usual care with standard documents only (control). The proprietary NLP system (Claude 3.5-based, Simply Onno GmbH) generates the summaries, which are medically reviewed for accuracy. Participants will then attend their scheduled doctor consultation. Outcomes will be assessed immediately post-visit using validated questionnaires: the Health Care Climate Questionnaire (HCCQ) for perceived clinician autonomy support, the Questionnaire on Quality of Physician–Patient Interaction (QQPPI/“FAPI”) for communication quality, a 10-item Difficult Doctor–Patient Relationship Questionnaire (DDPRQ-10) for physician-rated encounter difficulty, the patient’s self-rated understanding (5-point Likert), and the consultation duration recorded in minutes. Inclusion criteria are broad (adults ≥18, German-speaking, able to consent, with a relevant medical document). Key exclusion criteria include cognitive impairment or language barriers preventing questionnaire completion. With 150 patients per arm in each trial (N=300 each), the sample size provides >80% power to detect modest improvements (effect size d≈0.4) in primary outcomes at α=0.05. Recruitment will be consecutive. Randomization is computer-generated; blinding is not feasible due to the nature of the intervention. Ethics approvals were obtained in Feb 2025 (AI-INFOCARE: BOff(Mono)-EK-89022025; AI-MEDTALK: BOff(Mono)-EK-91022025), and both trials are registered (German Clinical Trials Register: DRKS00036810, DRKS00036814).

Results:

Recruitment starts on May 25, 2025 and will continue for 6 months. As of submission (May 2025), no participants have been enrolled and no results are yet available. We will report outcome data after trial completion. We hypothesize that providing layperson-friendly summaries will significantly improve patients’ understanding and satisfaction with information, foster a more autonomy-supportive communication climate, and reduce physicians’ perceived difficulty in the encounter, without unduly prolonging consultation time.

Conclusions:

These two RCTs will be the first to rigorously test an AI-based intervention for enhancing doctor–patient communication in real clinical settings. By focusing on patient comprehension of medical documents, the trials address a modifiable barrier to effective communication. If successful, this approach could be scaled to improve health literacy and patient engagement, leading to better care experiences and potentially improved health outcomes. The findings will inform best practices for integrating AI-generated patient-centered communication tools into routine care. Clinical Trial: Deutsches Register Klinischer Studien (DRKS): DRKS00036810 (AI-INFOCARE), DRKS00036814 (AI-MEDTALK).