Recent research has highlighted the potential of artificial intelligence (AI) in enhancing pediatric diagnostic accuracy, particularly for rare diseases. A study published in Pediatric Investigation on March 25, 2026, reveals that advanced AI models outperform pediatric clinicians in diagnosing rare conditions, while a combined human-AI approach delivers the highest level of success overall. This study, led by Dr. Cristian Launes of Hospital Sant Joan de Déu in Barcelona, Spain, underscores AI’s capability as a complementary tool in improving diagnostic precision and patient outcomes.
Pediatric diagnosis can be especially complex due to the subtle or overlapping symptoms of rare diseases, which often delays treatment and increases the risk of complications. Most prior studies on AI in healthcare utilized simplified or curated cases rather than authentic clinical data, leaving a gap in understanding AI’s real-world performance. Addressing this, Dr. Launes and his team evaluated AI models using genuine pediatric clinical cases, comparing the performance of four advanced language models with 78 pediatric clinicians across 50 cases.
The findings showed that AI models exhibited superior diagnostic accuracy compared to clinicians, particularly in rare disease cases where AI was adept at identifying correct diagnoses missed by human experts. Clinicians, however, demonstrated strengths in complex or context-dependent scenarios, highlighting the differing approaches to diagnostic reasoning between humans and AI. The study did not conduct a real-time human-plus-AI diagnostic workflow but estimated potential complementarity using a “union” approach. This method, which combined the top-5 diagnostic suggestions of both AI and clinicians, achieved a 94.3% accuracy, suggesting the value of AI as a second opinion, especially in challenging cases.
Incorporating AI into clinical practice requires careful governance, as medical diagnostic systems are considered high-risk applications under the European Union AI Act. The study emphasizes that AI should serve as an advisory tool within a framework of clear accountability and oversight to mitigate risks like variability and misleading outputs. The research also noted that the inclusion of additional clinical information, such as lab or imaging results, improved diagnostic performance for both AI and clinicians, emphasizing the importance of continuous clinical assessment.
Overall, this study demonstrates the promise of AI in pediatric healthcare, particularly in diagnosing rare diseases. By integrating AI into clinical workflows, healthcare systems can potentially foster more collaborative and data-driven decision-making. While challenges such as response variability and the need for robust oversight remain, the findings point to a promising future where AI acts as a supportive tool, enhancing the capabilities of human expertise in pediatric care.