Related Experiment Video
Updated: Aug 6, 2026
Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
Published on: June 13, 2025
Artificial intelligence for pediatric rare disease diagnosis: a multimethod study integrating published evidence and
Jungang Zhao1,2, Jiawei Luo1,2, Qiu Li3
1Chevidence Lab Child & Adolescent Health, Department of Pediatric Research Institute, Ministry of Education Key Laboratory of Child Development and Disorders, National Clinical Research Center for Children and Adolescents' Health and Diseases, Children's Hospital of Chongqing Medical University, Chongqing, China.
Insights
Artificial intelligence (AI) shows promise for diagnosing rare pediatric diseases, but implementation faces challenges. Bridging the gap between research metrics and clinical needs is crucial for AI adoption in pediatric rare disease diagnosis.
Area of Science:
- Medical Informatics
- Artificial Intelligence in Medicine
- Rare Disease Diagnosis
Background:
- Pediatric rare diseases present diagnostic challenges due to heterogeneity, often leading to delays and significant patient/family burden.
- While AI, including large language models (LLMs), offers diagnostic support potential, its real-world pediatric application is limited by a gap between research metrics and clinical implementation needs.
- This study addresses the need for an implementation-oriented framework for AI-assisted pediatric rare disease diagnosis by integrating evidence and clinician perspectives.
Purpose of the Study:
- To develop an implementation-oriented evidence-to-requirements framework for AI-assisted diagnosis of pediatric rare diseases.
- To identify and bridge the translational gap between AI research findings and the practical needs of pediatric clinicians.
- To inform future development and evaluation of AI tools for rare disease diagnosis in children.
Main Methods:
- A convergent multimethod design combining a PRISMA-ScR scoping review of 28 studies on AI-assisted pediatric rare disease diagnosis (PubMed, Embase, Web of Science, Scopus, up to Dec 2025) and semi-structured interviews with 21 pediatric clinicians.
- Inductive thematic analysis of interview transcripts to identify clinician perspectives on AI in diagnosis.
- Integration of scoping review findings and interview themes to identify convergences, divergences, and translational gaps.
Main Results:
- Scoping review indicates a trend towards multimodal and LLM-enabled AI for tasks like screening, phenotyping, and gene prioritization, but evidence for translation is uneven, with limited prospective evaluation and inconsistent reporting on fairness, safety, and context.
- Clinician interviews revealed AI is viewed as a cognitive extender, trust hinges on traceable evidence and transparent reasoning, and there's a demand for structured, actionable outputs with minimal workflow disruption.
- An implementation gap persists between metric-focused research reporting and clinician requirements for practical adoption of AI in diagnosing pediatric rare diseases.
Conclusions:
- This study identifies key implementation requirements for AI-assisted pediatric rare disease diagnosis, including representative multicenter data, prospective validation, evidence traceability, actionability, safety, fairness, and workflow integration.
- The findings guide future AI development and evaluation by highlighting the need to align technological advancements with the practical demands of clinical settings.
- Limitations include a focus on English-language studies, reliance on broad rare disease terminology, and a single-institution sample for clinician interviews.
Background:
Pediatric rare diseases are highly heterogeneous and are frequently associated with missed or delayed diagnosis, creating substantial burden for patients, families, and clinicians. Although artificial intelligence (AI), including large language model-enabled approaches, has shown potential for diagnostic support, translation into real-world pediatric care remains limited. A key gap is the mismatch between metric-centric evidence reporting and clinician-defined implementation needs. To address this, we integrated published evidence with clinician perspectives to derive an implementation-oriented evidence-to-requirements framework for AI-assisted pediatric rare-disease diagnosis.
Methods:
We used a convergent multimethod design with two complementary evidence sources. First, we conducted a PRISMA-ScR scoping review of four databases (PubMed, Embase, Web of Science, and Scopus) from inception to December 2025 and included 28 original studies on AI-assisted pediatric rare-disease diagnosis. Second, we conducted semi-structured interviews with 21 pediatric clinicians from 15 departments at a tertiary children's hospital in Chongqing, China, and analyzed transcripts using inductive thematic analysis. We then integrated findings side-by-side to identify convergences, divergences, and translational gaps.
Results:
The scoping review showed rapid movement toward multimodal and LLM-enabled approaches across several diagnostic task types, including screening or cohort identification, phenotyping, differential diagnostic support, and variant or gene prioritization. Translation-oriented evidence remained uneven, with limited prospective evaluation and inconsistent reporting of fairness, safety, and deployment context. Interview analysis identified four recurrent themes: diagnosis as time-pressured puzzle-solving; AI as a cognitive extender rather than replacement; trust dependent on traceable evidence and transparent reasoning; and demand for structured, actionable outputs with low workflow burden. Integrated analysis revealed a persistent implementation gap between metric-centric publication practices and clinician-defined requirements for real-world adoption.
Conclusions:
This scoping review and qualitative interview study does not establish clinical effectiveness of AI-assisted diagnosis. Instead, it identifies implementation requirements that may guide future development and evaluation, including representative multicenter data, prospective validation, evidence traceability, actionability, safety, fairness, and workflow fit. Main limitations include restriction to English-language studies, reliance on umbrella rare-disease terminology, possible missed studies among unscreened records after ASReview-assisted screening, and a single-institution clinician interview sample.