MAPPING EPIDEMIOLOGICAL EVIDENCE AND PRIMARY MUTATION LANDSCAPE OF LEBER HEREDITARY OPTIC NEUROPATHY USING AI-ENABLED EVIDENCE SYNTHESIS
Author(s)
Diwyashri Govindarajaperumal, B.Pharm1, Angeline Babitha Dhas, BS2, Viji Queen V, Sr., PharmD3, Revanth M, B.E.1, Swathirajan C R, Ph.D1, George Alisha, B.E.1, Meghan Oates-Zalesky, MSc4.
1MadeAi, Nagercoil, India, 2MadeAi, Cambridge, MA, USA, 3MadeAI, Nagercoil, India, 4Chief Marketing Officer, MadeAi, Cambridge, MA, USA.
1MadeAi, Nagercoil, India, 2MadeAi, Cambridge, MA, USA, 3MadeAI, Nagercoil, India, 4Chief Marketing Officer, MadeAi, Cambridge, MA, USA.
OBJECTIVES: To evaluate Leber hereditary optic neuropathy (LHON) prevalence, incidence, and primary mitochondrial mutation distribution using MadeAi™, an AI-powered literature review platform, in synthesizing fragmented epidemiological evidence.
METHODS: A dual-reviewer, double-blinded SLR (human vs. AI) was conducted in PubMed and Embase using MadeAi platform to identify studies reporting LHON epidemiology and primary mtDNA mutations. AI and human screening decisions, extraction, and synthesis were compared.
RESULTS: We included 29 studies comprising 15,824 patients across 19 countries. Asia contributed 55.2% of studies (16/29; China 34.5%, India 13.8%), followed by Europe (34.5%, 10/29). LHON prevalence ranged from 0.19 (Serbia) to 3.44 (India) per 100,000 persons (median: 1.93/100,000). Male prevalence exceeded females (median: 2.40 vs. 0.43/100,000). Incidence data was exclusively available from Japan (median: 0.07/100,000).m.11778G>A variant was the predominant mutation overall (median: 59.3%), proving more prevalent in Asian versus European cohorts (median: 76.3% vs. 56.0%). Conversely, m.3460G>A skewed heavily toward European compared to Asian cohorts (median: 15.0% vs. 3.8%), while m.14484T>C remained uniform across continents (median: 7.2%).AI-assisted title/abstract and full-text screening demonstrated 86% and 83% accuracy, respectively, matching human performance. Overall, AI workflow reduced total review time by 67% across all stages.
CONCLUSIONS: LHON remains a rare disease with substantial geographic variation in prevalence and mutation distribution. These findings highlight the importance of region-specific epidemiological evidence to support genetic counseling, disease awareness, and healthcare planning. Successful application of AI towards SLR enabled rapid synthesis of fragmented global evidence while maintaining reviewer-level accuracy, demonstrating its potential to accelerate epidemiological research in rare diseases where data are dispersed across diverse sources.
METHODS: A dual-reviewer, double-blinded SLR (human vs. AI) was conducted in PubMed and Embase using MadeAi platform to identify studies reporting LHON epidemiology and primary mtDNA mutations. AI and human screening decisions, extraction, and synthesis were compared.
RESULTS: We included 29 studies comprising 15,824 patients across 19 countries. Asia contributed 55.2% of studies (16/29; China 34.5%, India 13.8%), followed by Europe (34.5%, 10/29). LHON prevalence ranged from 0.19 (Serbia) to 3.44 (India) per 100,000 persons (median: 1.93/100,000). Male prevalence exceeded females (median: 2.40 vs. 0.43/100,000). Incidence data was exclusively available from Japan (median: 0.07/100,000).m.11778G>A variant was the predominant mutation overall (median: 59.3%), proving more prevalent in Asian versus European cohorts (median: 76.3% vs. 56.0%). Conversely, m.3460G>A skewed heavily toward European compared to Asian cohorts (median: 15.0% vs. 3.8%), while m.14484T>C remained uniform across continents (median: 7.2%).AI-assisted title/abstract and full-text screening demonstrated 86% and 83% accuracy, respectively, matching human performance. Overall, AI workflow reduced total review time by 67% across all stages.
CONCLUSIONS: LHON remains a rare disease with substantial geographic variation in prevalence and mutation distribution. These findings highlight the importance of region-specific epidemiological evidence to support genetic counseling, disease awareness, and healthcare planning. Successful application of AI towards SLR enabled rapid synthesis of fragmented global evidence while maintaining reviewer-level accuracy, demonstrating its potential to accelerate epidemiological research in rare diseases where data are dispersed across diverse sources.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR271
Topic
Methodological & Statistical Research, Real World Data & Information Systems, Study Approaches
Topic Subcategory
Artificial Intelligence, Machine Learning, Predictive Analytics
Disease
Genetic, Regenerative & Curative Therapies, Rare & Orphan Diseases, Sensory System Disorders (Ear, Eye, Dental, Skin)