Scholarlink

<p>Rare diseases (RD) affect 3.5–5.9% of the global population, yet patients often endure a diagnostic odyssey averaging around 5–6 years, navigating multiple specialties and referrals. This study investigates the use of information content (IC) as a screening tool to identify RD candidates using SNOMED CT, the clinical terminology standard for the electronic health record (EHR) in the Singapore health system. A longitudinal dataset of 1,274,199 patients (2018–2022) containing 35,898 unique SNOMED terms was analyzed. Rare disease patients (<i>n</i> = 17,575) were identified through SNOMED–Orphanet mappings. IC effectively distinguished RD patient profiles from the first clinical encounter by applying outcome-driven thresholds. The simplicity of implementing the proposed method enables a screening test that can achieve ~95% sensitivity, while maintaining the potential follow-up burden at a reasonable level (20% precision starting from 3 encounters, with an IC threshold of 8.17). This approach also surfaced 71 underdiagnosed rare diseases in the population (57 of genetic origin). To our knowledge, this is the first study to apply information-theoretic metrics to EHR for RD screening, demonstrating its potential as a screening tool at both hospital and health system levels.</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

Information content as a health system screening tool for rare diseases

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月25日

查看全文
  • Tudor Groza,
  • Peter N. Robinson,
  • Weng Khong Lim,
  • Kaavya Narasimhalu,
  • Jenny Hsieh,
  • Khung Keong Yeo,
  • Goh Bee Keow,
  • Terrence Thomas,
  • Tien Yin Wong,
  • Neerja Karnani,
  • Gareth Baynam,
  • Saumya Shekhar Jamuar

摘要

Rare diseases (RD) affect 3.5–5.9% of the global population, yet patients often endure a diagnostic odyssey averaging around 5–6 years, navigating multiple specialties and referrals. This study investigates the use of information content (IC) as a screening tool to identify RD candidates using SNOMED CT, the clinical terminology standard for the electronic health record (EHR) in the Singapore health system. A longitudinal dataset of 1,274,199 patients (2018–2022) containing 35,898 unique SNOMED terms was analyzed. Rare disease patients (n = 17,575) were identified through SNOMED–Orphanet mappings. IC effectively distinguished RD patient profiles from the first clinical encounter by applying outcome-driven thresholds. The simplicity of implementing the proposed method enables a screening test that can achieve ~95% sensitivity, while maintaining the potential follow-up burden at a reasonable level (20% precision starting from 3 encounters, with an IC threshold of 8.17). This approach also surfaced 71 underdiagnosed rare diseases in the population (57 of genetic origin). To our knowledge, this is the first study to apply information-theoretic metrics to EHR for RD screening, demonstrating its potential as a screening tool at both hospital and health system levels.