A weakly supervised transformer for rare disease diagnosis and subphenotyping from EHRs with pulmonary case studies - Takeaways - MDSpire

A weakly supervised transformer for rare disease diagnosis and subphenotyping from EHRs with pulmonary case studies

  • By

  • Kimberly F. Greco

  • Zongxin Yang

  • Mengyan Li

  • Han Tong

  • Sara Morini Sweet

  • Alon Geva

  • Kenneth D. Mandl

  • Benjamin A. Raby

  • Tianxi Cai

  • February 6, 2026

  • 0 min

Share

  • 1

    Rare diseases affect 300–400 million people globally, yet they remain underdiagnosed due to low prevalence and limited clinician knowledge.

  • 2

    The proposed WEST model utilizes weakly supervised transformers to enhance rare disease diagnosis and subphenotyping from electronic health records.

  • 3

    WEST combines expert-validated labels with probabilistic silver-standard labels to improve model calibration and performance.

  • 4

    Evaluation of WEST on rare pulmonary conditions demonstrates superior performance in phenotype classification and disease progression prediction.

  • 5

    By minimizing manual annotation, WEST facilitates efficient representation learning for accurate rare disease diagnosis from routine EHR data.

Original Source(s)

A weakly supervised transformer for rare disease diagnosis and subphenotyping from EHRs with pulmonary case studies

  • by Kimberly F. Greco, Zongxin Yang, Mengyan Li, Han Tong, Sara Morini Sweet, Alon Geva, Kenneth D. Mandl, Benjamin A. Raby, Tianxi Cai

  • February 6, 2026

Related Content

AI Scribes Lag Clinicians on Note Quality

A VHA study across 11 vendors finds AI-generated primary care notes score lower than clinician-written notes, with the largest deficits in thoroughness, organization, and usefulness

MMF Shows Promise in Juvenile Scleroderma

Mycophenolate mofetil demonstrated comparable disease control with fewer adverse effects.

Cutaneous Rhabdomyosarcoma May Mask Melanoma

Molecular and epigenetic profiling suggested most tumors in a rare cutaneous sarcoma cohort were better classified as transdifferentiated melanoma.