BIG DATA NATURAL LANGUAGE PROCESSING RESEARCH RETROSPECTIVE ELECTRONIC MINING MACHINE LEARNING TEXT MEDICAL STATISTICS RECORDS INTERPRETABLE CLASSIFICATION NARRATIVE ACCURACY CLINICAL DIAGNOSIS

Advantages

Our approach using NLP offers advantages over traditional methods of cohort identification, such as structured data or keyword searching. For many of our prior queries, ICD-9/10 codes, CPT codes, or e-codes do not exist. With this approach and using software and support from our team, users at BCH have completed many successful research projects:

Publication/project Patient cohort Case fraction missed using ICD-9/10
Hennely et al. Penetrating palate injury 83%
Aprahamian et al. Glass thermometer injury ICD-9/10 does not exist
Deanehan et al First time knee monoarthritis 26%
Kimia et al. First complex febrile seizure 48%
Rudloe et al. Peri-orbital cellulitis 33%
Nigrovic et al Lumbar puncture, excluding VP shunt and diabetes ICD-9/10 does not exist
Hodges et al. Dacryocystitis 54%
Hudgins et al. Lymphangitic streaking 69%
Miller et al Misplaced ET tubes after intubation ICD-9/10 does not exist
Lessenich et al Patients evaluated for intussusception 41%
Capraro et al. Unapproved abbreviations in text ICD-9/10 does not exist

Contact:

Amir Kimia, MD

Solution Architect, co-founder

Associate Physician in Medicine, Boston Children's Hospital

Assistant Professor of Pediatrics, Harvard Medical School

Email: amir.kimia@childrens.harvard.edu

Phone: 617-355-6624