Poster presented at: American College of Rheumatology Annual Meeting
Location and date: San Diego, CA; November 3-8, 2017
Authors: Atul Deodhar, MD,Cody Garges, MS,Oodaye Shukla, MS,Theresa Arndt, MLS,Tara Grabowsky, MD,Yujin Park, PharmD
- To develop a predictive mathematical model for AS based on features observed in the claims histories of patients with and without a diagnosis of AS, to aid in the earlier identification of AS.
- Predictive models for AS diagnosis were developed and refined in this US administrative claims database.
- Application of machine learning dramatically improves the performance of the model in comparison with traditional linear regression or purely clinical models.
- These predictive models may be vital for a timely diagnosis of AS despite their low PPV.
- Application of these models in a separate claims database and confirmation of AS diagnosis through in-person medical examinations are ongoing for model validation and use in real-world settings.