Objective: To develop an integrated staging methodology for Prodromal and Manifest Huntington’s disease (HD). Background: Once HD is diagnosed clinically (also known as, motor-onset/manifest), HD is categorized into five functional stages by an expert-derived scale (1). However, this classification omits pre-manifest (or, pre-motor onset) HD strata. Since the prodromal stage spans several decades, the inability to stage prodromal patients is a crucial rate-limiting step to earlier clinical trials. Here we describe a novel, agnostic staging model for classification of phenotypes from prodromal and manifest HD patients. Method: Four observational studies (PREDICT-HD, REGISTRY, TRACK-HD&ON, Enroll-HD) are integrated into a database of 19,136 participants with 2079 outcomes recorded longitudinally. We apply machine learning methods (Latent Variable Analysis and Continuous-Time Hidden Markov Models) to this dataset to develop a probabilistic staging model learned through variational Bayes method and EM algorithm. Results: HD progression covering ~36 years (~16 years pre- motor onset, ~10 years in transition, and ~10 years post- motor onset) is classified into nine distinct phenotypes. The annual transition probability between successive phenotypes ranges from 5% – 27% with higher transition probabilities associating with longer polyglutamine repeats. Among the earliest changes (10 – 15 years) prior to motor onset are striatal volume declines and deteriorating cognitive (e.g. SDMT, SCNT and SIT) and motor outcomes (e.g. TMS). Conclusion: HD phenotypes can be segregated into nine groups with varying combinations of motor, cognition and function measurements. Assignment to each of the nine groups predicts future outcome what implies that each of the groups is a diseases stage. Unlike the traditional classification that only considers HD patients post-onset, our model serves to classify a larger portion of HD gene carriers. We find that inter-group progression rates are modulated by CAG lengths with higher CAGs accompanied by higher transition probabilities. These varying transition probabilities between stages can be used to rationally guide the design of future HD clinical trials in premanifest participants. References:  Shoulson I, Fahn S Huntington disease: clinical care and evaluation. Neurology 1979;29:1-3  Ghosh, Soumya et al. AMIA Summits on Translational Science proceedings. 2017, 92-102.  Sun, Zhaonan et al. AMIA Annual Symposium Proceedings. 2017, 1635-1644.  Sun, Zhaonan et al. JAMIA Open. 2019, ooy060.