Add clinical target columns (UPDRS, MoCA, cognitive composite, PRS)
Adds derived clinical targets for the existing 1000 scans, joined from PPMI LONI study data. No imaging is touched.
What lands
clinical/ppmi_mini_clinical.parquet— a ~130 KB sidecar table, keyed onsample_id, withparticipant_idalongside for subject-level grouping. Load it and align onsample_id.README.md— a new Clinical targets section defining every column: instrument, range, direction of scoring, coverage, and a load snippet.
What's in it
Motor and cognitive instruments as _baseline (nearest the scan) and _slope_48m (annualized OLS slope over a −0.25 to 4.0 year window) pairs: MDS-UPDRS Part III (OFF-only and all-states), Hoehn & Yahr, MDS-UPDRS Part II, Schwab & England, MoCA, a five-test cognitive composite, and DAT-SPECT striatal binding ratio for striatum, putamen and caudate.
Cross-sectional columns: UPSIT, RBDSQ, CSF alpha-synuclein SAA status, and two PD polygenic risk scores.
Three recruitment covariates: enrolled_genetic_cohort, enrolled_hyposmia, enrolled_rbd. Not targets — PPMI recruits genetic carriers through a dedicated cohort and that membership correlates with PD polygenic risk at r=0.58 by construction, so anything that might vary with recruitment can be checked against them.
The join is on PATNO, which participant_id already encodes as sub-<PATNO>, so it hits 1000/1000 and PATNO itself is not published.
Full column definitions are in the dataset card rather than here, so they stay with the data.