B1646
Title: Statistical modelling for drivers and mediators of the characteristic rest tremor of Parkinson's disease
Authors: Kieran Baker - King's College London (United Kingdom) [presenting]
Chianna Umamahesan - Kings College London (United Kingdom)
Clive Weller - Kings College London (United Kingdom)
Sylvia Dobbs - Kings College London (United Kingdom)
John Dobbs - Kings College London (United Kingdom)
Andre Charlett - Public Health England (United Kingdom)
Steven Gilmour - KCL (United Kingdom)
Abstract: A global subjective assessment score (MDS-UPDRS) is commonly used clinically to quantify the severity of Parkinson's disease (PD). It is the sum of subscores based on an assessment of a range of features, including tremor. Accelerometers mounted on finger pulp have been used to objectively measure tremor, over the frequency range 3-14 Hz, at rest. Clinicians identified three key metrics, partitioned by frequency band, for tremor analysis: displacement of a device by tremor, total duration of tremor and number of pulses over a 20 second period. Available numerical integration schemes were compared to determine the best for converting acceleration signals to displacement. MDS-UPDRS rest tremor subscores have been evaluated against the corresponding displacement continuous-scale measurement. The discriminatory power of the key tremor metrics for PD-status was tested using logistic regression, with modelling process guided by a clinical understanding of PD. Whilst size of displacement discriminated for PD, the 4-7 Hz rest tremor pattern (characteristic in PD) was also evident in those not diagnosed with PD, potentially quantifying distance down-the-pathway to PD. This aligns with the long pre-presentation state of PD. Statistical and machine learning models will be used to map potential drivers and mediators (faecal metabolome, intestinal inflammation/barrier disfunction, immunome) of tremor.
