Spontaneous brink rate is an important digital biomarker

Abstract

The therapeutic outcome measure that reflects the efficacy of anti–parkinsonism medications is The Movement Disorder Society (MDS)–Sponsored Revision of the Unified Parkinson's Disease Rating Scale (MDS–UPDRS) Part III score. However, this can only be assessed in the clinic, and inter–rater variability can occur. Since optimization of dopamine replacement therapy leads to long–term improvement of patients' ADLs and maintenance of quality of life, the development of biomarkers to determine treatment efficacy is an urgent issue. Blinking is controlled by a neural circuit mediated by the basal ganglia. There are three types of blinking : voluntary blinking, spontaneous blinking, and reflex blinking. Spontaneous blinking is the involuntary blinking that occurs during arousal, the so–called “blink”. The spontaneous eye blink rate (sEBR) decreases after dopamine antagonists and increases after dopamine agonists. In patients with Parkinson disease (PD), sEBR is decreased compared to normal controls, but dopamine administration increases sEBR to the normal range. It has also been reported that plasma L-dopa concentration correlates with blink frequency in PD patients. We investigated whether blink information could be a biomarker for PD symptoms by focusing on the sEBR of 20 advanced PD patients with wear-off phenomenon (UMIN:000044246). Fluctuations in blood concentration of L-dopa correlated with fluctuations in blink rate. We constructed a model to estimate clinical symptoms of PD by machine learning using parameters such as duration and interval based on blink information. The model was successful in predicting in real time the presence or absence of dyskinesia and on/off symptoms in the patient diary from the blink information. Blink has the potential to provide a simple, non-invasive digital biomarker that reflects PD symptoms in real time, and further validation is required.