Digital biomarkers for non-motor symptoms in Parkinson’s disease: the state of the art

Digital biomarkers that remotely monitor symptoms have the potential to revolutionize outcome assessments in future disease-modifying trials in Parkinson’s disease (PD), by allowing objective and recurrent measurement of symptoms and signs collected in the participant’s own living environment. This biomarker field is developing rapidly for assessing the motor features of PD, but the non-motor domain lags behind. Here, we systematically review and assess digital biomarkers under development for measuring non-motor symptoms of PD. We also consider relevant developments outside the PD field. We focus on technological readiness level and evaluate whether the identified digital non-motor biomarkers have potential for measuring disease progression, covering the spectrum from prodromal to advanced disease stages. Furthermore, we provide perspectives for future deployment of these biomarkers in trials. We found that various wearables show high promise for measuring autonomic function, constipation and sleep characteristics, including REM sleep behavior disorder. Biomarkers for neuropsychiatric symptoms are less well-developed, but show increasing accuracy in non-PD populations. Most biomarkers have not been validated for specific use in PD, and their sensitivity to capture disease progression remains untested for prodromal PD where the need for digital progression biomarkers is greatest. External validation in real-world environments and large longitudinal cohorts remains necessary for integrating non-motor biomarkers into research, and ultimately also into daily clinical practice.

Proof of concept, one individual has been tested for one night.A smart pillow is developed to monitor body temperature, humidity, sleep position and duration, and turns.Limitation: data recorded every 5 minutes and all data more granular data within 5 min is lost.Limited accuracy, only one night testing in one patient.

Other developments:
-Nocturnal awakenings and sleep fragmentation were associated with PD progression.

Fatigue
The association between heart rate variability, reaction time, and indicators of workplace fatigue in wildland firefighters 11 n=10 participants in a 14-day period were examined to investigate the relationship between heart rate variability and incidence of fatigue, total sleep time, and reaction time in shift worker by using wrist actigraphy in laboratory setting.
There is significant association between HRV and incidence of fatigue and sleep.Efficient for monitoring daily cardiac function in response to stressful situation.
Smartphone-based gaze as digital biomarker for mental fatigue 12 Proof-of-concept, two mental fatigue checking systems were used, consisting of a language-independent, object-tracking Task system was used for 17 patients and proofreading task was used for 15 patients, data collected in lab.
Mental fatigue was predicted with 80% accuracy by smartphone-measured gaze (as a digital biomarker) by using 75-150 sec of gaze data.

Pain
No supplementary studies.Samsung Gear Sport smartwatch is compared (n=28) with Shimmer3 ECG device in free-living conditions for 24 hours.

Hyperhidrosis
Good correlations with HR and time-domain parameters, whereas LF, HF and other frequencydomain were less consistent and situation-dependent.

Bladder dysfunction
Smartphone App for In-home Uroflowmetry 19 Correlation between standard uroflowmetry and a smartphone app that analyzed urine voiding sounds to calculate flow rate and volume.47 individuals with overactive bladder or outlet obstruction, and 15 healthy controls, all performing ≥10 self-measurements.

Other developments:
In a recent review, a novel framework for the development of state-of-the-art digital biomarkers for erectile dysfunction is proposed.Options include skin temperature, arterial pulse using PPG, radial circumference and rigidity, oxygen saturation.A smart ring implemented with PPG, ECG and Thermistor to detect HR, RR, Spo2 and temperature.Validation against gold standard in-clinic for short period, in n=2 participants.
Proof of concept.Experimentally reported HR & RR had significant correlation with their related standards.But Spo2 had trial-dependent similarities and temperature measurements fell within normal range (correlation up to r=0.85, p<0.05).

Orthostatic Hypotension
Cuffless Blood Pressure Monitoring from an Array of Wrist Bio-Impedance Sensors Using Subject-Specific Regression Models: Proof of Concept 23 Proof of concept, n=10 healthy participants (age 18-30).Post-exercise BP measurement by using low-noise bio-impedance sensing hardware placed on radial and ulnar arteries of the wrist.
Wrist sensor capable of cuffless BP-monitoring, accuracy: Correlation up to 0.86 for systolic BP and 0.77 for diastolic BP.Better accuracy than two other studies with accuracies as high as +-7 and +-5 mmHg for systolic and diastolic blood pressure respectively. 24,25itivity of orthostatic hypotension measurement not investigated.
Smart Vest: wearable multi-parameter remote physiological monitoring system. 16lot study, n = 25 (healthy).Smart T-shirt contains ECG and photoplethysmography waveforms.Data of both sensors are transmitted to a remote physiological monitoring station along with the geo-location of the wearer, and analyzed there.To identify the relationship between cognitive functioning and sleep disturbances n=95 with idiopathic PD and n=48 healthy controls were recruited, to measure nocturnal sleep efficacy they wore actigraphy for 2 weeks.
Sleep efficiency is associated with specific cognitive impairments.Working (r = .28)and verbal (r = .23)memory were significantly associated with sleep efficiency, but verbal fluency and attentional setshifting were not associated.

Other developments:
-Smart homes with multi-room activity sensors show good correlation with clinician-scored cognitive status (r = .72) 33and show moderate correlations with some in-lab tests and scales.

Anxiety
No supplementary studies.

Depressive symptoms
Mobile Phone Detection of Semantic Location and Its Relationship to Depression and Anxiety 41 N = 208, using their mobile phone while their phone sensor data recorded for 6 weeks, to evaluate the relationship between semantic location visit patterns and depression and anxiety.
Low predictive value of semantic location (AUC 0.62).In combination with phone sensor data, AUC increased to 0.88.Depressed Mood Prediction of Elderly People with a Wearable Band 42 Proof-of-concept, n = 14 elderly without history of depression.71 days of measurement using the Empatica E4 wearable band, combining PPG and accelerometry, compared to depression scales.
Mean accuracy is 82.7% for PPG, 76.6% for accelerometry and 76.3% for the combination of PPG and accelerometry.The relationship between mobile phone location sensor data and depressive symptom severity 43 Validation study, n = 48 healthy students over a 10-week period.To investigate the correlation between GPS features/circadian activity and depressive symptoms.Revealing association at baseline, follow-up, and changes in symptoms severity over time.
GPS features can be a reliable predictor for severity of depressive symptoms.Measuring period of 10 weeks.Relatively short to be able to accurately predict onset of depression or depressive episodes.Severity of depressive symptoms is self-reported (PHQ-9).
A Sensor-Driven Visit Detection System in Older Adults Homes: Towards Digital Late-Life Depression Marker Extraction 44 Longitudinal study, n = 13 healthy participants (age 86 +-7.23).Detection of home visits as a predictor for social isolation and late-life depression.
Small sample size, assumption that visits are well associated with common geriatric depression scale screening tool (ρ=-0.89,p=0.001)

Other developments:
-Objective (passive) sleep characteristics in PD are not predictive of depressive symptoms. 45lated HRV measurements has accuracy up to 83% in accuracy for depressive symptom prediction 46 , although circadian rhythm in HRV was not associated with depressive symptoms. 47-Smartwatch-measured heart rate circadian rhythm is associated with depression severity. 48-Combination of passive smartphone use, voice samples, and social media use predicts depressive symptoms with a sensitivity of 0.75 and specificity of 0.79 49 .-More time at home is associated with depressive symptoms in young adults 50 , and geographic location distribution associates with depressive symptoms in bipolar disorder. 51

-
No clear to low associations between sleep parameters, circadian rhythm and mood symptoms 48,52,53 , although one recent study demonstrates predictive value of actigraphy-measured sleep and circadian rhythm disturbance on depressive symptoms in a longitudinal study. 54

Olfactory dysfunction
No supplementary studies.

Hallucinations
No supplementary studies.

Color vision disturbances
No supplementary studies.

Prodromal symptom Methods Validity, reliability and feasibility and longitudinal measurements
3Daytime napping on actigraphy does not correlate with subjective EDS severity as measured by Epworth Sleepiness Scale 9,10 -Excessive daytime sleepiness is associated with shows significant EDS based on MSL 4 -Frequency of daytime naps was associated with PD relative to controls but was not progressive over multi-year follow-up.3 3-Increased sleep latency and reduced efficiency and REM sleep were associated with PD. 4 RBD Towards a handy screening tool for REM sleep behavior disorder: RDBAct algorithm from wrist actigraphy data 5 Early validation study of automatic REM sleep without atonia (RSWA) detection.25 patients with PD (underwent video-PSG with bilateral wrist actigraphy.31 video recordings of RSWA and 18 without RSWA were used for training and testing.Wrist actigraphy had AUC of 0.67 in detecting RSWA.Other developments: -Expert-based visual interpretation of actigraphy outperformed quantitative actigraphy analyses. 6-Low sensitivity (20%) of REM sleep behavior disorder detection by actigraphy can be increased by adding an REM sleep behavior disorder questionnaire. 7-Number of wake bouts is higher in people with PD with REM sleep behavior disorder compared to people with without REM sleep behavior disorder. 8Excessive daytime sleepiness (EDS) Other developments:

20 Skin impedance
Validation study, N=10 healthy adults.Ingestible capsule containing motion sensors (SmartPill®).Measuring parameters of colonic motility and transit times.Capsule is validated against gold-standard scintigraphy.Capsule is sensitive to opioid effect on gut transit time.R=0.95 for scintigraphy vs. pill.More convenient and less invasive than scintigraphy.Difficult for individuals with dysphagia.
40-AUC of ≥0.93 for predicting MCI In multi-room environment over a period of 24 weeks.35-In-house vitoring can differentiate Alzheimer's disease from healthy controls 36 -Computer use and more sleep were associated with better cognition.37,whereasmore time in kitchen and looking in fridge and cabinets is associated with worse cognition. 38Higher latency in using punctuation and backspace keys was associated with worse processing speed between people with multiple sclerosis, but was not sensitive to progression over a 1-year period within participants. 39Rest-activity rhythms associate with cognitive function during morning and afternoon activity.40