Objective: Use a novel virtual reality (VR) protocol to evaluate for potential digital biomarkers of visual hallucinations (VH) in Parkinson’s Disease (PD).

Background: VH are common in PD, impacting 20-50% of patients (1,4). VH in PD patients is associated with cognitive decline (2) and increased risk for nursing home placement (3). Despite this, VH are often underdiagnosed, as diagnosis often relies on subjective recall (4).

Method: Ten patients with PD were studied: 6 with VH in the last year (PDVH) and 4 without (PDnoVH), defined by questionnaire. Demographic and clinical information was collected [table 1]. Patients described controlled scenes using VR. Eye tracking data was analyzed in R studio. Speech was converted to text using Amazon Web Services. Transcripts were analyzed in R using natural language processing which counted total nouns used. Independent sample t-tests compared number of fixations, number of objects described, and MoCA scores between PDVH and PDnoVH (normal distribution assumed) [figures1-3]. Multivariate regression was used to evaluate above dependent variables holding MoCA constant.

Results: PDnoVH used more nouns (M=267.25) than PDVH (M= 195.83), though this was not statistically significant (p = 0.123). There were more fixations in PDnoVH (M=551.75) compared to PDVH (M=525.50), also not statistically significant (p = 0.345). MoCA was higher in PDnoVH (M= 28.5) than PDVH (M=25.7) (p < 0.05). When holding MoCA constant, there was still a non-significant, inverse relationship between VH+ and number of fixations (p= 0.128), but not between presence of VH and number of objects (p=0.608).

Conclusion: While difference in verbal output and fixation between PDnoVH and PDVH is not statistically significant in this population size, there is still a difference, suggesting the existence of a potential digital biomarker for VH.   Future studies should evaluate additional features of language and visual system function in larger populations. PDVH patients had lower MoCA scores than PDnoVH (p = 0.01). Considering this, there is a high likelihood for multicollinearity of MoCA scores and VH+ on fixation and verbal output; although the weak relationship between fixation and VH+, even when accounting for MoCA, suggests this could be a feature of VH in PD irrespective of cognition. Larger studies are needed to further assess.

Table 1: Demographic and Clinical Information

Figure 1: Number of Fixations in PDVH vs PDnoVH

Figure 2: Objects Identified by PDVH vs PDnoVH

Figure 3: MoCA Scores in PDVH vs PDnoVH

References: 1. Eversfield, Charlotte Louise, and Llwyd David Orton. “Auditory and visual hallucination prevalence in Parkinson’s disease and dementia with Lewy bodies: a systematic review and meta-analysis.” Psychological Medicine 49.14 (2019): 2342-2353.

2. Santos-García, Diego, et al. “Risk of cognitive impairment in patients with Parkinson’s disease with visual hallucinations and subjective cognitive complaints.” Journal of clinical neurology (Seoul, Korea) 19.4 (2023): 344.

3. Goetz CG, Stebbins GT: Mortality and hallucinations in nursing home patients with advanced Parkinson’s disease. Neurology 1995, 45(4):669-671.

4. Gibson G, Mottram PG, Burn DJ, Hindle JV, Landau S, Samuel M, Hurt CS, Brown RG, M Wilson KC. Frequency, prevalence, incidence and risk factors associated with visual hallucinations in a sample of patients with Parkinson’s disease: a longitudinal 4-year study. Int J Geriatr Psychiatry. 2013 Jun;28(6):626-31. doi: 10.1002/gps.3869. Epub 2012 Aug 28. PMID: 22927195.

« Back to 2025 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/virtual-reality-to-assess-for-digital-biomarkers-of-visual-hallucinations-in-parkinsons-disease/