Using AI to find facial and speech signs of Parkinson's and related disorders

Study on Artificial Intelligence-Based Facial and Speech-Related Patterns in Parkinson's Disease and Their Digital Biomarkers

This project uses AI to analyze facial expressions and speech to try to find digital markers that help diagnose and tell apart Parkinson's disease, PSP, MSA, and healthy people.

Quick facts

• Study type: Observational
• Enrollment: 720 (estimated)
• Ages: 15 Years to 75 Years
• Sex: All
• Sponsor: Beijing Tiantan Hospital (Academic / other)
• Locations: 1 site (Beijing, Beijing Municipality)
• Trial ID: NCT07392411 on ClinicalTrials.gov

What this trial studies

This observational study records participants' facial expressions on video and speech audio to generate datasets for machine-learning analysis. Participants include people with Parkinson's disease across early and later stages, patients with progressive supranuclear palsy or multiple system atrophy, and healthy controls. AI models will be trained to identify patterns in facial and vocal features that may serve as digital biomarkers for diagnosis and differentiation. All data collection is performed in person at a single center with informed consent and standard clinical diagnostic criteria used for group assignment.

Who should consider this trial

Good fit: Adults aged 18–75 with clinically diagnosed Parkinson's disease, PSP, or MSA, or healthy controls, who can give informed consent and agree to facial video and speech recordings are ideal candidates.

Not a fit: People under 18 or over 75, those unable to consent or cooperate with recordings, or those whose face or speech is altered by unrelated conditions (for example recent facial surgery, stroke, or severe dementia) may not receive benefit from this work.

Why it matters

Potential benefit: If successful, the approach could provide noninvasive, low-cost digital tools to support earlier or more accurate diagnosis and differentiation of Parkinsonian disorders.

How similar studies have performed: Previous research using voice and facial analysis in Parkinson's disease has shown promising early results but remains experimental and not yet established as a clinical diagnostic standard.

Eligibility criteria

Inclusion Criteria:

1. Inclusion Criteria for Parkinson's Disease (PD) Group: (1) Diagnostic Criteria: Meet the diagnostic criteria for Parkinson's disease established by the Movement Disorder Society (MDS) in 2015. (2) Age Range: 18-75 years old. (3) Disease Severity: Early-stage PD: Hoehn-Yahr score ≤ 2.5 points. Mid-to-late-stage PD: Hoehn-Yahr score 2.5-5 points. (4) Consent for Data Collection: Willing to undergo facial expression video and speech audio recording. (5) Informed Consent: Signed informed consent form.
2. Inclusion Criteria for Parkinson's Plus Syndromes (MSA-P, PSP) Group: (1) Age Range: 18-75 years old. (2) Diagnostic Criteria: MSA patients must meet the diagnostic criteria established by the \[Chinese Expert Consensus on the Diagnosis of Multiple System Atrophy, 2017\]. PSP patients must meet the diagnostic criteria established by the \[Chinese Clinical Diagnostic Criteria for Progressive Supranuclear Palsy, 2016 Edition\]. (3) Consent for Data Collection: Willing to undergo facial expression video and speech audio recording. (4) Informed Consent: Signed informed consent form.

Exclusion Criteria:

1. History of cerebrovascular disease, head trauma, hydrocephalus, brain tumors, or intracranial surgery.
2. Presence of metal implants, cardiac pacemakers, or other metallic foreign bodies (applies to PD patients).
3. Severe dyskinesia in PD patients that would compromise cooperation with video/audio recording.
4. Mini-Mental State Examination (MMSE) score ≤ 24 points.

Where this trial is running

Beijing, Beijing Municipality

• 首都医科大学附属北京天坛医院 — Beijing, Beijing Municipality, China (Recruiting)

Study contacts

• Study coordinator: Lingyan Ma, M.D.

Conditions Parkinson's Disease; Progressive Supranuclear Palsy; Multiple System Atrophy; Healthy Control