As we recognize Tardive Dyskinesia (TD) Awareness Week, I’m reflecting on the significant challenges faced by both patients living with TD and the healthcare providers dedicated to their care. TD remains underdiagnosed despite affecting millions of Americans taking antipsychotic medications, with its involuntary movements often mistaken for nervousness, aging, or other conditions. This gap in recognition represents not just a clinical challenge, but a deeply human one that affects quality of life and treatment outcomes.
The Challenge of TD Detection
Tardive Dyskinesia presents unique screening challenges even for experienced clinicians. Traditional AIMS (Abnormal Involuntary Movement Scale) assessments, while valuable, face several real-world limitations:
- Assessments are often rushed in busy clinical settings
- Training and experience among raters is inconsistent
- Unconscious biases can affect evaluations (such as assumptions about which populations are affected)
- Ongoing monitoring is inconsistent due to limited capacity
Our research and work with clinics show that when different people check the same patients for TD symptoms, they often don’t agree on what they see. Even trained clinicians who haven’t been calibrated together achieve a Cohen’s Kappa score of only 0.37 (±0.05), suggesting significant inconsistency. Even after calibration, this improves to just 0.57 (±0.03). In other words, humans are inconsistent even when rating the same patient and symptoms.
This variability in assessment doesn’t just represent a data problem—it directly impacts patient care, potentially delaying crucial interventions that could improve quality of life.
How AI-Powered TD Screening Improves Detection Accuracy
At Videra Health, we’ve developed TDScreen, an AI-powered solution designed to address these challenges head-on. TDScreen uses advanced video analysis technology to evaluate patients for potential TD symptoms quickly and efficiently. Patients can take the screener on their own time, on their own device. The results have been remarkable, with our AI model demonstrating a Cohen’s Kappa of 0.61—outperforming even calibrated human raters.
TDScreen’s performance metrics show its power as a clinical support tool:
- AUC (Area Under Curve): 0.85 (0.83 – 0.97)
- Sensitivity/Specificity = 0.79 (0.70 – 0.93)/0.82 (0.73 – 0.92)
- Precision/Recall = 0.81 (0.73 – 0.92)/0.79 (0.70 – 0.93)
These numbers represent more than just statistical achievements—they translate to real-world benefits:
- Consistency: Unlike human raters whose assessments may vary, AI provides the same evaluation standards every time
- Accessibility: Patients can complete assessments from home on their own devices
- Efficiency: Providers save valuable clinical time while increasing screening frequency
- Earlier detection: Subtle symptoms can be identified before they become pronounced
- Continuous monitoring: Regular assessments track symptom progression or improvement
Moving Beyond Detection to Comprehensive Care
While TDScreen isn’t intended as a standalone diagnostic tool, it represents a significant advancement in meeting the standard of care for TD screening. The American Psychiatric Association guidelines recommend regular monitoring for TD in patients taking antipsychotics, but time constraints often make this challenging to implement consistently.
Our AI-assisted approach helps bridge this gap by making regular screening more feasible, ultimately leading to earlier intervention when symptoms are detected. This approach aligns perfectly with measurement-based care principles, providing objective data to inform treatment decisions.
The Human Element of Technology
It’s important to emphasize that TDScreen and our broader AI platform aren’t about replacing clinician judgment—they’re about enhancing it. By automating routine screenings, we free healthcare providers to focus on what matters most: the human connections and complex decision-making that drive quality care.
One provider using TDScreen shared, “I love it, I had a patient that took the screener, and it picked up on movements I would have missed. It works great!”
This Tardive Dyskinesia Awareness Week, I invite healthcare providers to consider how AI-powered tools like TDScreen can help them meet and exceed the standard of care for TD screening and monitoring. By embracing these innovations, we can collectively improve outcomes for the millions of patients at risk for or living with tardive dyskinesia.
At Videra Health, we’re committed to continuing this important work, combining advanced technology with compassionate care to transform how we understand, detect, and manage TD. Together, we can make significant progress in addressing this challenging condition and improving the lives of those affected by it.
To learn more about TDScreen and to sign up to use TDScreen at no cost in your practice, visit TDScreen.ai