Accurate and early diagnosis of tardive dyskinesia (TD), a serious movement disorder typically caused by exposure to antipsychotic medications, is imperative for long-term management.
“Keeping an eye on things and recognizing it early allows for potential interventions that will prevent it from being permanent or chronic,” said Tsao-Wei Liang, MD, of Thomas Jefferson University in Philadelphia.
Tardive dyskinesia is significantly under-diagnosed, noted Anthony Sterns, PhD, of Miami University in Oxford, Ohio. “There are probably 800,000 to 1.2 million people with TD but only 40,000 receiving the medication that controls it,” he said.
“A key factor contributing to this is that the necessary skills to accurately diagnose TD are not being adequately taught to psychiatrists, who are often the very specialists most likely to prescribe the antipsychotic medications that can lead to this condition,” Sterns told MedPage Today.
To help improve detection, new tools have recently been developed, some of which use artificial intelligence (AI).
Interactive Digital Tool
One interactive digital tool, dubbed Discover TD, became available in 2023. Developed in collaboration with a medical video game company, Discover TD features an online education experience to teach healthcare providers how to distinguish TD from other drug-induced movement disorders such as drug-induced parkinsonism, dystonia, and akathisia. Healthcare providers practice identifying the intricacies of how TD presents with virtual patients.
This can be particularly helpful for providers who may be less experienced in diagnosing and managing TD, said M. Mercedes Perez-Rodriguez, MD, PhD, of the Icahn School of Medicine at Mount Sinai in New York City, in a statement.
“In addition to the patient history, the key to differential diagnosis is knowing what the characteristic movements of different drug-induced movement disorders look like, and this tool helps users distinguish these movements through a simulated physical exam so that they can make a proper diagnosis and management plan,” she added.
New AI Diagnostic Aid
Another tool called TDScreen, which became available in June 2025, is an automated, video-based AI tool developed by Sterns’s company to help screen for and monitor TD symptoms.
While TDScreen isn’t intended as a standalone diagnostic tool, it may help healthcare providers improve early diagnosis, particularly in remote care. It was developed based on the Abnormal Involuntary Movement Scale (AIMS), a clinician-rated scale that assesses involuntary movements like those related to TD.
Using AI video technology, TDScreen assesses and generates a patient’s TD risk score in under 5 minutes, which can help the healthcare provider with clinical decision-making and management strategies. It is intended for patients at-risk for TD taking antipsychotics.
The tool’s role “is as a monitor to trigger a review by a diagnostician,” said Sterns, who led the study testing the technology. “We are working with movement disorder specialists to enhance the diagnostic capabilities by applying additional analysis based on better instruments. There is a role to support patient-centered care for ongoing monitoring of TD once diagnosed and for managing the VMAT2 medications which are given to control TD’s symptoms.”
“Our team was initially focused on improving medication adherence for antipsychotics, and through that work, we came to the conclusion that a crucial component was the need for effective monitoring for TD,” said Sterns. “This realization directly led us to explore innovative methods with our video-based AI partners, and we formed the group to spot TD earlier and expand more widespread detection.”
In studies testing the tool, over 350 adult participants completed a standard AIMS procedure while sitting across from a device on a stand. Patients were asked open-ended questions like “What kind of pet do you like?” The video captured the patient’s face, trunk, and hands while answering, which are three key places to focus when assessing a patient for TD.
This type of visual assessment usually is done by a healthcare provider. Patients with tardive dyskinesia typically present during a clinical exam with some involuntary, abnormal movement, but the presentation can span a variety of different abnormal movements, Liang told MedPage Today.
“The most important thing is to recognize typical features. Look at the lower face first, look carefully,” he said.
While TDScreen does not assess legs and feet directly, it provides a prediction of the total AIMS score using video data of the face, shoulders, trunk, arms, and hands. In the studies, the AI screening tool outperformed human raters for reliably detecting suspected TD.
Ultimately, new tools like this may help healthcare providers diagnose TD. “More importantly, it can spot the condition early and perhaps permanence can be avoided in the best case or at least mitigated,” Sterns said.
Please enable JavaScript to view the