--- title: AI Voice Models for Parkinson's Screening Impacted by Age Confounding description: Voice AI models for Parkinson's screening may be misled by age differences, highlighting the need for rigorous demographic controls in digital health evaluations. image: https://blog.jmir.org/hubfs/StoryTap%20Blog%20-%20Shikhar%20Shukla.png --- [![JMIR Publications logo](https://blog.jmir.org/hs-fs/hubfs/JMIR-Publications_logo_2020_original.png?width=550&height=109&name=JMIR-Publications_logo_2020_original.png "JMIR Publications logo")](http://www.jmirpublications.com) Open main menu Close main menu [Research News](https://blog.jmir.org/tag/research-news) # AI Voice Models for Parkinson's Screening Impacted by Age Confounding ![Liana Ramos, Marketing Associate](https://blog.jmir.org/hs-fs/hubfs/Liana%20Ramos.jpeg?width=45&name=Liana%20Ramos.jpeg) [Liana Ramos, Marketing Associate](https://blog.jmir.org/author/liana-ramos) Published Date: 7 October 2026 9:04 a.m. [Share this blog post on Twitter](https://twitter.com/intent/tweet?text=I+found+this+interesting+blog+post&url=https://blog.jmir.org/ai-voice-models-for-parkinsons-screening-impacted-by-age-confounding) [Share this blog post on Facebook](http://www.facebook.com/share.php?u=https://blog.jmir.org/ai-voice-models-for-parkinsons-screening-impacted-by-age-confounding) [Share this blog post on LinkedIn](http://www.linkedin.com/shareArticle?mini=true&url=https://blog.jmir.org/ai-voice-models-for-parkinsons-screening-impacted-by-age-confounding) ![AI Voice Models for Parkinson's Screening Impacted by Age Confounding](https://blog.jmir.org/hubfs/StoryTap%20Blog%20-%20Shikhar%20Shukla.png) Voice-based artificial intelligence models have generated significant interest for noninvasive screening of neurological conditions like Parkinson's disease and dementia. However, high classification performance in these models may be driven by age differences between study groups rather than true disease signatures. | --- **Key Takeaways** | | --- | | Age Masked as Performance: Age alone predicted Parkinson's status better than an 86.4M-parameter AI model (AUC 0.875 vs 0.843), proving demographic gaps create massive statistical shortcuts. | | True Signal Is Weaker: In age-restricted evaluation (60–80 years), age-only accuracy fell to chance while the AI model retained an AUC of 0.787, confirming a real but far weaker disease-specific vocal signal. | | Brittle Feature Extraction: Models trained on single feature pipelines degraded sharply when tested across preprocessing versions (AUC dropped to 0.618), highlighting vulnerability to technical shifts. --- | Listen to this blog post 1:47 ---   In a [methodological evaluation](https://www.jmir.org/2026/1/e95609/) published in the [Journal of Medical Internet Research](https://www.jmir.org/), lead researcher Shikhar Shukla and colleagues at [Indiana University](https://iu.edu/) and [Emory University](https://www.emory.edu/) audited the Bridge to Artificial Intelligence (Bridge2AI) Voice Dataset v3.0.0 to quantify how demographic imbalances affect deep learning models trained on vocal biomarkers. ### Auditing Bias in AI Voice Biomarkers How reliably can deep learning models distinguish pathological voice changes from normal vocal aging? Because speech characteristics naturally alter with age, comparing older disease cases against younger control groups creates a statistical shortcut that models can exploit. By applying age-restricted evaluations, site controls, and propensity-score matching, the research team demonstrated that while a disease-specific acoustic signal exists, true model performance is significantly lower than unadjusted full-cohort metrics suggest. Discover how rigorous demographic controls and reporting standards impact digital health AI evaluations: - [Read the Full Open-Access Article](https://www.jmir.org/2026/1/e95609) to explore the complete statistical auditing, spectrotemporal analysis, and proposed reporting standards. - [Watch the Author Video](https://www.youtube.com/watch?v=q7HRFhNdrgU) to hear the research team discuss demographic confounding and voice biomarker evaluation. | | In this video, Shikhar Shukla from Indiana University presents a methodological audit of the Bridge2AI Voice Dataset v3.0.0, evaluating voice-based AI models for Parkinson disease and dementia screening. | | | --- | --- | --- |   ### In the Author's Words > *"Before these tools reach a clinic, we need to know they are hearing the disease and not just hearing age."* > > > — Shikhar Shukla, BDS, MS (Indiana University)   --- Shukla S, Naliyatthaliyazchayil P, Gichoya J, Purkayastha S Demographic Confounding in Voice-Based Parkinson Disease Screening: Methodological Analysis of the Bridge2AI Voice Dataset J Med Internet Res 2026;28:e95609 URL: [https://www.jmir.org/2026/1/e95609](https://www.jmir.org/2026/1/e95609) DOI: [10.2196/95609](https://www.jmir.org/2026/1/e95609) --- JMIR Publications is thrilled to announce that the [*JMIR Aging*](https://aging.jmir.org/) team will be attending the [Gerontological Society of America (GSA) Annual Scientific Meeting](https://www.gsa2026.org/) this November 4-7, 2026 in National Harbor, MD for the very first time! If you'll be at GSA, we would love to connect in person. Drop by booth 214 to meet us, discuss your upcoming research, or learn more about publishing open-access with *JMIR Aging*. ![Liana Ramos, Marketing Associate](https://blog.jmir.org/hs-fs/hubfs/Liana%20Ramos.jpeg?width=150&name=Liana%20Ramos.jpeg) #### Liana Ramos, Marketing Associate [Follow me on LinkedIn](https://www.linkedin.com/in/lianaramos/) Liana Ramos is a Marketing Associate at JMIR Publications, dedicated to building data-driven systems that connect scholarly content with the communities that need it most. With five years of experience at the intersection of media, publishing, and audience behaviour, she handles automated lifecycle journeys, author recognition programs, and record-scale digital events for JMIR. Passionate about storytelling and digital strategy, she focuses on making open access digital health research discoverable, scalable, and highly engaging. ### Leave a Comment ## Related Articles [![Remote Cognitive Screening in Parkinson Disease: Validating the ORCA-PD Tool](https://blog.jmir.org/hubfs/StoryTap%20Blog%20-%20Avigail%20Lithwick%20Algon.png)](https://blog.jmir.org/remote-cognitive-screening-in-parkinson-disease-validating-the-orca-pd-tool) [Research News](https://blog.jmir.org/tag/research-news) ### [Remote Cognitive Screening in Parkinson Disease: Validating the ORCA-PD Tool](https://blog.jmir.org/remote-cognitive-screening-in-parkinson-disease-validating-the-orca-pd-tool) Parkinson disease (PD) is predominantly recognized for its motor symptoms, yet covert cognitive decline affects many patients long before dementia occurs. Traditional... ![Liana Ramos, Marketing Associate](https://blog.jmir.org/hs-fs/hubfs/Liana%20Ramos.jpeg?width=45&name=Liana%20Ramos.jpeg) [Liana Ramos, Marketing Associate](https://blog.jmir.org/author/liana-ramos) [Read More](https://blog.jmir.org/remote-cognitive-screening-in-parkinson-disease-validating-the-orca-pd-tool) [![Patient-Powered Progress: Revolutionary Digital Strategies for Optimal Patient Engagement and Healthcare Delivery,](https://blog.jmir.org/hubfs/DCI%20JoPM%20v2.png)](https://blog.jmir.org/unlocking-patient-powered-progress-featuring-leaders-from-the-journal-of-participatory-medicine) [Events](https://blog.jmir.org/tag/events) ### [Unlocking Patient-Powered Progress: Featuring Leaders from the Journal of Participatory Medicine](https://blog.jmir.org/unlocking-patient-powered-progress-featuring-leaders-from-the-journal-of-participatory-medicine) The landscape of healthcare is evolving, and at the forefront of this transformation is the powerful concept of patient engagement. "[Patient-Powered Progress:...](https://www.dcinetwork.org/patients2025) ![JMIR Publications](https://blog.jmir.org/hs-fs/hubfs/jmir_icon_100px.png?width=45&name=jmir_icon_100px.png) [JMIR Publications](https://blog.jmir.org/author/jmir-publications) [Read More](https://blog.jmir.org/unlocking-patient-powered-progress-featuring-leaders-from-the-journal-of-participatory-medicine) #### About - Menu Item One - Menu Item Two - Menu Item Three #### Services - Menu Item One - Menu Item Two - Menu Item Three #### News - Menu Item One - Menu Item Two - Menu Item Three Follow us on Facebook Follow us on LinkedIn Follow us on Twitter Follow us on Instagram ```json { "@context" : "https://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "Liana Ramos, Marketing Associate", "url" : "https://blog.jmir.org/author/liana-ramos" }, "dateModified" : "2026-10-07T13:06:01.846Z", "datePublished" : "2026-10-07T13:04:46.000Z", "headline" : "AI Voice Models for Parkinson's Screening Impacted by Age Confounding", "image" : [ "https://blog.jmir.org/hubfs/StoryTap%20Blog%20-%20Shikhar%20Shukla.png" ], "mainEntityOfPage" : { "@id" : "https://blog.jmir.org/ai-voice-models-for-parkinsons-screening-impacted-by-age-confounding", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject", "url" : "https://blog.jmir.org/hubfs/JMIR-Publications_logo_2020_original.png" }, "name" : "JMIR Publications" } } ```