---
title: "Real-time human–AI platform links EEG and smartphone to comanage epilepsy"
id: "nature-2-human-artificial-intelligence-collaboration-for-comanaging-epilepsy-in-real-time"
canonical_url: "https://medichelpline.com/clinical-feed/nature-2-human-artificial-intelligence-collaboration-for-comanaging-epilepsy-in-real-time"
content_type: "clinical_feed_article"
specialty: "Neurology"
source_name: "Nature Medicine"
source_url: "https://www.nature.com/articles/s41591-026-04680-4"
published_at: "2026-10-13T10:14:13.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Real-time human–AI platform links EEG and smartphone to comanage epilepsy
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/nature-2-human-artificial-intelligence-collaboration-for-comanaging-epilepsy-in-real-time
- **Specialty:** [Neurology](https://medichelpline.com/clinical-feed/neurology.md)
- **Primary Source:** Nature Medicine
- **Source URL:** [Original Journal Publication](https://www.nature.com/articles/s41591-026-04680-4)
- **Published At:** 2026-10-13T10:14:13.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- Investigators developed an inpatient prototype that connects scalp and intracranial **EEG** to a secure, large language model–based **AI** and a smartphone app for bidirectional patient communication. - The system processes biomarkers including sleep, medication adherence, cognition and **seizure** probability and notifies patients of events and trends. - The platform records patient-reported symptoms and responses; those responses were used to retrain the seizure detector and reduce false alarms in simulation. - The team validated the prototype prospectively in 13 adult epilepsy monitoring unit inpatients (scalp EEG n = 11; intracranial EEG n = 2) without changing standard-of-care treatment. - Investigators ran a pseudoprospective simulation showing that patient feedback alone improved detector precision, without expert intervention. - Design goals included intelligible presentation of sensor data, actionable answers tailored to patient function, low-latency conversational surveys, use of responses for annotation and retraining, governance of output safety and truthfulness, and clinician review pathways. - The study was single-center and observational; participants were encouraged to interact with the platform during continuous video-EEG monitoring. - The source did not report full technical specifications, detailed algorithms, quantitative performance metrics beyond the qualitative improvement claim, or long-term outpatient testing results. - The authors propose the system as a scalable, safeguarded platform adaptable across epilepsy and other conditions, but further validation in diverse ambulatory cohorts and regulatory review are implied next steps.
## Clinical Analysis & Structured Key Points
## Your privacy, your choice We use essential cookies to make sure the site can function. We also use optional cookies for advertising, personalisation of content, usage analysis, and social media, as well as to allow video information to be shared for both marketing, analytics and editorial purposes. By accepting optional cookies, you consent to the processing of your personal data - including transfers to third parties. Some third parties are outside of the European Economic Area, with varying standards of data protection. See our [privacy policy](https://www.nature.com/info/privacy) for more information on the use of your personal data. Manage preferences for further information and to change your choices. Accept all cookies Reject optional cookies [Skip to main content](https://www.nature.com/articles/s41591-026-04680-4#content) Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Advertisement [ ![Nature Medicine](https://media.springernature.com/full/nature-cms/uploads/product/nm/header-95e59e63930e5d6009bad2c23a42ab2d.svg) ](https://www.nature.com/nm) * [ View all journals ](https://www.nature.com/siteindex) * [ Saved research ](https://www.nature.com/saved-research) * [ Search ](javascript:;) ## Search Search articles by subject, keyword or author Show results from All journals This journal Search [ Advanced search ](https://www.nature.com/search/advanced) ### Quick links * [Explore articles by subject](https://www.nature.com/subjects) * [Find a job](https://www.nature.com/naturecareers) * [Guide to authors](https://www.nature.com/authors/index.html) * [Editorial policies](https://www.nature.com/authors/editorial_policies/) * [Log in](https://idp.nature.com/auth/personal/springernature?redirect_uri=https://www.nature.com/articles/s41591-026-04680-4) * [ Content Explore content ](javascript:;) ## Explore content * [ Research articles ](https://www.nature.com/nm/research-articles) * [ Reviews & Analysis ](https://www.nature.com/nm/reviews-and-analysis) * [ News & Comment ](https://www.nature.com/nm/news-and-comment) * [ Podcasts ](https://www.nature.com/nm/podcast) * [ Current issue ](https://www.nature.com/nm/current-issue) * [ Collections ](https://www.nature.com/nm/collections) * [Follow us on Facebook ](https://www.facebook.com/Nature-Medicine-193691346949/) * [Follow us on X ](https://twitter.com/naturemedicine) * [Sign up for alerts ](https://journal-alerts.springernature.com/subscribe?journal_id=41591) * [ RSS feed ](https://www.nature.com/nm.rss) * [ About the journal ](javascript:;) ## About the journal * [ Aims & Scope ](https://www.nature.com/nm/aims) * [ Journal Information ](https://www.nature.com/nm/journal-information) * [ Editors ](https://www.nature.com/nm/editors) * [ Journal Metrics ](https://www.nature.com/nm/journal-impact) * [ Research Cross-Journal Editorial Team ](https://www.nature.com/nm/research-cross-journal-editorial-team) * [ Reviews Cross-Journal Editorial Team ](https://www.nature.com/nm/reviews-cross-journal-editorial-team) * [ Statistical Advisory Panel ](https://www.nature.com/nm/statistics-advisory-panel) * [ Our publishing models ](https://www.nature.com/nm/our-publishing-models) * [ Editorial Values Statement ](https://www.nature.com/nm/editorial-values-statement) * [ Editorial Policies ](https://www.nature.com/nm/editorial-policies) * [ Content Types ](https://www.nature.com/nm/content) * [ Web Feeds ](https://www.nature.com/nm/web-feeds) * [ Contact ](https://www.nature.com/nm/contact) * [ Publish with us ](javascript:;) ## Publish with us * [ Submission Guidelines ](https://www.nature.com/nm/submission-guidelines) * [ For Reviewers ](https://www.nature.com/nm/for-reviewers) * [ Language editing services ](https://authorservices.springernature.com/go/sn/?utm_source=For+Authors&utm_medium=Website_Nature&utm_campaign=Platform+Experimentation+2022&utm_id=PE2022) * [Open access funding](https://www.nature.com/nm/open-access-funding) * [Submit manuscript ](https://mts-nmed.nature.com/cgi-bin/main.plex) * [ Sign up for alerts ](https://journal-alerts.springernature.com/subscribe?journal_id=41591) * [ RSS feed ](https://www.nature.com/nm.rss) 1. [nature](https://www.nature.com/) 2. [nature medicine](https://www.nature.com/nm) 3. [articles](https://www.nature.com/nm/articles?type=article) 4. article Human–artificial intelligence collaboration for comanaging epilepsy in real time [ Download PDF ](https://www.nature.com/articles/s41591-026-04680-4.pdf) [ Download PDF ](https://www.nature.com/articles/s41591-026-04680-4.pdf) * Article * [Open access](https://www.springernature.com/gp/open-science/about/the-fundamentals-of-open-access-and-open-research) * Published: 13 October 2026 # Human–artificial intelligence collaboration for comanaging epilepsy in real time * [Zack Goldblum](https://www.nature.com/articles/s41591-026-04680-4#auth-Zack-Goldblum-Aff1-Aff2-Aff3) [ORCID: orcid.org/0000-0003-2803-7051](https://orcid.org/0000-0003-2803-7051)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2),[3](https://www.nature.com/articles/s41591-026-04680-4#Aff3) [na1](https://www.nature.com/articles/s41591-026-04680-4#na1), * [Haoer Shi](https://www.nature.com/articles/s41591-026-04680-4#auth-Haoer-Shi-Aff1-Aff2) [ORCID: orcid.org/0009-0000-5273-8371](https://orcid.org/0009-0000-5273-8371)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2) [na1](https://www.nature.com/articles/s41591-026-04680-4#na1), * [Zhongchuan Xu](https://www.nature.com/articles/s41591-026-04680-4#auth-Zhongchuan-Xu-Aff1-Aff2)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2), * [William K. S. Ojemann](https://www.nature.com/articles/s41591-026-04680-4#auth-William_K__S_-Ojemann-Aff1-Aff2)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2), * [Carlos A. Aguila](https://www.nature.com/articles/s41591-026-04680-4#auth-Carlos_A_-Aguila-Aff1-Aff2)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2), * [Kevin Long](https://www.nature.com/articles/s41591-026-04680-4#auth-Kevin-Long-Aff4) [ORCID: orcid.org/0000-0001-8881-2481](https://orcid.org/0000-0001-8881-2481)[4](https://www.nature.com/articles/s41591-026-04680-4#Aff4), * [Kevin Xie](https://www.nature.com/articles/s41591-026-04680-4#auth-Kevin-Xie-Aff1-Aff5-Aff6)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[5](https://www.nature.com/articles/s41591-026-04680-4#Aff5),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6), * [Kerry C. Nix](https://www.nature.com/articles/s41591-026-04680-4#auth-Kerry_C_-Nix-Aff1-Aff7)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[7](https://www.nature.com/articles/s41591-026-04680-4#Aff7), * [Katherine G. Walsh](https://www.nature.com/articles/s41591-026-04680-4#auth-Katherine_G_-Walsh-Aff1-Aff2)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2), * [Ellie Chang](https://www.nature.com/articles/s41591-026-04680-4#auth-Ellie-Chang-Aff1-Aff2)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2), * [Sarah B. Lavelle](https://www.nature.com/articles/s41591-026-04680-4#auth-Sarah_B_-Lavelle-Aff1)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1), * [Brandon Bach](https://www.nature.com/articles/s41591-026-04680-4#auth-Brandon-Bach-Aff1)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1), * [Kathryn A. Davis](https://www.nature.com/articles/s41591-026-04680-4#auth-Kathryn_A_-Davis-Aff1-Aff6) [ORCID: orcid.org/0000-0002-7020-6480](https://orcid.org/0000-0002-7020-6480)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6), * [Nishant Sinha](https://www.nature.com/articles/s41591-026-04680-4#auth-Nishant-Sinha-Aff1-Aff6-Aff8) [ORCID: orcid.org/0000-0002-2090-4889](https://orcid.org/0000-0002-2090-4889)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6),[8](https://www.nature.com/articles/s41591-026-04680-4#Aff8), * [Lauren H. Hammer](https://www.nature.com/articles/s41591-026-04680-4#auth-Lauren_H_-Hammer-Aff1-Aff2-Aff6) [ORCID: orcid.org/0000-0002-4045-490X](https://orcid.org/0000-0002-4045-490X)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6), * [Erin C. Conrad](https://www.nature.com/articles/s41591-026-04680-4#auth-Erin_C_-Conrad-Aff1-Aff6)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6) & * … * [Brian Litt](https://www.nature.com/articles/s41591-026-04680-4#auth-Brian-Litt-Aff1-Aff2-Aff3-Aff6) [ORCID: orcid.org/0000-0003-2732-6927](https://orcid.org/0000-0003-2732-6927)[1](https://www.nature.com/articles/s41591-026-04680-4#Aff1),[2](https://www.nature.com/articles/s41591-026-04680-4#Aff2),[3](https://www.nature.com/articles/s41591-026-04680-4#Aff3),[6](https://www.nature.com/articles/s41591-026-04680-4#Aff6) Show authors [_Nature Medicine_](https://www.nature.com/nm) (2026) [Cite this article](https://www.nature.com/articles/s41591-026-04680-4#citeas) [ Save article ](https://www.nature.com/articles/s41591-026-04680-4/save-research?_csrf=Oab6zKWLdhtBukuF6x3jhYEmj3b87qMu) [ View saved research ](https://www.nature.com/saved-research) ## Abstract One-third of the world’s 70 million people with epilepsy have seizures uncontrolled by medication, and implantable devices are a promising treatment. These devices can detect impending seizures, missed medication and impaired cognition, but, like most medical devices, they cannot share this information with patients in real time. This is a missed opportunity for devices to learn from clinical events and teach patients how their behaviors affect their health. Here we present a device platform that converses with patients to help them manage their epilepsy. The inpatient prototype links scalp and intracranial electroencephalography (EEG) to a secure, large language model-based artificial intelligence (AI) that communicates bidirectionally through a smartphone app. The AI processes biomarkers of sleep, medication, cognition and seizure probability, notifies patients of clinical events and trends, records symptoms and responses, and uses their feedback to retrain its seizure detector, reducing false alarms. We validated this platform in 13 patients who are undergoing video-EEG monitoring for epilepsy (scalp, _n_ = 11; intracranial, _n_ = 2). In a pseudoprospective simulation, seizure detectors improved precision from patient feedback alone, without expert intervention. We propose this system as a scalable, safeguarded platform for medical devices that can rapidly adapt to patient and provider needs across epilepsy and other medical conditions. ### Explore related subjects Discover the latest articles and news in related subjects. * [Biomedical engineering](https://www.nature.com/subjects/biomedical-engineering) * [Epilepsy](https://www.nature.com/subjects/epilepsy) * [Translational research](https://www.nature.com/subjects/translational-research) ## Main More than one-third of the world’s 70 million people with epilepsy continue to have seizures despite aggressive treatment with medications. Some are candidates for functional surgery (removing or laser-ablating seizure-generating brain regions), but only half will become seizure free after two years[1](https://www.nature.com/articles/s41591-026-04680-4#ref-CR1 "Jette, N. & Wiebe, S. Update on the surgical treatment of epilepsy. Curr. Opin. Neurol. 26, 201–207 \(2013\)."). Implantable devices are an exciting, less invasive alternative that improve seizure control without surgery’s most severe side effects, such as cognitive impairment or mood disorders. However, seizure freedom is rare with devices. Even for the two-thirds of patients with medication-responsive epilepsy, living ‘seizure free’ but reliant on medication presents ongoing challenges. These individuals are at risk for seizure relapses triggered by common life events—missed medications, sleep deprivation, pregnancy or taking commonly prescribed medications that increase seizure risk. There is a need for technologies that help people live with epilepsy, because clinical treatment often fails to provide complete seizure freedom. FDA approved only three devices to treat epilepsy. Whether open loop (stimulating in fixed cycles agnostic to neural activity) or closed loop (stimulating in response to preprogrammed detection criteria), they reduce seizures gradually, plateauing at a median seizure reduction of up to 75% 9 years after deployment in the most positive studies[2](https://www.nature.com/articles/s41591-026-04680-4#ref-CR2 "Nair, D. R. et al. Nine-year prospective efficacy and safety of brain-responsive neurostimulation for focal epilepsy. Neurology 95, e1244–e1256 \(2020\)."). Roughly 30% patients improve minimally or not at all. Although some devices can store and transmit small amounts of data asynchronously, none share it in real time. This is a missed opportunity to present useful information to patients that the device ‘knows’ but has no way to convey. This could include changes in seizure risk, the presence of known seizure precipitants or detecting specific patient behaviors that may be either beneficial (for example, getting more sleep) or detrimental (for example, consuming alcohol or some medications) to controlling seizures. Similarly, current devices prevent patients from configuring them to their needs and lifestyle; they lack mechanisms to richly annotate data and cannot efficiently retrain algorithms with minimal physician interaction. New minimally invasive electroencephalography (EEG) devices are poised to dramatically change seizure management with their ability to continuously collect data from patients during normal daily life. These devices vary from small, battery-powered single-channel scalp motes to subgaleal implants and wearable devices[3](https://www.nature.com/articles/s41591-026-04680-4#ref-CR3 "Rehman, M., Higdon, L. M. & Sperling, M. R. Long-term home EEG recording: wearable and implantable devices. J. Clin. Neurophysiol. 41, 200–206 \(2024\)."). Some devices can transmit continuous EEG data to smartphones and subsequently to the cloud[4](https://www.nature.com/articles/s41591-026-04680-4#ref-CR4 "Halliday, A. J. et al. The UMPIRE study: a first-in-human multicenter trial of bilateral subscalp monitoring for epileptic seizure detection. Epilepsia 66, 3426–3439 \(2025\)."), where the data are processed and viewable by clinicians. Wearable devices monitor movement, heart rate and other peripheral parameters to alert families to convulsive seizures. Yet none offer bidirectional interfaces that let patients communicate and interact with them in real time. To our knowledge, no prior literature describes interactive epilepsy devices that communicate bidirectionally with patients or their specifications. Early work instead focuses on warning of impending seizures and forecasting seizure-free periods[5](https://www.nature.com/articles/s41591-026-04680-4#ref-CR5 "Snyder, D. E., Echauz, J., Grimes, D. B. & Litt, B. The statistics of a practical seizure warning system. J. Neural. Eng. 5, 392–401 \(2008\)."),[6](https://www.nature.com/articles/s41591-026-04680-4#ref-CR6 "Cook, M. J. et al. Prediction of seizure likelihood with a long-term, implanted seizure advisory system in patients with drug-resistant epilepsy: a first-in-man study. Lancet Neurol. 12, 563–571 \(2013\)."). Grounding our design in the published literature on the lived experience and stated priorities of people with epilepsy and their caregivers[7](https://www.nature.com/articles/s41591-026-04680-4#ref-CR7 "Mitchell, J. W., Doherty, J., Batchelor, R., Noble, A. & Marson, A. What matters most to people living with epilepsy? A rapid review of qualitative research relating to health outcomes. Epilepsia 66, 4122–4138 \(2025\)."),[8](https://www.nature.com/articles/s41591-026-04680-4#ref-CR8 "Hill, C. E., Vanderboll, K., Parent, J. M., Skolarus, L. E. & Zahuranec, D. B. Health priorities and treatment preferences of adults with epilepsy: a narrative literature review with a systematic search. Epilepsy Behav. 166, 110359 \(2025\)."),[9](https://www.nature.com/articles/s41591-026-04680-4#ref-CR9 "Rapport, F., Clement, C., Doel, M. A. & Hutchings, H. A. Qualitative research and its methods in epilepsy: contributing to an understanding of patients’ lived experiences of the disease. Epilepsy Behav. 45, 94–100 \(2015\)."),[10](https://www.nature.com/articles/s41591-026-04680-4#ref-CR10 "Camarillo, I. A., Misiewicz, S., Siegel, K. & Ottman, R. The concept of ‘control’ in people with epilepsy: a qualitative study. Epilepsy Behav. 161, 110059 \(2024\)."),[11](https://www.nature.com/articles/s41591-026-04680-4#ref-CR11 "Boele, F. et al. Health-related quality of life and unmet needs of people with epilepsy and their family caregivers: a systematic scoping review. Epilepsy Behav. 171, 110601 \(2025\)."), together with our group’s clinical experience, we propose that an optimal interactive epilepsy device should (1) present sensor-derived information (here scalp or intracranial EEG (iEEG)) intelligibly; (2) answer patient queries with actionable text and figures matched to their level of function; (3) provide an intuitive, low-latency conversational interface for surveys and two-way voice or text; (4) use patient responses to annotate data and retrain adaptive algorithms; (5) govern the appropriateness, truthfulness, safety and security of its output; and (6) let experts efficiently review and annotate data for patient care and to improve performance. In the following sections, we describe a platform that performs these functions and report results from prospective testing on 13 patients admitted to the Hospital of the University of Pennsylvania epilepsy monitoring unit (EMU) for continuous video and scalp (_n_ = 11) or iEEG (_n_ = 2) monitoring. ## Results ### Patient recruitment We conducted a prospective, single-center observational study to evaluate a real-time patient–artificial intelligence (AI) interface for clinical utility and usability (see [Methods](https://www.nature.com/articles/s41591-026-04680-4#Sec15) for detailed study procedures). Thirteen adult EMU inpatients undergoing continuous EEG monitoring to characterize seizures for presurgical evaluation were enrolled (scalp EEG, _n_ = 11; iEEG, _n_ = 2; Table [1](https://www.nature.com/articles/s41591-026-04680-4#Tab1) and Supplementary Fig. [1](https://www.nature.com/articles/s41591-026-04680-4#MOESM1)) without altering standard-of-care treatment. Throughout their EMU stay, participants were encouraged to interact with the platform and provide feedback on their experience. **Table 1 Patient demographics (_n_ = 13)** [Full size table](https://www.nature.com/articles/s41591-026-04680-4/tables/1) ### Device platform The platform comprises the following three systems: (1) software that extra
## Related Clinical Research

- [Scalp EEG Signatures of Anterior Thalamic DBS: Reproducible Gamma Enhancement](https://medichelpline.com/clinical-feed/biorxiv-0-acute-scalp-eeg-signatures-of-anterior-thalamic-dbs-in-epilepsy-reproducible.md)
- [AI-Assisted Auto-Categorization Cuts Response Time for Patient Portal Messages in Neurology](https://medichelpline.com/clinical-feed/medrxiv-14-ai-tools-to-assist-with-management-of-patient-portal-messages-auto.md)
- [NED-Net: iterative annotation-training for rodent EEG seizure detection](https://medichelpline.com/clinical-feed/biorxiv-18-ned-net-an-annotation-training-loop-for-rodent-eeg-event-detection.md)
- [Intraoperative functional ultrasound distinguishes seizure-onset cortex and tracks responses to re](https://medichelpline.com/clinical-feed/medrxiv-13-interictal-functional-ultrasound-imaging-in-human-epilepsy-resection-and.md)
- [Safety of Stroboscopic Light Stimulation: Observational Evidence and Guidelines](https://medichelpline.com/clinical-feed/plos-one-23-stroboscopic-light-stimulation-safety-within-and-beyond-laboratory-settings.md)

## Navigation
- [← Back to Neurology Feed](https://medichelpline.com/clinical-feed/neurology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.