Innov Clin Neur. 2025;22(10–12):78–85.

November 2–5, 2025 in Boston, Massachusetts.

Artificial intelligence (AI)/machine-based learning

Digital phenotyping with real-world data to identify candidates for anti-CGRP migraine treatment

Presenters: Giorgio Pietro Biondetti; Jigar Bandaria; Joseph Zabinski; Carl Marci; and Costas Boussios

Introduction: Migraines affect approximately 15 percent of the total US population, but fewer than 20 percent of patients are correctly diagnosed and fewer use preventive treatment. This is a novel artificial intelligence (AI)-based phenotyping method applied to real-world data (RWD) to identify anti-calcitonin gene-related peptide (CGRP) treatment-naïve patients clinically similar to those being treated to assess whether digital phenotyping can identify patients in need.

Methods: This study used a real-world United States (US) dataset comprising linked claims and EHR data (OM1, Inc.) from May 1, 2018, to July 1, 2022. All patients were aged 18 years or older and diagnosed with migraine or hemicrania. Patients with a qualifying anti-CGRP medication constituted the positive cohort and those remaining constituted the negative cohort. This led to 220,977 patients in the positive and 435,293 in the negative cohort. An AI digital phenotyping platform (OM1 PhenOM®) was calibrated to identify treatment-positive patients based on all available data at index by isolating common shared characteristics.

Results: The AI model identified treatment-positive patients with an area under the curve (AUC) of 0.82. Factors comprising the treatment profile included history of medications to manage migraine; documentation of chronic migraine; and prior exposure to antidepressant medications.

Conclusion: This study applied a digital phenotyping model to large real-world datasets and showed that a distinct profile from migraine patients before anti-CGRP treatment could be used to find a “treatment phenotype” in patients not exposed to anti-CGRP medications. This novel study shows that digital phenotyping can be useful in identifying treatment phenotypes in RWD.

Funding/financial disclosures: Not provided.

Using real-world data to map the application of GLP-1 therapies across multiple diseases

Presenter: Gen Li, PhD

Affiliation: Dr. Li is president and founder of Phesi.

Objective: To map the use of glucagon-like peptide-1 (GLP-1) therapies using real-world data to understand the potential for a single intervention to address multiple related conditions. This could have wide implications for trial sponsors and clinicians, impacting the treatment and prevention of disease.

Methods: The analysis was undertaken using real-world data from Phesi’s AI-powered Trial Accelerator™. A Digital Patient Profile was constructed by analyzing 1,896,194 patients about to be treated by GLP-1 inhibitors over 20 years, providing a precise understanding of GLP-1 patient characteristics including age, sex, comorbidities, outcome measures, and concomitant medications.   

Results: Analysis of 670 clinical trials shows more than 100 diseases are being studied using GLP-1 therapies. This includes trials directly evaluating GLP-1s as interventions, as well as those exploring comorbidities and modulatory effects. GLP-1s began use in treating diabetes, moved into obesity, and interest is now gaining momentum as a broader modulatory pathway across multiple indications including cardiovascular disease, polycystic ovary disease, osteoarthritis and Alzheimer’s disease. Patients were found to be presenting overlapping conditions, such as hypercholesterolemia and hyperlipidaemia, suggesting GLP-1s are increasingly used to target systemic disease clusters rather than single indications.

Conclusion: The industry is in the early stages of understanding disease convergence, and this analysis shows potential for patient benefits to extend far beyond obesity, opening doors to a new era of multidisease prevention and care. It highlights the need for organizations across the biopharmaceutical and healthcare ecosystem to work together, using the growing source of real-world data to gain a better understanding of the relationship between different diseases.

Funding/financial disclosures: There are no funding and/or presenter conflicts related to the content of this poster.

Harnessing conversational AI and methodological rigor to accelerate connected sensor selection in neuroscience trials

Presenter: Megan Parisi

Affiliation: Ms. Parisi is with Syneos Health.

Introduction: Connected sensors can provide objective, continuous data in neuroscience clinical trials, but selecting the right sensor is complex and resource intensive. Although frameworks and guidance exist to support this process, stakeholders still face a crowded sensor landscape, incomplete evidence, uncertainty around regulatory acceptability, operational challenges, and limited access to information. The objective of this work is to develop and evaluate a conversational artificial intelligence (AI)-based tool to support the early stages of systematic sensor selection.

Methods: The conceptual model integrates authoritative data sources, including regulator digital health technology guidance, Digital Medicine Society’s (DiMe) V3 framework, peer-reviewed publications, databases, internal sensor catalogs, and a published aligned framework for clinical outcome assessment (COA) selection. These sources will be curated into a structured knowledge base and embedded into the AI tool interface enriched with contextual metadata such as country acceptability, validation status, technical specifications, and regulatory considerations. Users will query the system in natural language, while structured ontologies and validation steps supported by guardrails and human oversight ensure reliability.

Results (theoretical): The AI tool is expected to generate a structured table of candidate sensors aligned to trial needs, offering a reproducible starting point for evaluation. Outputs are intended for triage and shortlisting, not final selection, and will support expert review and informed discussions with technology providers.

Conclusion: By combining conversational AI with methodological rigor, regulatory alignment, and safeguards, this approach could broaden access to evidence-based sensor selection, reduce manual burden, and accelerate design of patient-centered neuroscience trials pending real-world validation.

Funding/financial disclosures: No funding was received. The author declares no conflicts of interest.

Leveraging AI to accelerate clinical data review: a comparative study of automated discrepancy detection vs. traditional methods

Presenters: Matthew Purri, PhD; Amit Patel; and Erik Deurell, MD

Affiliations: All authors are with Octozi in New York, New York.

Introduction: This study evaluated whether AI-powered clinical discrepancy detection algorithms could enhance data review efficiency and accuracy compared to traditional spreadsheet methods in clinical trials. Clinical data cleaning represents a critical bottleneck throughout trial lifecycles, particularly during interim analyses and database lock periods. Patients often present complex medical histories, comorbidities, and medication regimens requiring careful review. Current manual methods using spreadsheets are inadequate for handling this complexity and prone to human error. Poor data cleaning practices can delay treatment development and compromise safety by missing critical signals during active treatment periods.

Methods: Ten experienced clinical scientists and medical monitors participated in a controlled comparison study using a synthetic dataset derived from a Phase III prostate cancer study, containing electronic data capture and laboratory data.

Results: Six categories of systematically introduced discrepancies represented common clinical dataset issues. Participants completed two 30-minute data cleaning sessions using different patient cohorts: one with traditional spreadsheet and visualization methods, another with the automated AI system. Primary metrics assessed throughput and accuracy, with participants completing useability and task-load surveys.

Conclusion: The AI-assisted approach demonstrated a six-fold increase in data cleaning throughput, with participants achieving 95.7 percent accuracy compared to 57.3 percent accuracy using traditional methods. Poststudy surveys revealed 100 percent of participants preferred the automated system across all measures. AI-assisted clinical data review provides transformative improvements in efficiency and accuracy. The six-fold throughput increase with near-perfect accuracy has significant implications for accelerating clinical trial timelines and improving patient access to treatments while maintaining rigorous data quality standards and regulatory compliance.

Funding/financial disclosures: Not provided.

Assessment devices and tools

Developing a brain-based metric of depression severity using portable brain measurement

Presenters: Julien Dubois; Ryan M. Field; Erin M. Koch; Zahra M. Aghajan; Naomi Miller; Katherine L. Perdue; and Moriah Taylor

Affiliations: All authors are with Kernel in Culver City, California.

Introduction: Measuring depression severity is central to evaluating the effectiveness of novel interventions. An objective measure of depression severity would enable better tracking across trial sites and stages of clinical development.

Methods: We employed a state-of-art time-domain functional near-infrared spectroscopy (TD-fNIRS) device, Kernel Flow, in eight clinical sites and recorded brain activity and Montgomery–Åsberg Depression Rating Scales (MADRS) of patients with major depressive disorder (MDD) before and after treatment. Data were collected from 86 patients at baseline and end of depression treatment. Brain measurements included both resting-state and task-based recordings. The relationship between change in MADRS and change in neural features were evaluated using Pearson correlation and we used an ordinary least squares (OLS) model to estimate the proportion of variance in MADRS change that is explained by neural metrics.

Results: We found that a number of functional brain metrics were sensitive to post-treatment changes in MADRS (p<0.05, fdr-bh corrected). Further, a fitted OLS model found that 59.8 percent variation in MADRS differences could be explained by a linear combination of differences in brain activity (F: 2.78; p<0.001).

Conclusion: Neural metrics collected across multiple clinic locations and from patients spanning a range of treatments were sensitive to changes in depressive states. The high explainability of MADRS score changes (from pre- to post-depression treatment) highlight the potential of functional brain activity to serve as a reliable adjunct to subjective assessments like MADRS and might improve CNS trials.

Funding/financial disclosures: All authors are employees of Kernel.

Standardization of the Trier Social Stress Test administration to enhance reproducibility

Presenters: Franziska Schmitt¹; Mark P. Berry²*; Michael C. Chen²; and Juliane Hellhammer¹

Affiliations: ¹daacro, Trier, Germany; ²Seaport Therapeutics, Boston, MA, USA

*Presenting author

Introduction: The Trier Social Stress Test (TSST) enables the induction and investigation of psychosocial stress in experimental settings, but human interaction can introduce variability in test administration and results. Here, we describe a refined TSST training and enhanced monitoring approach designed to minimize variability and ensure consistent stress induction.

Methods: The TSST was deployed in a randomized placebo-controlled Phase IIa trial (NCT05129865) to evaluate the potential of GlyphAlloTM (SPT-300 or GlyphTM Allopregnanolone) to reduce physiological stress. Two experienced trainers traveled to the research site where site team members were given a theoretical introduction to stress reactivity. The site team observed trial TSSTs conducted by trainers before the site team conducted practice TSSTs with role reversal, receiving feedback after each rehearsal. In an enhancement of the methodology, the TSST panel was videotaped throughout the trial and evaluated by the trainers. Individual feedback and retraining were performed if any issues were identified during video reviews.

Results: Six categories of systematically introduced discrepancies represented common clinical dataset issues. Participants completed two 30-minute data cleaning sessions using different patient cohorts: one with traditional spreadsheet and visualization methods, another with the automated AI system. Primary metrics assessed throughput and accuracy, with participants completing usability and task-load surveys.

Conclusion: This structured TSST training, which included the addition of a systematic video monitoring approach and personalized feedback, ensured high protocol compliance while reducing the risk of variability, supporting high-quality, consistent and reproducible stress induction across participants.

Funding/financial disclosures: MPB and MCC are currently employed by and hold stock in Seaport Therapeutics. JH and FS are currently employed by daacro, which provided training activities related to this Phase IIa TSST clinical trial. This study was supported by Seaport Therapeutics and, previously, Seaport Therapeutics’ predecessor, PureTech Health. ClinicalTrials.gov Identifier: NCT05129865.

Biomarkers

Improving success rates of CNS clinical trials through predictive biomarker stratification

Presenters: Klara Czobor and CK Singla

Affiliations: Both authors are with PA1.AI Inc.

Introduction: Over the past decade, numerous late-phase clinical trials for novel treatments of major neuropsychiatric disorders—including schizophrenia, bipolar disorder, depression, and Alzheimer’s disease—have failed. A primary driver of these failures is the declining effect size relative to placebo, underscoring the urgent need for predictive biomarkers to guide trial design.

Objective: This study evaluated the potential impact of biomarker-driven stratification strategies on trial design in neuropsychiatric disorders, using schizophrenia as a representative example.

Methods: We analyzed data from recently failed schizophrenia trials in which treatment arms demonstrated small, numerically favorable but statistically nonsignificant differences from placebo. Effect sizes for these comparisons were estimated, and post-hoc power analyses were conducted to determine the sample sizes required to achieve statistical significance in exact replications. We then modeled prospective trial scenarios incorporating biomarkers with varying levels of predictive accuracy, assessing their potential impact on sample size requirements and statistical power.

Results: Post-hoc power analyses (targeting 80% power, α=0.05) showed that replicating these failed trials without design changes would require sample sizes 1.6 to 9 times larger than originally enrolled, making such trials impractical. In contrast, incorporating biomarkers with realistic predictive accuracy (65–75% true positive rates) could substantially reduce sample size requirements in marker-positive populations while achieving 80 percent power.

Conclusion: Our findings suggest that some failed drug candidates might still prove to be effective in biomarker-defined subpopulations. Even moderately accurate biomarkers could substantially enhance trial efficiency, supporting their development as a critical tool for revitalizing central nervous system (CNS) drug discovery in light of widespread trial-failures.

Funding/financial disclosures: The authors gratefully acknowledge Pal Czobor, PhD, for providing statistical advice for the computations.

Decentralized and virtual clinical trials

Advancing remote symptom assessment in myasthenia gravis through digital health

Presenters: Ram Kinker Mishra, PhD¹; İlkay Yıldız Potter, PhD¹; Amanda Guidon, MD, MPH²; Ashkan Vaziri, PhD¹

Affiliations: ¹BioSensics LLC, Newton, MA; ²Department of Neurology, Division of Neuromuscular Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA

Objective: To develop and validate a multimodal digital health technology for remote monitoring of disease symptoms in myasthenia gravis (MG).

Methods: MG is a chronic autoimmune neuromuscular disease characterized by muscle weakness and fatigue. MG symptoms are typically evaluated by neuromuscular experts through in-person neurological examinations that are time-consuming, require disease expertise, and capture only a snapshot in time. We developed a multimodal digital health technology (DHT) called BioDigit MG, for monitoring MG symptoms. BioDigit MG includes tablet-guided speech and video-based assessments, electronic patient-reported outcomes relevant to MG, as well as a wearable sensor to measure physical activity and posture during activities of daily living. We assessed the real-world feasibility of BioDigit MG by conducting a clinical study with patients with MG who used the developed DHT to collect data related to their symptoms. To evaluate technology acceptance and usability, we conducted face-to-face interviews with the patients with MG and expert clinicians, each with over 13 years of practice.

Results: The study included 20 patients with MG, and five expert clinicians. During the study, a total of 219 speech tasks and 119 videos were successfully collected by the DHT, achieving 100 percent reliability in data collection and transfer. The patient and clinician participants found the DHT highly effective, easy to use, and well-suited to their needs.

Conclusion: The study illustrates the promising role of digital health technologies in revolutionizing the management of MG. By providing continuous, objective, and easily accessible monitoring, the BioDigit MG system can enhance the capability to manage MG effectively.

Funding/financial disclosures: Research reported in this publication was supported in part by the National Institute Of Neurological Disorders and Stroke of the National Institutes of Health under Award Number R44NS122672. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Investigative drug compounds and therapies

TerXT, a novel prodrug approach to a once-daily oral and long-acting injectable combination of xanomeline and trospium, shows improved pharmacokinetics in animal models

Presenter: Sam Clark, MD, PhD

Affiliation: Dr. Clark is founder and CEO of Terran Biosciences.

Introduction: The US Food and Drug Administration recently approved Cobenfy (Bristol Myers Squibb, New York, New York), a twice-daily combination of xanomeline tartrate and trospium chloride for schizophrenia. However, trospium chloride has very poor absorption, which results in a strong food effect and perisistent gastroinestinal side effects, and limits the feasibility of improved dosing forms such as once-daily oral and long-acting injectable (LAI) forms. We applied a prodrug strategy to xanomeline and trospium to improve absorption and pharmacokinetic (PK) matching, and reduce side effects. These novel prodrugs enable once-daily oral and long-acting injectable administration.

Methods: We designed and synthesized over 1,000 examples of xanomeline and trospium prodrugs, which were tested in rodent, dog, and nonhuman primate models and compared to the original parent compounds in Cobenfy.

Results: We identified a lead prodrug combination of xanomeline/trospium that outperformed the parent compounds on both oral and intramuscular metrics. After oral dosing, this combination showed a 9.5-fold improved area under the curve (AUC) in dogs and a 5.5-fold improved AUC in nonhuman primates without a food effect, enabling once-daily dosing. The lead prodrug combination also showed improved stability in a wide range of pH buffered solutions.

Conclusion: Our study showed that the limitations of Cobenfy (xanomeline/trospium) could be overcome through a prodrug approach, enabling once-daily oral administration and a long-acting injection, with the potential for elimination of the food effect and reduced side effects through better PK matching. These results support the further development of these prodrugs in trials for patients with schizophrenia.

Funding/financial disclosures: Dr. Sam Clark is a full-time employee and stockholder of Terran Biosciences.

Lumateperone as adjunctive therapy in patients with major depressive disorder and anxious distress

Presenters: Suresh Durgam, MD¹; Willie R. Earley, MD¹; Susan G. Kozauer, MD¹*; Changzheng Chen, PhD¹; Renee Rotolo, PhD¹; Gary S. Sachs, MD^2,3

Affiliations: ¹Intra-Cellular Therapies, a Johnson & Johnson Company, Bedminster, NJ; ²Massachusetts General Hospital, Boston, MA; ³Signant Health, Plymouth Meeting, PA

Introduction: Lumateperone is FDA-approved to treat schizophrenia and depressive episodes associated with bipolar I or II disorder. This analysis of a positive, Phase III, randomized, double-blind, placebo-controlled trial (Study 501, NCT04985942) investigated adjunctive lumateperone 42mg efficacy in patients with major depressive disorder (MDD) with inadequate antidepressant therapy (ADT) response and anxious distress.

Methods: Adults with DSM-5–diagnosed MDD with inadequate response to 1-2 ADT in the current depressive episode were randomized to six-week oral lumateperone 42mg+ADT or placebo+ADT. Anxiety was assessed using Generalized Anxiety Disorder-7 (GAD-7) Total score. In patients with DSM-5–diagnosed anxious distress Montgomery-Asberg Depression Rating Scale (MADRS) Total, Clinical Global Impression-Severity Scale (CGI-S), and Quick Inventory of Depressive Symptomatology-Self Report-16 item (QIDS-SR-16) Total scores were evaluated.

Results: Of 481 patients (modified intent-to-treat [mITT] population), 207 (43.0%) had anxious distress (lumateperone, 109; placebo, 98) at baseline. Lumateperone+ADT improved MADRS Total score (least squares mean difference vs placebo [LSMD]= −6.8; effect size [ES]= −0.85; p<0.0001) and CGI-S score (LSMD= −0.9; ES=−0.91; p<0.0001) from baseline to Day 43 vs placebo+ADT in patients with anxious distress. Lumateperone+ADT significantly improved self-reported depression (QIDS-SR-16 Total score; LSMD= −3.7; ES= −0.80; p<.0001) at Day 43 vs. placebo+ADT in patients with anxious distress. Patient-reported anxiety (GAD-7 Total score) improved at Day 43 vs placebo+ADT in the ITT (baseline mean: lumateperone+ADT, 9.9; placebo+ADT, 9.6; LSMD= −1.6; ES= −0.43; p<0.0001) and subgroup with anxious distress (LSMD= −3.24; ES= −0.88; p<0.0001).

Conclusion: Lumateperone 42mg+ADT improved symptoms of depression and anxiety vs placebo+ADT, indicating lumateperone as a promising adjunctive therapy in patients with MDD with anxious distress.

Funding/financial disclosures: S Durgam, WR Earley, C Chen, and R Rotolo are full-time employees of Intra-Cellular Therapies, a Johnson & Johnson Company. SG Kozauer is a former employee of Intra-Cellular Therapies, a Johnson & Johnson Company. GS Sachs is an employee of Signant Health.

Long-term adjunctive lumateperone treatment in major depressive disorder: results from a six-month open-label extension study

Presenters: Willie R. Earley, MD¹; Suresh Durgam, MD¹; Susan G. Kozauer, MD¹*; Changzheng Chen, PhD¹; Joseph Fawole, MD¹; Andrew J. Cutler, MD²

Affiliations: ¹Intra-Cellular Therapies, a Johnson & Johnson Company, Bedminster, NJ; ²Department of Psychiatry, SUNY Upstate Medical University, Lakewood Ranch, FL

*Former employee

Introduction: Lumateperone is FDA-approved to treat schizophrenia and bipolar depression. Adjunctive lumateperone 42mg efficacy and safety was demonstrated in patients with major depressive disorder (MDD) with inadequate antidepressant therapy (ADT) response (Study 501/NCT04985942; Study 502/NCT05061706). This Phase III open-label extension (OLE) Study 503 (NCT05061719) investigated long-term lumateperone 42mg+ADT safety in patients who completed Studies 501/502.

Methods: Studies 501/502 enrolled adults with DSM-5–defined MDD with inadequate response to 1-2 ADT in the current depressive episode; patients completing six-week double-blind treatment enrolled in the OLE received 26-week lumateperone 42mg+ADT. The primary endpoint was safety/tolerability. The secondary endpoint was depression by Montgomery-Asberg Depression Rating Scale (MADRS) Total and Clinical Global Impression-Severity Scale (CGI-S) scores.

Results: Of 809 patients, 84.5 percent completed treatment. Treatment-emergent adverse events (TEAEs) occurred in 548 patients (67.7%). Treatment discontinuation due to AEs occurred in 7.4 percent of patients. TEAEs (≥5%) were headache (16.6%), dizziness (10.6%), dry mouth (8.0%), nausea (7.7%), somnolence (7.2%), diarrhea (6.2%), and nasopharyngitis (5.2%). Most TEAEs (98.9%) were mild-to-moderate severity. Extrapyramidal symptom (EPS)-related TEAE rates were low (3.8%) with no mean increase in EPS scales. Changes from baseline to end-of-treatment were minimal and not clinically meaningful for body morphology (eg, weight, body mass index, waist circumference), cardiometabolic parameters, prolactin, blood pressure, and electrocardiogram measures. No emergence of serious suicidal ideation occurred. Depressive symptoms improved, by mean change from Study 501/502 baseline to Week 26 in MADRS Total (−22.9; p<0.0001) and CGI-S (−2.7; p<0.0001) scores.

Conclusion: Lumateperone 42mg+ADT was safe and effective during 26-week treatment in patients with MDD and inadequate ADT response.

Funding/financial disclosures: WR Earley, S Durgam, C Chen, and J Fawole are full-time employees of Intra-Cellular Therapies, a Johnson & Johnson Company. SG Kozauer is a former employee of Intra-Cellular Therapies, a Johnson & Johnson Company.

AJ Cutler has served as a consultant/on an advisory board for: AbbVie, Acadia, Actinogen, Alfasigma, Alkermes, Anavex Life Sciences, Arrivo BioVentures, Autobahn Therapeutics, Axsome, Biogen, Biohaven, Boehringer Ingelheim, Brii Biosciences, Bristol Myers Squibb, Cerevel, Cognitive Research Corporation, Collegium Pharmaceutical, Corium, Delpor, Evolution Research Group, 4M Therapeutics, Intra-Cellular Therapies, J&J Innovative Medicine, Jazz Pharma, Karuna, Knight Therapeutics, LivoNova, Lundbeck, Luye Pharma, MapLight Therapeutics, MedAvante-ProPhase, Mentavi, Neumora, Neurocrine, Neuroscience Education Institute, NeuroSigma, Noven, Otsuka, PaxMedica, Relmada, Sage Therapeutics, Sirtsei Pharmaceuticals, Supernus, Teva, Thynk, Tris Pharma, Vanda Pharmaceuticals, and VistaGen; Served on a speaker’s bureau for: AbbVie, Alfasigma, Alkermes, Axsome, Boehringer Ingelheim, Bristol Myers Squibb, Corium, Intra-Cellular Therapies, Ironshore Pharmaceuticals, J&J, Lundbeck, Neurocrine, Noven, Otsuka, Supernus, Teva, Tris Pharma, and Vanda Pharmaceuticals; Owns stock options/equity with: 4M Therapeutics; Served on a data safety monitoring board for: Alar Pharma, COMPASS Pathways, Freedom Biosciences, and Pain Therapeutics.

Adjunctive lumateperone 42mg treatment in major depressive disorder: efficacy in anhedonia and across broad range of depressive symptoms

Presenters: Willie R. Earley, MD¹; Suresh Durgam, MD¹; Susan G. Kozauer, MD¹*; Changzheng Chen, PhD¹; Hassan Lakkis, PhD¹; Robert Hayes, PhD¹; Michael E. Thase, MD²

Affiliations: ¹Intra-Cellular Therapies, a Johnson & Johnson Company, Bedminster, NJ; ²Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA

*Former employee

Introduction: Lumateperone is an FDA-approved antipsychotic to treat schizophrenia and bipolar depression. A positive Phase III, randomized, double-blind, placebo-controlled trial (Study 501, NCT04985942), investigated adjunctive lumateperone 42mg in patients with major depressive disorder (MDD) with inadequate antidepressant therapy (ADT) response. This analysis evaluated efficacy of lumateperone+ADT across Montgomery-Asberg Depression Rating Scale (MADRS) anhedonia factor and single-item scores.

Methods: Adults had DSM-5–diagnosed MDD with inadequate response to 1-2 ADT in the current depressive episode and had MADRS Total score≥24 and Clinical Global Impression-Severity Scale score≥4. Patients received 6-week oral placebo+ADT or lumateperone 42mg+ADT. Post-hoc analyses evaluated anhedonia according to MADRS anhedonia factor (sum of apparent sadness, reported sadness, concentration difficulties, lassitude, and inability to feel). Prospective analyses investigated change in individual MADRS items.

Results: Lumateperone+ADT significantly improved MADRS Total score vs placebo+ADT at Day 43 (lumateperone+ADT, n=239; placebo+ADT, n=242; least squares mean difference vs placebo+ADT [LSMD]= −4.9; ES= −0.61; p<0.0001). Lumateperone+ADT significantly improved anhedonia symptoms vs placebo+ADT at every visit according to MADRS anhedonia factor score (Day 43: LSMD= −2.5; ES= −0.50; p<0.0001). All MADRS items comprising the anhedonia factor significantly improved with lumateperone+ADT by Day 29.

At Day 43, nine of the 10 MADRS items significantly improved with lumateperone+ADT vs placebo+ADT (p<0.01): apparent sadness (LSMD= −0.6), reported sadness (LSMD= −0.6), inner tension (LSMD= −0.6), reduced sleep (LSMD= −0.9), reduced appetite (LSMD= −0.4), concentration difficulties (LSMD= −0.4), lassitude (LSMD= −0.5), inability to feel (LSMD= −0.4), and pessimistic thoughts (LSMD= −0.4).

Conclusion: Lumateperone 42mg+ADT significantly improved anhedonia and a broad range of depression symptoms vs placebo+ADT in patients with MDD with inadequate ADT response. 

Funding/financial disclosures: WR Earley, S Durgam, C Chen, H Lakkis, and R Hayes are full-time employees of Intra-Cellular Therapies, a Johnson & Johnson Company. SG Kozauer is a former employee of Intra-Cellular Therapies, a Johnson & Johnson Company.

ME Thase has served as an advisor or a consultant for Autobahn Therapeutics; Axsome Therapeutics, Inc.; Clexio Biosciences; Gerson Lehman; GH Therapeutics; H. Lundbeck, A/S; Janssen Pharmaceuticals, Inc.; Johnson & Johnson; Luye Pharma Group, Ltd.; Merck & Company, Inc.; Object Pharma; Otsuka Pharmaceutical Company, Ltd.; Pfizer, Inc.; Sage Pharmaceuticals; Seelos Pharmaceuticals; Takeda Pharmaceutical Company, Ltd.; has received grants from Acadia Inc.; Alkermes; Axsome Therapeutics Inc.; Intracellular, Inc.; Janssen Pharmaceuticals, Inc.; Myriad; National Institute of Mental Health; Otsuka Pharmaceutical Company, Ltd.; Patient-Centered Outcomes Research Institute (PCORI); Takeda Pharmaceutical Company, Ltd.; and has received royalties from the American Psychiatric Foundation; Guilford Publications; Herald House; Kluwer-Wolters; W.W. Norton & Company, Inc., and Spouse’s Employment with Open Health, which does business with most major pharmaceutical companies.

A Phase III, randomized, double-blind, placebo-controlled, parallel group trial to evaluate the safety and efficacy of X/T for the treatment of psychosis associated with Alzheimer’s disease (ADEPT-4)

Presenters: Minsu Kang¹; Jeffrey L Cummings²; Shawna Fox¹; George Grossberg³; Ron Marcus¹

Affiliations: ¹Bristol Myers Squibb, Princeton, NJ; ²Chambers-Grundy Center for Transformative Neuroscience, Department of Brain Health, Kirk Kerkorian School of Medicine, University of Nevada Las Vegas, Las Vegas, NV; ³Department of Psychiatry and Behavioral Neuroscience, Saint Louis University School of Medicine, St Louis, MO

Introduction: Alzheimer’s disease (AD) is the most common cause of dementia, and neuropsychiatric symptoms, such as psychosis, are among the most common symptoms. AD psychosis is often treated using off-label therapies with safety concerns and limited efficacy, highlighting unmet medical need.

In previous research, the M1/M4 preferring muscarinic receptor agonist xanomeline improved cognition and psychosis in people with AD, but peripheral cholinergic-related adverse events halted development. Xanomeline combined with the peripherally restricted muscarinic receptor antagonist trospium chloride (X/T) is approved to treat schizophrenia. X/T reduced symptoms of psychosis, improved cognitive symptoms, and showed lower risk of adverse events associated with conventional antipsychotics. These data suggest X/T could be effective in treating AD symptoms.

Our aim is to evaluate the safety and efficacy of X/T in adults with mild-to-severe AD dementia with moderate-severe psychosis.

Methods: ADEPT-4 (NCT06585787) is a 14-week, Phase III, randomized, double-blind, placebo-controlled, parallel group trial of participants aged 55 to 90 years who are diagnosed with AD with biomarker confirmation, exhibit no other central nervous system diseases via imaging, and have at least two months of psychosis prior to screening.

Results: This trial is currently recruiting and estimated to enroll about 406 participants. The primary endpoint is mean change from baseline to end of treatment in the Neuropsychiatric Inventory-Clinician: Hallucinations and Delusions score. The key secondary endpoint is change from baseline in Clinical Global Impression-Severity scale.

Conclusion: X/T has the potential to be the first in a new class of muscarinic receptor agonists for the treatment of AD psychosis.

Funding/financial disclosures: JLC has provided consultation to Acadia, Acumen, ALZpath, Annovis, Artery, Axsome, Biogen, Biohaven, Bristol Myers Squibb, CervoMed, Eisai, Fosun, GAP Foundation, Green Valley, Hummingbird Diagnostics, IGC, Janssen, Kinoxis, Lighthouse, Lilly, Lundbeck, LSP/eqt, Merck, MoCA Cognition, Novo Nordisk, NSC Therapeutics, Optoceutics, Otsuka, Praxis, ReMYND, Roche, Scottish Brain Sciences, Signant Health, Simcere, Sinaptica, T-Neuro, TrueBinding, and Vaxxinity pharmaceutical, assessment, and investment companies. JLC is cofounder of CNS Innovations and Mangrove Therapeutics. JLC owns the copyright of the Neuropsychiatric Inventory. JLC is supported by NIA R35AG71476, NIA R25AG083721-01, Alzheimer’s Disease Drug Discovery Foundation (ADDF), Ted and Maria Quirk Endowment, and Joy Chambers-Grundy Endowment. GG has been a consultant for Axsome, Biogen, BioXcel, BMS, Eisai, Karuna, Lundbeck, MapLight, Otsuka, and Takeda; received research support from the NIA and Functional Neuromodulation; been on the safety monitoring committee for Anavex, Merck, Newron, Oligomerix, and Quince; and has been on the speakers bureau for Biogen, Eisai, and Otsuka. MK, SF, and RM are employees of Bristol Myers Squibb.

Benefit of NOE-105 monotherapy in adults with childhood onset fluency disorder

Presenters: Robert Lasser¹; George Garibaldi²; Smiljana Ristic; and Gerald Maguire³

Affiliations: ¹Noema Pharma; ²Garibaldi Consulting GmBH; ³Professor of Psychiatry and Neuroscience at the University of California, Riverside School of Medicine in Riverside, California.

Introduction: Childhood Onset Fluency Disorder (COFD/stuttering/stammering), profoundly impacts at least one percent of individuals worldwide, considered related to cortical and subcortical dopaminergic dysfunction. NOE-105, an investigational selective PDE10A inhibitor which targets GABAergic spiny projection neurons in the striatum to impact dopaminergic circuits was studied in NOE-CFD-201.

Methods: NOE-CFD-201 was a 10-week study of adult male individuals with COFD (US and Australia). Following screening, participants entered a single-blind placebo run-in period. Randomization (1:1) occurred at Day 1, 15, or 29, based on placebo nonresponse. NOE-105 was individually titrated to 15mg/d and maintained for up to 10 weeks. Primary endpoint was treatment group difference in least square (LS) mean change from baseline to Week 6 on the Maguire-Leal-Garibaldi Self-rated Stuttering Scale (MLGSSS) Total score. Post hoc analyses examined outcomes in participants who maintained moderate-to-severe symptoms (MLGSS total score >24) at randomization.

Results: 75 participants entered the placebo run-in period, 65 were randomized; main reasons for discontinuation were adverse events or withdrawal by subject. Primary analysis was not statistically significant: –0.39 points [-6.1±1.8 NOE-105, -5.7±1.7 placebo; 95% confidence interval [CI]: –5.22 to 4.43]; p=0.873. For participants with moderate-to-severe symptoms (MLGSS total score >24, n= 37), the primary analysis showed a notably larger treatment difference of –5.9 points [–12±2.7 NOE-105, –6.1±2.1 placebo, 90% CI: –11.74 to –0.02]; one-sided p=0.049. Adverse events were mild-to-moderate in severity.

Conclusion: TNOE-105 was safe and well-tolerated in this first controlled trial in COFD; participants with moderate-to-severe symptoms exhibited a substantial treatment effect supporting further exploration in larger controlled trial.

Funding/financial disclosures: This work was funded by Noema Pharma AG.

Intranasal 5-MeO-DMT concomitant with SSRI for treatment-resistant depression: a proof-of-concept trial

Presenters: Mathieu Seynaeve; Fiona Dunbar; Anna O. Ermakova; Chandni Hindocha; Mimi Pierce; Robert R. Conley; and Claire Roberts

Affiliations: All authors are with Beckley Psytech Ltd. as employees or consultants.

Introduction: Anecdotal evidence has raised questions about the safety and efficacy of co-administering psychedelics with selective serotonin reuptake inhibitors (SSRIs). BPL-003, a novel intranasal formulation of 5-methoxy-N,N-dimethyltryptamine (5-MeO-DMT.benzoate), is a short-acting psychedelic with acceptable safety and promising efficacy as monotherapy in patients with treatment-resistant
depression (TRD).

This study assessed the safety, tolerability and impact on depressive symptoms of ascending doses of BPL-003, in TRD participants on a stable dose of an SSRI.

Methods: As part of an open-label Phase IIa study, eligible participants received a single dose of 10 or 12mg intranasal BPL-003, with psychological support, with a 12-week follow up. Safety was assessed via treatment emergent adverse events (TEAEs). Symptom improvement was determined by changes in the Montgomery Åsberg Depression Rating Scale (MADRS) Total score.

Results: Twelve subjects with TRD on stable doses of an SSRI participated in the trial: seven female subjects, five male subjects, 92 percent White. TEAEs were similar for both doses with the most common events being administration site symptoms and gastrointestinal effects. Rapid and sustained reductions in MADRS Total scores were observed after both doses. In the 10mg cohort, 66.7 percent of participants were responders at the first post-dose assessment on Day 2 and 83 percent were responders at Day 85. Similar results were observed in the 12mg cohort.

Conclusion: Single doses of 10mg and 12mg BPL-003 administered concomitantly with SSRIs, in participants with TRD, were well tolerated and demonstrated preliminary evidence for a sustained improvement in depression.

Funding/financial disclosures: The study was Sponsored and funded by Beckley Psytech Ltd. Trial Registration (NCT05660642)

Efficacy and safety of strategies for switching to xanomeline and trospium chloride from standard of care atypical antipsychotics: design of an open-label trial in people with schizophrenia

Presenters: David P. Walling¹; Pierre Nicolas²; Eliesha Daniels²*; Lauren White³; Naomi Marbot²

Affiliations: ¹CenExel CNS, Los Alamitos, CA; ²Bristol Myers Squibb, Princeton, NJ; ³CenExel, Atlanta, GA

*At the time of trial design

Introduction: The dual M1/M4 muscarinic receptor agonist xanomeline combined with the pan-muscarinic receptor antagonist trospium chloride is approved by the US FDA for treatment of schizophrenia in adults. Evidence-based guidance on safely and effectively switching from standard-of-care atypical antipsychotics (AAs) to xanomeline/trospium (X/T) is needed. This is the first study to evaluate switching from AAs to an approved muscarinic compound. The eight-week, open-label, multicenter, outpatient trial (NCT06924255) will assess the efficacy, safety, and tolerability of X/T in adults with schizophrenia switching from AAs.

Methods: Approximately 100 adults aged 18 to 65 years with a primary diagnosis of schizophrenia, stable symptoms, stable oral AA regimen for ≥6 weeks, and baseline scores of ≤80 on the Positive and Negative Syndrome Scale (PANSS) and ≤4 on the Clinical Global Impression-Severity (CGI-S) scale will be enrolled. Exclusion criteria include history of antipsychotic therapy resistance, psychiatric hospitalization of more than 30 days within 12 months of screening, and prior xanomeline or trospium exposure. The design employs oral AA therapy de-escalation and X/T titration to a therapeutic dose. Following a ≤2-week screening period, participants will be randomized to either an accelerated (by Week 2) or slower (by Week 4) cross-titration switch with X/T treatment for up to eight weeks. Safety follow-up will occur one week post treatment.

Results: The primary endpoint is all-cause discontinuation. Secondary endpoints include adverse events incidence and change from baseline in PANSS total, CGI-S, and Personal and Social Performance scale scores.

Conclusion: Results will inform clinical decision-making when switching from oral AAs to X/T.

Funding/financial disclosures: DPW and LW are employees of CenExel. PN and NM are employees of Bristol Myers Squibb. ED was an employee of Bristol Myers Squibb at the time the trial was designed.

Wearables and mobile applications (APPS)

Cough and lung sounds as digital endpoints: remote monitoring of respiratory health using novel wearables

Presenters: Nick Delmonico; Shane Krauss; Gabe Steerman; and Yu Kan Au, MD

Introduction: Stethoscope auscultation has been a gold standard for over 200 years in assessing respiratory health but is limited to episodic, in-person visits. Strados Labs’ RESP® Biosensor (FDA 510(k) cleared) continuously and passively collects lung sounds in daily life, allowing for novel digital endpoints in clinical trials alongside patient reported outcomes (PROs). Strados Labs has also developed the RESP® Watch for seamless, unobtrusive cough monitoring.

Methods: Retrospective case reviews and observational studies were conducted at two separate facilities (Philadelphia, Pennsylvania and Asheville, New York) in patients with exacerbations of chronic lung diseases, including nine with asthma, 17 with chronic obstructive pulmonary disease (COPD), two with asthma-COPD overlap syndrome (ACOS), and two with COVID-19. RESP® was deployed for continuous lung sound monitoring throughout hospitalization, with some patients continuing use postdischarge.

Results: Among hospitalized patients, RESP® provided real-time lung sound analysis, allowing for objective monitoring of disease progression. In a case of severe asthma, RESP® detected increasing wheezing before the patient perceived their own symptom worsening, leading to readmission and eventual intubation. In an adult patient with AE-COPD, RESP® captured an increase of lung sounds up to 119 percent from baseline prior to hospital discharge and during the postdischarge period. He was ultimately readmitted five days after discharge.

Conclusion: Continuous monitoring of lung sounds might offer a valuable clinical tool that can provide novel insight into patients’ respiratory status. In pharmaceutical clinical trials, cough and lung sounds can be collected as digital endpoints and might enable greater understanding of treatment effects.

Funding/financial disclosures: Not provided.

Accurate, real-world gait assessment from a chest-worn sensor to reduce patient burden in CNS trials

Presenters: Brett Meyer¹; Reed Gurchiek²; Ryan McGinnis³; and Melissa Ceruolo¹

Affiliations: ¹Medidata Solutions, A Dassault Systemes Company; ²Department of Bioengineering, Clemson University; ³Center for Remote Health Monitoring, Wake Forest University School of Medicine and Department of Biomedical Engineering, Wake Forest University School of Medicine

Introduction: Effectively measuring disease progression in CNS disorders is challenging, as traditional, episodic evaluations fail to capture symptom fluctuations in daily life. Sensors enable passive collection of continuous, real-world evidence for developing sensitive digital endpoints in CNS clinical trials and digital therapeutics. Our goal was to validate a novel method for accurate gait assessment using a low-burden, single chest-mounted accelerometer.

Methods: Gait data were collected from 14 healthy adults and four individuals with Huntington’s disease (HD) during walking tasks. Participants wore a chest-mounted accelerometer while a gold-standard motion capture system collected ground truth data. A novel algorithm leveraging wavelet analysis segmented stride events (eg, stride and stance time), and a deep learning model estimated stride length.

Results: The approach achieved high accuracy against ground truth, with a root mean square error of 0.043 seconds for stride time, 0.067 seconds for stand time, and 0.117 meters for stride length. The algorithm demonstrated robust performance across both healthy and HD cohorts, with the ability to capture clinically meaningful changes and significantly outperforming existing public algorithms which failed to generalize to the impaired population.

Conclusion: This work validates a practical and robust method for collecting objective, quantifiable data needed for next-generation digital endpoints. By enabling continuous passive monitoring, this approach provides deeper insights into a patient’s real-world function and symptom variability. This is critical for developing more sensitive measures of therapeutic impact, ultimately enhancing the development and validation of novel therapeutics for CNS disorders by providing a more complete view of the patient experience.

Funding/financial disclosures: Not provided.

Monitoring and prediction of disease activities in Friedreich ataxia using wearable sensors

Presenters: Ram Kinker Mishra, PhD¹; Ana Enriquez¹; Victoria R. Profeta^2,3; McKenzie Wells^2,3; David Lynch, MD^2,3; and Ashkan Vaziri, PhD¹

Affiliations: ¹BioSensics LLC, Newton, MA; ²Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA; ³Departments of Pediatrics and Neurology, The Children’s Hospital of Philadelphia, Abramson Research Center, Philadelphia, PA

Objective: To investigate wearable sensors for objective monitoring physical activity and upper extremity function in Friedreich ataxia (FRDA) patients.

Methods: FRDA causes progressive impairments in gait, upper extremity function, and speech. Current in-clinic assessments are subjective, time-consuming, and provide limited insights into patients’ daily disabilities. We recruited 39 patients with FRDA (average age 26.8 ± 1.6 years; 19 female) and remotely monitored them using wearable sensors for seven consecutive days. We compared sensor-derived metrics of lower and upper extremity function during daily activities with clinical measures (eg, mFARS, FA-ADL) and biological markers (GAA and frataxin levels). Additionally, we developed machine learning models using demographic, biological, and sensor-derived data to predict disease severity.

Results: The sensor-derived physical activity metrics demonstrated significant correlations with clinical and biological outcomes. Median steps per walking bout were strongly correlated with FA-ADL (ρ=–0.63, p<0.001) and mFARS (ρ=–0.61, p<0.001). Regarding biological markers, the percentage of sitting time was positively correlated with frataxin levels (r=0.64, p=0.002). Metrics related to goal-directed movements, such as median velocity, also correlated significantly with mFARS (ρ=–0.56, p=0.004) and 9-HPT (ρ=–0.60, p=0.001). The machine learning models including sensor-derived metrics improved predictive accuracy, achieving R² values up to 0.77 and 0.81 for predicting GAA and frataxin levels, respectively.

Conclusion: Wearable sensors demonstrate potential for objectively assessing disease severity and motor dysfunction in FRDA, and offer a more comprehensive and continuous monitoring approach than traditional in-clinic assessments.

Funding/financial disclosures: The study was supported in part by BioSensics LLC, and in part by the Friedrich’s Ataxia Research Alliance.

Digital and wearable technologies for remote, multidomain monitoring in ALS

Presenters: Ram Kinker Mishra, PhD¹; İlkay Yıldız Potter, PhD¹; Zachary Simmons, MD²; Andrew Geronimo, PhD²; Ashkan Vaziri, PhD¹

Affiliations: ¹BioSensics LLC, Newton, MA; ²Department of Neurology, Penn State College of Medicine, Hershey, PA

Objective: To develop a remote monitoring system using wearable technology to track changes in physical activity, fine motor function, patient-reported outcomes (PROs), and speech in individuals with amyotrophic lateral sclerosis (ALS).

Methods: ALS participants were followed for up to 12 months, with clinical visits at baseline and every three months. After each visit, participants wore a PAMSys™ pendant and two PAMSys ULM™ wrist sensors to monitor daily physical activity and hand movements for one week. Biweekly digital assessments of speech, handwriting, and pattern-tracing were conducted using BioDigit Home tablets. Data were analyzed using repeated measures correlation and linear mixed models.

Results: Data from 18 ALS participants (mean age 66.3 ± 9.6 years; 6 female) across 51 visits were analyzed. Physical activity sensor data correlated significantly with ALSFRS-R gross motor scores (r=0.41 to 0.60, p<0.05). Speech metrics, including articulatory rate and intelligibility, showed moderate-to-strong associations with ALSFRS-R bulbar and respiratory scores. High compliance rates were observed: 91.9 percent for speech tasks, 96.5 percent for pendant sensors (>18 hours worn), and 88 percent for handwriting and pattern-tracing tasks.

Conclusion: These pilot results demonstrate the feasibility of multimodal, at-home monitoring for ALS. Digital assessments offer objective measurement, high compliance, scalability, and cost-effectiveness, supporting efforts to improve ALS care and health equity. A comprehensive assessment of neurological symptoms and disease progression via digital assessments can improve patient care and readiness for clinical trials in ALS.

Funding/financial disclosures: This research is supported in part by BioSensics LLC.