A Multimodal Data Capture System for Assessing Outcomes in Autism Clinical Trials

Panel Presentation
Thursday, May 2, 2019: 11:45 AM
Room: 517A (Palais des congres de Montreal)
G. J. Pandina, Janssen Research & Development, Titusville, NJ
Background: Autism Spectrum Disorder (ASD) is a complex, heterogeneous neurodevelopmental disorder with no approved medications for core symptoms. Objective measures have potential to de-risk drug development by improving patient selection and enhancing sensitivity to drug response. Mobile and web-based approaches could increase signal detection and decrease study burden by allowing fewer in-person visits, using momentary assessment and other novel measurement techniques. Lab- and home-based sensors could create novel, objective endpoints that help with stratification and change measurement in ASD clinical trials.

Objectives: The study objective was to demonstrate the utility of the Janssen Autism Knowledge Engine (JAKE®), a standardized set of mobile/web-based tools and lab- and home-based experimental, proof-of-concept sensor array, to identify potential markers of population enrichment and symptomatic change in ASD. The Autism Behavior Inventory, a caregiver-based rating scale administered through the My JAKE web portal was compared with other common ASD scales. Other experimental mobile assessments assessed caregiver-selected most troubling symptoms, global symptom severity, and important factors (sleep, mood/affect, etc.). Results were compared with lab-based sensor-derived endpoints, using a battery of computer tests specifically designed to detect social communication problems and repetitive behaviors/interests, and via passive sensing at home.

Methods: A total of N=144 children and adults (aged ≥6 to 53 years) with ASD were assessed in an 8-10 week prospective non-interventional trial (US, 9 sites) with the JAKE system. JAKE includes: 1) a web-and-mobile assessment of clinical symptoms (ABI, ASD-related events, mood report, medical/developmental history) and; 2) an array of experimental home-based wearable sensors for measuring activity and sleep, and lab-based computer tasks designed to provoke ASD-specific physiological signals on a variety of sensors. Participants completed battery of clinical rating scales. ASD results were compared with age and sex-matched typically developing (TD) controls (N=41), who completed rating scales and lab-based test battery once.

Results: Data from rating scales, lab-based tasks, and home- and lab-based web and mobile assessments yielded multiple findings that discriminated between ASD and TD. ASD participants were able to complete the lab-based task battery and produce valid data. ASD and TD participants showed differences in sensor based biologic responses on lab-based tasks. These include reduced facial affect expressivity (both passively and on-demand), a lack of face preference in dynamic social vs. non-social stimuli and a lack of preference for biological motion (both measured by gaze fixation), and differences in resting state electrical activity in the brain and coherence between brain regions during social perceptual tasks. Relationships between symptoms (rating scales) and lab and home-based sensor assessments were modest, suggesting these tools may measure different aspects of ASD.

Conclusions: Assessment of ASD subjects across a range of severity and age using a coordinated system of web-and-mobile clinician and caregiver tools, and lab/home based sensor technologies, is feasible and offers a robust method for assessing a broad range of ASD symptoms. This multimodal approach produces a complex blueprint of ASD behavior of potential utility in clinical trials.