SLIDE 1
Disclosures
- Full time employee of Janssen Research & Development LLC
- Johnson & Johnson Stockholder
- The opinions expressed in this presentation are those of Dr. Pandina,
not Janssen or Johnson & Johnson
2
The Role of Digital Medicine in Autism Spectrum Disorder Clinical - - PowerPoint PPT Presentation
The Role of Digital Medicine in Autism Spectrum Disorder Clinical - - PowerPoint PPT Presentation
The Role of Digital Medicine in Autism Spectrum Disorder Clinical Trials Gahan J. Pandina, PhD ISCTM-ECNP Joint Autumn Conference 6 September 2019 Copenhagen, Denmark Disclosures Full time employee of Janssen Research &
SLIDE 2
SLIDE 3
Agenda
3
- Role of digital medicine in ASD clinical trials
- Digital data collection – strengths and challenges
- Sample sensor data / correlation with clinical results
- Complex analytic approaches
- Conclusions
SLIDE 4
Digital Medicine and ASD Clinical Trials
4
Zhang et al (2018). Digital medicine: Emergence, definition, scope, and future. Digit Med, 4:1-4.
Describes the intersection of autism assessment, diagnosis, and intervention and digital technologies
Health-based video games Wearables and trackers Electronic health records Other Biosensors Web-and-mobile apps
SLIDE 5
Digital Medicine and ASD Clinical Trials
5
Zhang et al (2018). Digital medicine: Emergence, definition, scope, and future. Digit Med, 4:1-4.
Describes the intersection of autism assessment, diagnosis, and intervention and digital technologies
Academia, Sites, Foundations, Sponsor Companies Regulatory Agencies CROs & Partners Health-based video games Wearables and trackers Electronic health records Other Biosensors Web-and-mobile apps
INTEGRATION
SLIDE 6
Digital Medicine and ASD Clinical Trials
6
Zhang et al (2018). Digital medicine: Emergence, definition, scope, and future. Digit Med, 4:1-4.
Describes the intersection of autism assessment, diagnosis, and intervention and digital technologies
Academia, Sites, Foundations, Sponsor Companies Regulatory Agencies CROs & Partners Health-based video games Wearables and trackers Electronic health records Other Biosensors Web-and-mobile apps
INTEGRATION
Biomarkers & Other Endpoints
PoC Participant Selection Change Measurement Population Stratification
SLIDE 7
Use of Digital Medicine in ASD Clinical Trials
Strengths and Challenges
7
Zhang et al (2018). Digital medicine: Emergence, definition, scope, and future. Digit Med, 4:1-4.
Capture of digital endpoints – vs. standard paper rating scales Strengths Capture data in real time Central tracking Reduced burden of data entry, scoring Reduced data errors / loss Optimize point-of-collection (clinical site, home, etc.) Pre-programmed queries/skip-outs (no “maybe”) Full audit trail / data QC Maintain control over comments and writing in the margin Passive, “automatic” data capture Challenges Cost of development, implementation, and maintenance
- Hardware, software
Site and participant training, support, and account management Managing a compliant, global framework Back-up approach in case of problems
SLIDE 8
8
*Please note that JAKE is for research use only **Ness et al (2019). Frontiers Neurosci: 13:111. https://doi.org/10.3389/fnins.2019.00111
Overview - Janssen Autism Knowledge Engine*
**
SLIDE 9
Autism Biomarker Study Design
9
Ness et al (2019). Frontiers Neurosci: 13:111. https://doi.org/10.3389/fnins.2019.00111
Key Inclusion/ Exclusion
- Individuals with a diagnosis of ASD
- Aged 6 years to adult
- 9 US centers
- IQ of 60+
- Ongoing behavioral and pharmacologic
therapies allowed
- English speaking
- Capable of participation in study procedures
N=144 ASD Treatment
- No restrictions on concurrent therapies
- 3 site visits w/ BSWB (BL, Wk 4, Endpoint)
14-day Screening Phase
8 or 10 Week Prospective Monitoring N=41: Typically Developing Controls
- BSWB, rating scales, single assessment
Single visit 8-10 weeks
ASD TD N 144 41 Male le 112 (77.8) 27 (65.9) Mea ean Age e (SD (SD) 14.58 (7.83) 16.27 (13.176)
SLIDE 10
Autism Behavior Inventory (ABI; v1.1)
10
1 Bangerter et al (2019). J Autism Dev Disorders, online at https://doi.org/10.1007/s10803-019-03965-7 2 Data on file
Core Domains Associated Domains
5 Symptom Domains
Moo
- od &
& Anx nxie iety
- 62
62-item scale le (mon (monthly ly)
- 24
24-item Sh Short
- rt form
- rm (bi
(bi-weekly ly)
- Ra
Rated 0 to
- 3 (0
(0 to
- ma
max.)
Self lf Regula latio ion Chall allenging Behavior
- rs
Soc
- cia
ial l Com
- mmunic
icatio ion Restric icted Behavior
- r
ABI
ABI / Other Scale Domain
- Correl. (r)
ABI Core --- SRS Total 0.81 ABI-RRB --- RBSR Total 0.76 ABI-RRB --- SRS RI, RB 0.75 ABI-SC --- SRS SC, I 0.66 ABI / Other Scale Domain
- Correl. (r)
ABI-MA --- CASI Anxiety 0.77 ABI-SR --- ABC Hyper, NonCom 0.88 ABI-SR --- ABC Inapp Speech 0.64 ABI-CB --- ABC Irritability 0.76
SLIDE 11
11
Visual Exploration Test
Manyakov et al (2018). Autism Research 11: 1554-1566.
11
SLIDE 12
12
Visual Exploration Test – Broad Differences in ASD vs. TD
Exploration (#/% valid time)
- Less exploration of all images
- Greater perseveration on high autism interest items
- Perseveration on HAI items in both social and non-
social arrays P<0.001 Participants with ASD (vs. TD) showed significantly: Example: Exploration
Manyakov et al (2018). Autism Research 11: 1554-1566.
12
SLIDE 13
13
Visual Exploration Test – Broad Differences in ASD vs. TD
Exploration (#/% valid time)
- Less exploration of all images
- Greater perseveration on high autism interest items
- Perseveration on HAI items in both social and non-
social arrays P<0.001 Participants with ASD (vs. TD) showed significantly: Example: Exploration More complex features may tell us more about how visual perceptual processing affects social behavior … does combining variables or constructing new features result in better as biomarkers ? BUT…. Visual processing is a complex behavioral phenomenon!
Manyakov et al (2018). Autism Research 11: 1554-1566.
13
SLIDE 14
14
Eye-Tracking Data May Benefit from More Complex Analytic Approach
RQA (Recurrence Quantification Analysis) approach to identify eye-tracking patterns
Manyakov et al (2018). Autism Research 11: 1554-1566.
14
SLIDE 15
15
Eye-Tracking Data May Benefit from More Complex Analytic Approach
RQA (Recurrence Quantification Analysis) approach to identify eye-tracking patterns
Manyakov et al (2018). Autism Research 11: 1554-1566.
Recurrence Rate
Patterns and duration of refixations
Center of Recurrence Mass
Timing between fixations
Recurrent fixations close in time Recurrent fixations distant in time RR only RR and detail
- rientation
Determinism
Repetitive patterns of gaze shift
Pairs of consecutive fixations Repeated shift between images
Laminarity
Fixation and rescanning pattern
Single fixation, then rescanned in detail Detailed first fixation, then quickly rescanned 15
SLIDE 16
16
Eye-Tracking Data May Benefit from More Complex Analytic Approach
RQA (Recurrence Quantification Analysis) approach to identify eye-tracking patterns
Manyakov et al (2018). Autism Research 11: 1554-1566.
Recurrence Rate
Patterns and duration of refixations
Center of Recurrence Mass
Timing between fixations
Recurrent fixations close in time Recurrent fixations distant in time RR only RR and detail
- rientation
Determinism
Repetitive patterns of gaze shift
Pairs of consecutive fixations Repeated shift between images
Laminarity
Fixation and rescanning pattern
Single fixation, then rescanned in detail Detailed first fixation, then quickly rescanned Statistically lower in ASD vs. TD, correlates to symptoms 16
SLIDE 17
Funny Videos
Do individuals with ASD have an atypical affective response?
- Measurement of spontaneous affective
response at time proximal to “event”
- May be more relevant to social
exchange than prompted affect
- Not perceived as “tests”
- Appear to have durable response, even
with repetition
Bangerter et al. (submitted).
17
SLIDE 18
18 1 Ban
angerter et al, al, in in pr preparation
Funny Videos – FACET Results
Differential spontaneous affective response in ASD subgroups
- Wide variability in ASD affective
response to funny videos
- ASD – 2 subgroups emerge
- “Over-responders” (red)
- “Under-responders” (blue)
Lip corner puller Cheek raiser
Bangerter et al. (submitted).
18
SLIDE 19
19 1 Ban
angerter et al, al, in in pr preparation
Funny Videos – FACET Results
Differential spontaneous affective response in ASD subgroups
- Wide variability in ASD affective
response to funny videos
- ASD – 2 subgroups emerge
- “Over-responders” (red)
- “Under-responders” (blue)
Lip corner puller Cheek raiser
Under-responders (vs. TD group)*
- AU6 avg: p <0.001, r = - 0.25
- AU12 avg: p <0.001, r = - 0.36
*Linear regression (sex, age)
Over-responders (vs TD group)*
- AU6 avg: p <0.001, r = 0.62
- AU12 avg: p <0.001, r = 0.3
Bangerter et al. (submitted).
19
SLIDE 20
Resting state EEG
- Both eyes open (hourglass) and eyes closed conditions
Janssen Research & Development Confidential Janssen Research & Development – data in preparation
20
SLIDE 21
EEG Power
Resting State, Eyes Open
Hypothesis
Wang et al. 2013
Results
Significant difference is seen mainly in Alpha at posterior regions
Preliminary results Janssen Research & Development – data in preparation
21
SLIDE 22
EEG Power
Resting State, Eyes Open Correlation between alpha power and symptom severity (r~ -.21)
Preliminary results Janssen Research & Development – data in preparation
22
SLIDE 23
EEG Asymmetry
Resting state, left/right spectral power (frontal) Eyes open, frontal region
delta gamma beta alpha theta
In ASD left/right asymmetry is higher than TD (ASD[left/right] < TD[left/right])
Preliminary results
Significant Not significant
Janssen Research & Development – data in preparation
23
SLIDE 24
EEG regional connectivity in ASD vs. TD
Resting State, Eyes Closed Example: Theta, eyes closed ASD > TD: only connections (coherence) with sig. differences shown
- dashed lines – p<0.1
- solid lies – p<0.05)
In ASD, coherence between frontal and other regions is increased in α, θ, δ and decreased in β, γ (vs. TD)
Janssen Research & Development – data in preparation Preliminary results
24
SLIDE 25
25
Data Reduction & Prediction
Analysis of JAKE Biosensor Workbench Data
25
Tasks Activity monitoring Biological motion Social orienting task Social vs. non-social videos NIMSTIM direct/averted gaze Visual exploration task Evoke potential responses Resting state Facial expression Funny videos Auditory stimuli
Jagannatha et al (2019). Stat Biopharm Res, DOI: 10.1080/19466315.2018.1527247
# Features 539 3738 972 4502 12741 859 2094 3626 342 684 1014 Total = 31,111
Eliminate near zero variance = 25,022 Eliminate Collinearity w/in each task = 9,705 Eliminate Collinearity across tasks = 3,827 Eliminate features missing >50% of data = 9,705 Imputation possible only w/ ~70% non-missing obsns = 2,569 Sensors = EEG, ECG, Eye- tracking, FACET, EDA
Top 20 Features explain 80% of ASD vs. TD variance
3 common variables
- VET: ECG heart rate
- VET: social images explored
- Social videos: % face view
25
SLIDE 26
Conclusions
- ASD is complex, heterogeneity affects
development new, effective therapies
- Digital medicine approaches may
streamline data collection (web, mobile), produce higher quality data
- Biomarkers research should include single
major features, multiple features within tasks, and combination approaches
- Sensor-based technologies may allow
more objective means of stratification and measurement of change
26