Dig igital Phenotyping for Psychiatric Dis isorders: Ready for Primetime? Gregory P. Strauss, Ph.D. Assistant Professor Department of Psychology University of Georgia Email: gstrauss@uga.edu
Disclosures ACKNOWLEDGMENTS & DISCLOSURES ▪ Receive royalties and consultation fees from ProPhase LLC in connection with commercial use of the BNSS and other professional activities; these fees are donated to the Brain and Behavior Research Foundation. ▪ Last 12 Months: Speaking/consultation with Minerva, Lundbeck, Acadia ▪ Grant support from NIH and Brain & Behavior Research Foundation
LIM LIMITATIONS OF F CLIN LINICAL RATING ADVANTAGES OF F DIG IGITAL PHENOTYPING SCALES ALL rating scales have certain limitations PRACTICAL & COST EFFECTIVE : Data collection and ▪ Cognitive impairment analysis can be automated, reducing cost of rater ▪ Rater biases training, time and processing. ▪ Social desirability ▪ Halo effects that limit precision ASSESSMENT IN REAL TIME: Avoids influence of ▪ Arbitrariness of item anchors retrospective memory limitations, halo effects, biases, ▪ Limited scope of anchor scores culture and other potential “noise”. ▪ Practicality (training, time, expense) HIGH RESOLUTION/OPTICS: Allows for scalable resolution to view symptom dynamics over user- defined time and situation. SPECIFICITY : Allows high “spectral” resolution for parsing symptoms from each other, and from global cognitive and other impairments. FACILITATES DATA MINING FOR FUTURE ANALYSIS: Raw data is available for data mining, to fuel preliminary analyses, “proof of concept” studies and to uncover subtle medication effects and explain null findings.
DIG IGIT ITAL PHENOTYPING METHODS SPECIAL DATA ANALYSIS/FEATURE DATA DATA COLLECTION DATA TYPES PROCESSING NEEDS EXTRACTION DISSEMINATION Active Digital Self-Report Data None Phenotyping: (automated & “offline”) Subjective Vocal Acoustics User Patient Interface Requires minor human (Requires simple, shareware processing Interface processing (e.g., tools) transcribing, optimizing Lexical Analysis audio, video signal). Active Digital (Requires simple text-search tool). Phenotyping: Proprietary “Dictionary” needed All analyses can be Objective batched and automated Semantic “Coherence” Analysis (Requires proprietary software and and conducted “off - line” (hence, “off the shelf” “corpus”) software solutions can Video Analysis be used). (facial, head, eye analysis) Requires Proprietary software . Geolocation (automated & “offline”) Passive Digital None Phenotyping: Accelerometry Objective (automated & “offline”) Ambient Acoustics (automated & “offline”)
ACT CTIVE DIG IGITAL PHENOTYPING: SURVEYS ▪ 7 Days of EMA ▪ Scheduled within 90 minute epochs 9AM-9PM ▪ 15min window ▪ Takes ~5min per survey to complete
SURVEY ADHERENCE AND ACCEPTABILITY Strauss et al, in prep: Acceptability Study Authors/Year # Surveys per Day/ % Adherence # Days Tolerability How Positive How Negative (0-10) (0-10) Brenner et al., 2014 6/7 98.1% SZ 8.9 (1.0) 1.0 (1.4) Visser et al., 2018 4/6 90.2% CN 8.5 (1.2) 1.4 (1.2) Ben-Zeev et al., 2012 6/7 97.7% Moran, Culbreth, Barch, 2017 Granholm et al., 2013 4/7 72.1% Moran et al., 2016 4/7 80% Sanchez et al., 2014 4/7 80.6% Oorschot et al., 2013 10/6 80% Granholm et al., 2019 7/7 85% Strauss et al., 2019 4/6 90.2% Palmier-Claus et al., 2012 6/7 82% Edwards et al., 2018 7/6 71% Summary: Sum • Compliance is high Summary 5.6 surveys, 6.6 days 84.3% • The method is highly tolerable • Minimally disruptive • Non-completion due to unavailability or missing prompt • High negative sx, older, less educated, lower IQ can complete surveys Correlation w/ # surveys: r = -.07; # days: r = -.12
ACT CTIV IVE DIG IGIT ITAL PHENOTYPING: : SURVEY VALID IDIT ITY Associated with clinical interview based measures of the same construct and reward processing mechanisms underlying the symptoms Moran, Culbreth, Barch, 2017
ACT CTIV IVE DIG IGIT ITAL PHENOTYPING: : DATA FROM VID IDEO • Videos recorded concurrently with Facial Expression, Gesture, Eye movement Analysis each survey • “Give a step by step description of what you did in the last hour. Please talk for a full 30 seconds" • Submitted to automated facial and vocal analysis Lexical Analysis Semantic Analysis Acoustic Analysis Blunted Affect Emotional Alogia Motivated Behavior Mood Symptoms Experience (Mania, Depression) Disorganization Cognition
ACT CTIV IVE DIG IGIT ITAL PHENOTYPING: : DATA FROM VID IDEO Signal is highly variable over time: But shows consistency for context/time: Patients Versus Controls: Mod odel elli ling Group Dif Differences es: 10-folds Training Validation: Predicting Group status 339 control cases 271 patient cases Average Accuracy = 74% Cohen et al., under review)
ACT CTIV IVE DIG IGIT ITAL PHENOTYPING: : DATA FROM VID IDEO BN BNSS: Blu Blunted ed Affect BNSS: Alo BN logia ia EMA Paranoid De Delu lusions EMA Anxie xiety 222 control cases 243 control cases 207 control cases 180 control cases 49 patient cases 28 patient cases 15 patient cases 42 patient cases R 2 = .33 R 2 = .36 R 2 = .18 R 2 = .12 Average Accuracy = 82% Average Accuracy = 91% Average Accuracy = 92% Average Accuracy = 82% Highly correlated with Disorganization Self-Harm Cohen et al., in prep)
ACT CTIV IVE DIG IGIT ITAL PHENOTYPING: : COGNITION NLP: Speech to Text and Semantic Analysis can be automated (> 80% accuracy) NLP-based measures of Cognition Measures of Cognitive State ICC (3 sessions): Mean: 0.68 r with Neuropsych Measures 0.60 Holmlund et al., 2019
PASSIVE DIG IGITAL PHENOTYPING: ACCELEROMETRY ▪ Change in ACL was Representative Data Across 24Hrs Equipment passively measured on the phone. SZ ▪ Participants wore an Empatica wristband, which passively collected ACL as a constant force in g , which was later converted to m/s 2 . ▪ Both measures of ACL were collected over x, y, and z axes, from which a magnitude score was calculated, using: 𝑦 2 +𝑧 2 +𝑨 2 ACL = 3
ACCELEROMETRY Cohen’s d Effect Sizes Validity: At Rest vs Not 0.35 10.7 SZ CN 0.30 10.6 Phone Mean 10.5 0.25 Cohen's d Effect Size 10.4 0.20 10.3 0.15 10.2 0.10 10.1 0.05 10.0 9.9 0.00 Resting Not Resting Phone Mean Phone SD Band Mean Band SD Activity Type ACL Variable ACL Band Mean w/ Negative Symptoms: r = -.56, p<.001 Strauss et al in prep Control: n = 54; Schizophrenia: n = 54
PASSIVE DIG IGITAL PHENOTYPING: GEOLOCATION Variable Abbreviation(s) Definition Home time Amount of time spent at home Distance change Δd Distance traveled in meters from previous sample calculated by Haversine formula Distance from Δmh Meters from home for each sample calculated by home Haversine formula Stationary location NC Number of distinct geographical locations sampled clusters Location variance LV Variance within locations traveled to Entropy ENT Amount of movement and randomness in locations Transition time* TT Time spent moving between distinct locations
GEOLOCATION Depp et al 2019 Strauss et al under review Control (n Schizophre Z, P-Value Cohen’s D Control: n = 54; Schizophrenia: n = 54 = 56) nia (n = 86) Median 23.8 12.3 2.5, <.001 0.80 Cohen’s d Effect Sizes Distance (17.6) (10.4) Traveled Median 19.8 8.1 2.3, <.001 0.88 Distance (16.6) (9.0) from home % Samples 51.1% 74.4% 1.9, p<.001 0.72 at Home (0.38) (0.25) CAINS Negative Symptoms Median Distance Traveled -.35*** Median Distance from home -.35*** % Samples at Home 0.29**
VALIDITY SUMMARY • The active and passive digital phenotyping measures demonstrate validity via: • Group differences between clinical and control groups • Correlations with symptom, functional outcome, and cognitive measures from the same constructs measured via standard clinical instruments • Correlations between active and passive digital phenotyping measures that are temporally proximal
Sensitivity to Change in Clinical Trials • Bell et al (2020) Blended EMA/I for Coping with Voices in Psychosis
Sensitivity to Change in Clinical Trials • Bell et al (2020) Blended EMA/I for Coping with Voices in Psychosis
Sensitivity to Change in Clinical Trials Moore et al., 2016: Anxiety & Depression
Sensitivity to Change in Clinical Trials Munsch et al., 2009: Binge Eating
SUMMARY ▪ DP is a reliable and valid objective measure of psychiatric symptoms and functional outcome ▪ Objective measures overcome several limitations of traditional clinical rating scales ▪ More contextually sensitive and temporally dynamic than clinical rating scales. ▪ Combining active and passive methods may be most powerful approach ▪ Near-continuous recordings and large numbers of samples enhance probability of detecting treatment effects via enhanced power, offering promise for clinical trials. ▪ Evidence for sensitivity to change in clinical trials
LIM IMITATIONS AND IS ISSUES TO BE RESOLVED • Do these tools measure something meaningful to those with psychiatric conditions? • Industry-ready, compliant, validated tools • Norms and scaling • Automated scoring • Data analysis • Data management • Compliance with FDA regulations (audit trails and security) • Optimal use of phone vs other technologies (e.g., band)
Recommend
More recommend
