Visual mapping of body image disturbance in anorexia nervosa reveals objective markers of illness severity

Abstract

Body image disturbance (BID) is a core feature of eating disorders, for which there are few objective markers. We examined the feasibility of a novel digital tool, “Somatomap”, to index BID related to anorexia nervosa (AN) severity. Fifty-five AN inpatients and 55 healthy comparisons (HC) outlined their body concerns on a 2-Dimensional avatar. Next, they indicated sizes/shapes of body parts for their current and ideal body using sliders on a 3-Dimensional avatar. Physical measurements of corresponding body parts, in cm, were collected for reference. We evaluated regional differences in BID using proportional z-scores to generate statistical body maps, and multivariate analysis of covariance to assess perceptual discrepancies for current body, ideal body, and body dissatisfaction. The AN group demonstrated greater regional perceptual inaccuracy for their current body than HC, greater discrepancies between their current and ideal body, and higher body dissatisfaction than HCs. AN body concerns localized disproportionately to the chest and lower abdomen. The number of body concerns and perceptual inaccuracy for individual body parts was strongly associated with Eating Disorder Examination Questionnaire (Global EDE-Q) scores across both groups. Somatomap demonstrated feasibility to capture multidimensional aspects of BID. Several implicit measures were significantly associated with illness severity, suggesting potential utility for identifying objective BID markers.

Full text



Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports
Visual mapping of body image
disturbance in anorexia nervosa
reveals objective markers of illness
severity
Christina Ralph‑Nearman1,2*, Armen C. Arevian3, Scott Moseman4, Megan Sinik1,
Sheridan Chappelle1, Jamie D. Feusner3,6,7 & Sahib S. Khalsa1,5*
Body image disturbance (BID) is a core feature of eating disorders, for which there are few objective
markers. We examined the feasibility of a novel digital tool, “Somatomap”, to index BID related to
anorexia nervosa (AN) severity. Fifty‑ve AN inpatients and 55 healthy comparisons (HC) outlined
their body concerns on a 2‑Dimensional avatar. Next, they indicated sizes/shapes of body parts
for their current and ideal body using sliders on a 3‑Dimensional avatar. Physical measurements of
corresponding body parts, in cm, were collected for reference. We evaluated regional dierences
in BID using proportional z‑scores to generate statistical body maps, and multivariate analysis of
covariance to assess perceptual discrepancies for current body, ideal body, and body dissatisfaction.
The AN group demonstrated greater regional perceptual inaccuracy for their current body than HC,
greater discrepancies between their current and ideal body, and higher body dissatisfaction than
HCs. AN body concerns localized disproportionately to the chest and lower abdomen. The number
of body concerns and perceptual inaccuracy for individual body parts was strongly associated with
Eating Disorder Examination Questionnaire (Global EDE‑Q) scores across both groups. Somatomap
demonstrated feasibility to capture multidimensional aspects of BID. Several implicit measures were
signicantly associated with illness severity, suggesting potential utility for identifying objective BID
markers.
Body image disturbance (BID), dened as disruption of how one’s body size is experienced, is a core diagnostic
feature for anorexia nervosa (AN)1, which is among the most deadly of all psychiatric illnesses1,2. BID is associ-
ated with a higher risk of relapse3,4, which occurs in nearly half of all patients5, suggesting that a greater insight
into the perceptual mechanisms underlying this process may be key for more eective and durable treatments.
BID is a multifaceted construct that can be loosely divided into perceptual6–9 and attitudinal10–12 components.
As there is a subjective element of this construct, it is dicult to assess the more objective sensory and aec-
tive components of how an individual perceives their own body, and few assessment tools exist to capture this.
Clinical measures of BID (e.g., the Body Image States Scale)10, commonly require patients to select from a
predetermined menu of word indicators to describe their mentalized representation of their body. Perceptual
inaccuracy (i.e., discrepancies between the person’s receipt of body signals and their corresponding interpreta-
tion) has been suggested to be a core characteristic of AN for decades13–15, and has been demonstrated empirically
in multiple studies6. ese perceptual distortions lead to both distress and subsequent eating disorder behaviors.
For instance, an individual with AN may view their overall body or certain parts of their body as extremely large,
when it is actually very thin. ese perceptual inaccuracies and distortions are oen related to negative emotions.
Eating disorder behaviors may partly reect attempts to correct misperceptions of body size and shape, and to
temporarily reduce negative emotions, which are ultimately ineective. erefore, assessing the degree of these
perceptual distortions as an indicator of eating disorder illness severity may inform treatment targets. Yet, it is still
OPEN
Laureate Institute for Brain Research (LIBR), Tulsa, OK, USA. Department of Psychological and Brain Sciences,
University of Louisville, Louisville, KY, USA. Jane and Terry Semel Institute for Neuroscience and Human Behavior,
University of California Los Angeles, Los Angeles, CA, USA. Laureate Psychiatric Clinic and Hospital, Tulsa, OK,
USA. Oxley College of Health Sciences, University of Tulsa, Tulsa, OK, USA. Centre for Addiction and Mental
Health, Toronto, ON, Canada. Department of Psychiatry, University of Toronto, Toronto, ON, Canada. *email:
ChristinaRalphNearman@gmail.com; skhalsa@laureateinstitute.org
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
unclear whether and to what degree these abnormalities are due to a primary dysfunction of sensory encoding
or another perceptual misrepresentation mechanism. Moreover, BID in AN has been linked to disrupted activity
and connectivity in visual processing and parietal association networks16–20, raising the possibility that these brain
areas play a disproportionate role in developing and/or maintaining over-estimations of one’s body size and shape.
Visual silhouette assessments21–24 are one commonly used method to overcome reliance on language-based
inferences. Some gure scales have strong psychometric properties, provide quick assessments of overall body
dissatisfaction24, and measure distinct dimensions of body dissatisfaction independently (e.g., adiposity/mus-
cularity)21,22. Yet, they do not measure other details such as the types of body concerns (e.g., fat, acne, sweat),
lack a focus on specic body parts (e.g. abdominal protrusion, thighs, neck, etc.), and they do not specically
measure the associated emotional impact of body concerns. Visual computer-based body-shaping tools have
been developed, but they also typically only assess overall body size perception because they use multiple body
parts that scale together, or else use preset virtual avatars25–29. is precludes identication of which aspects of
the body are key to generating the disturbance, making it impossible to pinpoint the precise body areas that are
misperceived as problematic, distorted, or obese, and to link these misperceptions with negative aect and eat-
ing disorder behaviors. Moreover, most tools also obscure the head/neck, preventing the identication of BID
related to these features. e available studies assessing distinct body part distortions have found that individu-
als with AN overestimate individual body parts more than the whole body size6,30, suggesting a need for visual
tools with this capability.
We recently created a multi-platform digital tool called “Somatomap”, which utilizes a visual representation of
specic body concern areas on 2-dimensional (2D) and 3-dimensional (3D) avatars to facilitate BID assessment
related to specic regions across the entire body. In a proof-of-concept study with female professional fashion
models and nonmodels, this tool identied patterns of body image concerns localized to discrete body areas
(see Ralph-Nearman etal. 201931 for development details). ese results raised the possibility that Somatomap
might have utility for investigation of BID in clinical populations, such as in individuals with eating disorders.
e current study examined the feasibility of measuring BID characteristics in individuals with AN using
Somatomap. In transitioning Somatomap for use from a nonclinical (female fashion models31) to a clinical sample
we made several renements to the original tool, including increasing the number of 3D body part measure-
ments to 23 independently manipulatable body areas and assessing additional body image measures (ideal body
image perception). is allowed us to estimate a measure of “body dissatisfaction”, dened by subtracting the
ideal from the currently perceived body rating. We also evaluated the relationship between Somatomap BID
measures and traditional estimates of illness severity (i.e., EDE-Q). We predicted that inpatients with AN (relative
to healthy comparisons, HC) would exhibit greater BID on the visual measures in Somatomap as evidenced by a
signicantly higher proportion of body image concern areas (Somatomap 2D), and signicantly lower percep-
tual accuracy for the size/shape of individual body parts (Somatomap 3D). We also predicted that the degree
of BID on Somatomap 2D and 3D would be associated with eating disorder symptom severity as measured by
the EDE-Q. Finally, we employed a usability scale with the prediction that both groups would report acceptable
levels of usability for each measure.
Methods
Participants. We recruited individuals diagnosed with AN from the Laureate Eating Disorders Inpatient
Program in Tulsa, Oklahoma, and HCs from the Tulsa community. Clinical inclusion criteria included a primary
diagnosis of AN by the treating psychiatrist as noted in their medical records during admission, body mass index
(BMI)
≥
14, female sex (a requirement for treatment), age from 13 to 64years, and independent ambulation.
Exclusion criteria included active suicidal ideation or a comorbid psychotic disorder. HCs were recruited from
the local community, screened for the presence of DSM-5 diagnoses via the MINI structured clinical interview32,
and met the eligibility criteria for age, sex, and independent ambulation.
BID measurement via Somatomap. Somatomap 2D. Participants rst used a mouse to outline an area
of body concern directly on a 2D human avatar presented on a computer screen. ey subsequently selected the
specic concerns (e.g., “too fat”, “acne”, etc.) and the emotions they related to each area (e.g., “fearful”, “disgusted”,
“other [be specic]”, etc.) from a menu of visual icons. In addition to the 39 specic types of body concerns and
the 26 specic emotions icon word-pairs they could select from, a free text-entry option was available for par-
ticipants to indicate any other specic body concern areas or any other specic emotions related to their body
concerns not listed in the menu.
Participants also rated the level of defectiveness, emotion intensity, and perceived distress surrounding their
body concern using visual-analogue scales (from 0 = not at all to 100 = extremely). is process was repeated
for each individual body concern. ere was no limit on the number of body concerns that they could outline
and rate, and no time limit.
Somatomap 3D. Aer personalizing the avatar to match their own hair and skin color, the 3D avatar was dis-
played on screen. e size/shape of the 23 adjustable body parts were randomly permuted for the initial avatar
to prevent potential priming eects associated with viewing anatomically proportioned bodies33. Participants
used sliders to adjust the shape of each 3D avatar body area independently, in order to show how they perceived
their current body, and then again to show how they would like their ideal body to look. Adjustable body areas
included: neck girth, neck length, shoulder width, bust girth, chest girth, biceps girth, upper arms length, fore-
arms girth, lower arms length, wrists girth, hands girth, hands length, torso length, waist, stomach form, hips,
thighs girth, thighs length, calves girth, calves length, ankles, feet width, and feet length.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
e protocol for obtaining the physical body measurements of participants was applied from our prior
Somatomap study31 (adapted from the PhenX toolbox34), with the inclusion of 16 additional body areas for a
total of 23 body areas. Areas with multiple parts (e.g., hands, feet, thighs, biceps, forearms, etc.) were measured
individually and then averaged. We used the InBody bioimpedance scale (Cerritos, California) and a stadiometer
to calculate each participant’s actual body mass index (BMI).
Eating disorder symptom assessment. To estimate the degree of eating disorder symptom severity,
each participant completed the Eating Disorder Examination Questionnaire (EDE-Q 6.0)35, which includes four
subscales: Shape, Weight, Eating, and Restraint concern. A summed average of the four subscale ratings yields
an EDE-Q Global score; all scores range across a scale from 0-none to 6-severe. Scores ≥ 4.0 have oen been
considered to reect clinically severe levels of eating disorder psychopathology36,37.
Data collection procedure. is study was approved by the Western Institutional Review Board, and all
methods were carried out in accordance with relevant guidelines and regulations. Prior to the experiment, each
participant provided written informed consent, and informed consent was obtained from a parent and/or legal
guardian for the participants under 18years of age (Trial Registration: ClinicalTrials.gov #NCT03758326). All
participants completed demographic questions, BID assessments (Somatomap 2D and 3D in counterbalanced
order), and self-report scales at LIBR on a computer. User experience surveys were collected aer completing
the 2D and 3D measures, and physical measurements were completed last. e data for the Somatomap app
and survey was collected on the Chorus app platform38. Chorus is a visual development platform supporting
the creation of web-based digital health applications that can be accessed on various devices including mobile,
tablet and desktop.
Statistical analysis. Somatomap 2D. To create group-level proportional maps of body concerns, all pix-
elwise body concern tracings were merged for each person (overlapping pixels were set to 1) and the ensuing
binary maps were overlapped as in our prior Somatomap study procedure31. To evaluate between-group dier-
ences in the body maps we used custom Matlab soware (Mathworks, Inc.) to calculate the test statistic for each
pixel using the z-formula for proportion, and employed cluster correction (6 pixel threshold) and smoothing
following our previously described procedures31,39,40. Across both groups, a linear regression model (lm) con-
ducted in R (version 3.6.2) examined the relationship between the number of individual body concerns traced
and eating disorder symptomatology (EDE-Q Global Scores).
Somatomap 3D. e 3D assessment procedure yielded six distinct scores across each of the 23 body parts: (1)
actual body, (2) current perceived body, (3) ideal perceived body, (4) current body discrepancy (2 − 1), (5) ideal
body discrepancy (3 − 1), and (6) body dissatisfaction (3 − 2). Both current and ideal perceived body measure-
ment scores were converted into centimeters from arbitrary units using piecewise linear interpolation as previ-
ously described31. Similar to our previous study, we used multivariate analysis of covariance (MANCOVA) to
determine if there were group dierences in each of these measurements. Covariates included BMI, height, and
weight. If the MANCOVA results were signicant, a follow-up analysis using analysis of covariance (ANCOVA)
with the Benjamini–Hochberg multiple comparison correction was used to determine which variables showed
statistically signicant dierences between ANs and HCs. Multiple linear regression models were used to exam-
ine relationships between body dissatisfaction body part measures and body discrepancy body part measures
(separately) with measures of psychopathology severity (EDE-Q), across both groups. Multiple linear regression
models were used to examine whether intersecting measures on 2D and 3D together might explain more vari-
ance of symptom severity than when applied with 2D concerns and 3D body dissatisfaction and body discrep-
ancy separately.
User experience. Usability surveys were collected aer completing both measures to assess the ease and enjoy-
ability of use, the degree to which the 2D tool eectively captured participants’ BID concerns and associated
emotions, and the degree of identication with the original and nal 3D avatar.
Results
Participant demographics. Fiy-ve AN and 55 HC females completed testing (see Table1 for key demo-
graphic data; Fig.S.1. for Consort diagram). e AN group had a signicantly lower BMI than HC (P < 0.001),
which was driven by dierences in body weight (P < 0.001) but not height (P = 0.70). No signicant dierences
for overall race/ethnicity (χ24 = 7.36; P = 0.061), or education levels (χ25 = 10.90; P = 0.053) between the groups
were identied.
Somatomap 2D. Proportional body maps showed that AN participants perceived body concerns across
a broader area compared with HCs (Fig.1A). e statistical body map analysis revealed that the AN group
perceived signicantly more concerns localized to the lower abdomen and chest than HCs (P < 0.001) (Fig.1B).
e number of body concerns outlined ranged from 1 to 12 per individual for ANs (M = 3.20; SD = 2.45) and
from 1 to 7 per individual for HCs (M = 1.55; SD = 1.17), which signicantly diered between groups (t108 = 4.52,
P < 0.001). Across both groups, a linear regression revealed a signicant relationship between the Global EDE-Q
score and the number of body concern areas outlined (R2 = 0.24, F = 33.67, P < 0.001; Fig.1C). EDE-Q Global
concerns were reliable (Cronbach’s
α
=0.98). Participants took an average of 2.2min (SD = 1.6) to complete
Somatomap 2D (AN M = 2.6, SD = 1.8; HC M = 1.8, SD = 1.3).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
e number of body concern types identied (e.g., “too fat”, “bloated”) ranged from 1 to 73 per individual for
ANs (M = 15.65; SD = 15.74) and from 0 to 20 per individual for HCs (M = 3.53; SD = 4.11), which was signi-
cantly dierent between groups (t108 = 1.98, P < 0.001). e number of aective labels selected (e.g., “frustrated”,
“angry”, “disgusted”, etc.) for all concerns ranged from 0 to 93 per individual for ANs (M = 16.3; SD = 18.2) and
from 1 to 22 per individual for HCs (M = 2.6; SD = 3.9); which was also signicantly dierent between groups
(t109 = 5.4, P < 0.001; TableS.1).
Emotion ratings associated with body concerns were signicantly higher for ANs relative to HCs (Ps < 0.001)
with regards to the level of defectiveness (AN (M = 60.70; SD = 23.24); HC (M = 21.34; SD = 16.26)), emotion
intensity (AN (M = 56.70; SD = 26.05); HC (M = 23.35; SD = 27.73)), and degree of distress caused by body con-
cerns (AN (M = 63.25; SD = 25.61); HC (M = 12.33; SD = 14.42)).
Somatomap 3D. Actual body, current perceived body, and ideal perceived body. ere were multiple sta-
tistically signicant group dierences in the measurements of the actual body, current perceived body, and ideal
perceived body. A complete listing of these results is provided in the Supplement. Here, we summarize the sta-
tistically signicant dierences observed. e AN group had a signicantly smaller measured bust girth, chest,
biceps, waist, “stomach form” (abdominal protrusion), thighs, and calves, and had a shorter upper arm length
compared with HCs (MANCOVA: F(23, 83) = 33.73, Wilks Λ = 0.097, P < 0.001; TableS.2). However, they per-
ceived various current body girths to be signicantly larger than the HC group (i.e., chest, biceps, waist, stomach
form, thighs, calves, neck, forearms, hips, and ankles; MANCOVA: F(23, 83) = 1.73, Wilks Λ = 0.675, P = 0.037;
TableS.3). e AN and HC groups showed few dierences in idealized body girth and length characteristics,
except that the AN group desired narrower shoulders, a longer torso, smaller hips, calves, and feet width than
the HC group (MANCOVA: F(23, 83) = 3.40, Wilks Λ = 0.515, P < 0.001; TableS.4).
Current body discrepancy, ideal body discrepancy, and body dissatisfaction. ere were multiple statistically
signicant dierences in the current body discrepancy, ideal body discrepancy, and body dissatisfaction meas-
urements. A complete listing of these results is provided in the Supplement. In summary, the AN group had
signicantly higher discrepancy scores (overestimated current body sizes) for neck girth, biceps, stomach form,
Table 1. Demographic and clinical characteristics of anorexia nervosa inpatients and healthy comparisons.
GED general educational development, OCD obsessive compulsive disorder, PTSD posttraumatic stress
disorder. a Hispanic/Latino descent, Lebanese, American Indian/Alaska Native, Native Hawaiian or other
Pacic Islander, Jewish, Black, White/Caucasian, Asian (including East Indian) listings per the PhenX34. b For
brevity, only comorbid diagnoses with > 10% frequency are listed.
AN, mean (SD) HC, mean (SD) t (df)P-value
Gender 55 females 55 females
Age (years) 25.25 (11.00) 23.42 (4.98) 1.13 (75) 0.26
Height (cm) 163.92 (6.60) 164.47 (8.40) − 0.38 (102) 0.70
Weight (kg) 50.67 (7.75) 61.13 (7.27) − 7.30 (108) < 0.001
Body mass index (kg/m2) 18.85 (2.83) 22.61 (2.25) − 7.70 (103) < 0.001
Race/ethnicity (n, %) χ24 = 7.36 0.061
White/Caucasian 50 (90.90) 40 (72.73)
Black 0 (0) 2 (3.63)
Asian (including East Indian) 0 (0) 2 (3.63)
More than one race/ethnicitya5 (9.10) 11 (20.00)
Education χ25 = 10.90 0.053
Graduate school 7 (12.73) 8 (14.55)
University graduate 9 (16.36) 19 (34.55)
Some university 22 (40.00) 23 (41.82)
High school/A level/GED 11 (20.00) 3 (5.45)
Some high school/A level 4 (7.27) 2 (3.64)
Less than high school/A level 2 (3.64) 0 (0)
EDE-Q Global 4.25 (1.09) 0.65 (0.66) 20.95 (108) < 0.001
Age of onset (years) 14.7 (3.1) –
Illness duration (years) 10.2 (11.4) –
Psychotropic medication (%) 89.1 –
Comorbid diagnoses (n, %)b
Major depressive disorder 24 (42.6) –
Generalized anxiety disorder 21 (38.2) –
OCD 17 (30.9) –
PTSD 9 (16.4) –
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
hips, thighs, calves, ankles, feet length, and overall body compared to HCs (MANCOVA: F(23, 83) = 11.89, Wilks
Λ = 0.232, P < 0.001; Fig.2, TableS.5).
Relative to HCs, the AN group had signicant dierences in ideal body discrepancy scores such that they
desired narrower shoulders, a thinner chest, a larger bust, a thinner waist, narrower feet, longer upper arms,
torsos, and hands, compared with their actual body part sizes and shapes. Both groups desired a thinner stomach
form, but the AN group desired it to be signicantly thinner relative to their actual body (MANCOVA: F(23,
83) = 6.37, Wilks Λ = 0.362, P < 0.001; Fig.3, TableS.6).
e AN group showed signicantly greater body dissatisfaction scores for neck girth, chest, biceps, forearms,
waist, hips, thighs, calves, ankles, feet width, and overall body (towards the thin ideal) relative to the HC group
(MANCOVA: F(23, 83) = 2.59, Wilks Λ = 0.582, P < 0.001; Fig.4, TableS.7). ey also had greater dissatisfaction
for their shoulders (preferring them narrower), torso and calves (preferring them longer) (see Table2 for a sum-
mary of signicant group dierences for each of the six measures). Somatomap 3D took participants 6.6min on
average to complete (AN M = 6.9; SD = 2.2; HC M = 6.4; SD = 2.0).
Regression analysis. 3D body dissatisfaction and ED severity. We conducted a multiple linear regression
using the EDE-Q Global as the dependent variable, with body dissatisfaction scores on 3D for all body parts
as independent variables, across groups. Body dissatisfaction parts on 3D were signicantly associated with
Figure1. (A) (top le) Proportional body maps displaying the majority of body concern areas for each group
(le anorexia nervosa (AN) group; middle healthy comparisons (HC)). (B) (top right) Statistical body map
evaluating dierences in body image concerns between AN group (in cool colors) and HCs (in warm colors;
statistical threshold = P < 0.001); (C) (bottom) Association between Number of Body Concerns and Eating
Disorder Symptomatology (EDE-Q Global Scores) across ANs and HCs (P < 0.001).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
Figure2. Summary of current body discrepancy score dierences between AN and HC. Of the total set of
23 body part measurements only those showing signicant dierences are displayed (P < 0.05, corrected for
multiple comparisons using the Benjamini–Hochberg procedure). Positive values indicate overestimation of
true body part size, negative values indicate underestimation, and zero indicates correct estimation. Standard
errors are represented by error bars.
Figure3. Ideal body discrepancy (ideal body minus actual body measurements (cm)) in female AN and HC
groups. Of the total set of 23 body part measurements only those showing signicant dierences are displayed
(P < 0.05, corrected for multiple comparisons using the Benjamini–Hochberg procedure). Positive values
indicate ideal body part size is larger than true body part size, negative values indicate ideal body part size is
smaller than true body part size, and zero indicates the ideal and true body part size is the same. Standard errors
are represented by error bars.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
participants’ eating disorder psychopathology severity on the EDE-Q (R2 = 0.50, F(23,86) = 3.71, P < 0.001) (see
Fig.S.2).
3D current body discrepancy and ED severity. Next, we conducted a multiple linear regression to investigate
the relationships between EDE-Q scores and the current body discrepancy scores on 3D for all body parts,
across groups. Body discrepancy on 3D was signicantly associated with eating disorder symptom severity on
the EDE-Q, accounting for 67% of the variance of global psychopathology (R2 = 0.67; F(23,86) = 7.57, P < 0.001)
(see Fig.S.3).
Aggregated 2D with 3D body dissatisfaction and ED severity. To explore the degree to which aggregating the
number of body concerns on 2D and all of the body dissatisfaction parts on 3D, were related to eating disorder
severity, we conducted a multiple linear regression using the EDE-Q Global scores as the dependent variable,
with the number of 2D body concerns plus all of the body dissatisfaction parts on 3D as independent variables,
across groups. Together, the number of 2D concerns with all of the 3D body dissatisfaction body parts explained
59% of the variance of overall ED psychopathology (R2 = 0.59, F(24,85) = 5.03, P < 0.001).
3D chest and waist body dissatisfaction with number of 2D concerns and ED severity. Chest and waist were
body areas identied by both the 2D and 3D approaches as being signicantly related to BID in AN relative to
HCs. We conducted a multiple linear regression between the number of 2D body concerns and 3D chest and
waist body dissatisfaction and the EDE-Q Global scores, which explained 40% of the variance of overall ED
psychopathology (R2 = 0.40, F(3,106) = 23.4, P < 0.001).
Aggregated 3D current body discrepancy with number of 2D concerns and ED severity. Finally, to examine the
degree to which aggregating relevant 2D and 3D measures explained symptom severity, we conducted a multiple
linear regression using the EDE-Q Global as the dependent variable, with the number of 2D body concerns plus
all of the current body discrepancy parts on 3D as independent variables, across groups. Together, the number of
2D concerns with all of the 3D current body discrepancy body parts explained a high percentage of the variance
Figure4. Body Dissatisfaction Score (ideal minus current perceived (cm)) in female AN and HC groups. All
body parts refer to girth except where indicated for length. Of the total set of 23 body part measurements only
those showing signicant dierences are displayed (P < 0.05, corrected for multiple comparisons using the
Benjamini–Hochberg procedure). Standard errors are represented by error bars.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
Table 2. Summary of statistically signicant AN vs. HC body part dierences across all six Somatomap 3D
measures. 23 body part assessments (15 girth/8 length) were examined for each measure.
Measure Girth Co hen’s f Length C ohe n’s f
Actual body
AN narrower than HC AN shorter than HC
Bust 0.37 Upper arms 0.387
Chest 0.350
Biceps 0.318
Wai st 0.288
Stomach form 0.405
ighs 0.367
Calves 0.299
Current perceived
AN overestimated No dierences
Neck 0.246
Chest 0.351
Biceps 0.358
Forearms 0.275
Wai st 0.359
Stomach form 0.238
Hips 0.282
ighs 0.296
Calves 0.318
Ankles 0.350
Current body discrepancy
AN overestimated AN overestimated
Neck 0.289 Feet length 0.157
Biceps 0.481
Stomach form 0.447
Hips 0.308
ighs 0.495
Calves 0.403
AN underestimated
Ankles 0.264
Ideal perceived
AN desired smaller AN desired longer
Shoulders 0.311 Torso 0.243
Hips 0.269
Calves 0.245
Feet width 0.258
Ideal body discrepancy
AN desired smaller AN desired longer
Shoulders 0.316 Torso 0.307
Chest 0.240 Upper arms 0.306
Wai st 0.335 Hands 0.238
Stomach form 0.302
Feet width 0.252
AN desired larger
Bust 0.374
Body dissatisfaction
AN desired smaller AN desired longer
Neck 0.252 Torso 0.257
Shoulders 0.238 Calves 0.251
Chest 0.348
Biceps 0.345
Forearms 0.243
Wai st 0.401
Hips 0.420
ighs 0.351
Calves 0.426
Ankles 0.273
Feet width 0.296
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
of overall ED psychopathology (R2 =0.73; F(24,85) = 9.73, P < 0.001). All pre-checks for all regression models
performed showed minimal concern evidence for of multicollinearity (variance ination factor < 5).
Usability assessment. Participants rated the 2D body map and 3D avatar as easy to use, as very reective of their
body concerns (2D), and that the nal 3D avatar was mostly reective of their actual perceived body (TablesS.8,
S.9). HCs reported signicantly higher levels of ease (2D: t91 = 2.48, P = 0.015; 3D: t108 = 2.69, P = 0.008), and
enjoyment (2D: t91 = 4.04, P = 0.001; 3D: t108 = 3.69, P < 0.001), and how closely the nal 3D avatar reected their
body (t108 = 2.92, P < 0.004) than the AN group. Participants “liked least” that Somatomap: “was lacking muscle
tone”, “started randomized”, and “liked best” that it was: “easy to use”, “enabled me to express myself”, and “very
realistic”.
Discussion
e current study tested the feasibility of a novel digital tool for assessing multidimensional BID characteris-
tics, including objective measurements of perceptual disturbance for body part size estimation and subjective
experiences of dissatisfaction, in acutely ill individuals with AN relative to HCs. We also tested the relationships
between these measures and illness severity estimates. We found, as hypothesized, that there were signicant
BID dierences between groups in terms of perceptual overestimations, thin-ideal body dissatisfaction, and body
concerns (for individual body part size/shape on Somatomap 3D as well as the total number of body concerns
on Somatomap 2D), and in terms of the aective valence associated with body concerns. Finally, Somatomap
2D and 3D dierence measures were associated with symptom severity measures across groups when examined
individually and when aggregrated, with aggregated measures explaining the greatest symptom variance.
Analysis of the Somatomap 2D data revealed that this tool detected a greater degree of BID concerns in
AN than HCs for the chest and abdomen, with greater levels of perceived distress, intensity, and defectiveness
related to AN body concerns. Moreover, across both groups, the number of outlined body concerns was posi-
tively associated with eating disorder symptom severity on the EDE-Q when controlling for age and BMI. is
result raises the intriguing notion that a single perceptual rating about body concerns (involving a few mouse
clicks and minutes) could provide information about an individual’s eating disorder psychopathology sever-
ity. Such an implicit approach could potentially augment or replace more time-consuming screening methods
which explicitly signal the search for a disorder, although additional study is needed to verify this possibility.
At a minimum, this approach provides a clear indication of the visual perceptual body mapping and associated
emotional characteristics related to BID in AN, at the individual and group levels.
Analysis of the Somatomap 3D data identied that, overall, the AN group (relative to HCs) showed a general
over-estimation of current body girth and a preference for a thinner body than what they currently perceived.
is was apparent from the current perceived body, current body discrepancy, ideal perceived body discrepancy,
and body dissatisfaction scores, which were associated with symptom severity. ese ndings support the value
of assessments of individual body parts to better understand perceptual and attitudinal aspects of BID. ey raise
the possibility of impaired cognitive and/or visual mechanisms operating in individuals with AN7,20,41, which
might decrease their ability to accurately perceive their own body.
Our observation that the AN group overestimated body sizes and reported increased negative emotion
compared with HCs supports the self-discrepancy theory42, which suggests that mismatches between actual
and ideal internal self-representations produce negative aect and poor health outcomes. ese results also sup-
port a recent meta-analysis which found that more negative and less positive aect ranges for self-discrepancy
were signicantly related to psychopathology43. We might speculate that the AN inpatients’ overestimations of
their body size compared to HCs in our current study indicates an inability or diculty to calibrate their body
perception in relation to the general population. Another possibility is that they may ‘re-calibrate’ their body
perception towards other underweight individuals present in their current treatment setting. e development
and/or maintenance of body size and shape over-estimations in AN may be related to previously demonstrated
brain activation and connectivity abnormalities in visual systems. For example, BID in AN may also be due to a
disturbance in brain connectivity and abnormal functioning in body processing networks44,45. is might involve
reduced connectivity between the le fusiform body area and extrastriate body area44–46, reduced neural activity
in cortical visual systems and hyperconnectivity in dorsal visual and parietal networks20, which may be associ-
ated with body size misperception44–46 in AN. e connectivity of these networks could be further investigated
in future studies utilizing Somatomap. For a discussion of the relevance of these ndings to theories of body
perception in HCs, see the Supplement.
Two truncal areas of AN body concern were identied across Somatomap 2D and 3D: the abdomen and chest.
Excessive fullness and/or bloating of the stomach and intestines is a frequent complaint for individuals with AN,
which are oen identied in clinical settings as motivators of restrained eating25. e chest is another focus area
given that it is a region containing adipose tissue in females. Both regions show a lower percentage of body fat
in the underweight stage followed by an increased ratio of truncal/extremity fat during weight restoration26,
which may help to explain the common overlap across measures. Our exploratory aggregated regression analy-
ses revealed that these measures explained a substantial degree of symptom severity on the EDE-Q. Specically,
while the chest and waist body dissatisfaction (3D) together with the number of 2D body concerns explained
40% of the variance of EDE-Q, and body dissatisfaction parts (3D) together with the number of 2D body con-
cerns explained 59% of the EDE-Q variance, it was the 3D current discrepancy body parts (together with the
number of 2D body concerns) that explained the most variance of symptom severity: 73% of the EDE-Q Global
score variance. Although exploratory, these results suggest that combining convergent measures focusing on
the visual perceptual characteristics or body dissatisfaction of discrete body parts, and total number of body
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
parts of concern, might be used to make meaningful inferences about the degree of symptom severity across the
spectrum of eating psychopathology.
Objectively distinguishing the perceptual and aective components of BID may be critical to better under-
standing AN, in terms of screening, prevention, diagnostic assessment, prognostic prediction, intervention
targeting, and relapse prevention monitoring. Clinicians currently measure BID by relying upon subjective,
language-based questionnaires or visual analogue scales which assess body dissatisfaction and body distortion
constructs in a global manner. e present study tested the feasibility of a dierent approach, one that included
quantication of visual perceptual distortions and body dissatisfaction for discrete body parts. is revealed: (1)
both Somatomap 2D and 3D detected visual perceptual and aective aspects of BID in AN, explained unique
variance in eating disorder symptom severity and greater variance when combined across groups, (2) both were
tolerable, with no one quitting the measurement during test/retest, (3) usability was generally good, although the
signicantly reduced usability rating in AN (compared with HCs) means there is room for participant-suggested
improvements, and (4) Use of a multi-platform digital tool means it can be remotely deployed such as in post-
acute settings for monitoring and detecting illness trajectory. In addition to providing more specic information
about distinct combinations of BID and illness severity, this tool may equip researchers to visually, statistically,
and remotely pinpoint BID parameters that are uniquely related to ED severity.
Limitations, strengths and future directions. is study has several limitations. First, AN participa-
tion in this study was limited to females from a single inpatient treatment facility, and therefore, testing with a
broader range of facilities, care settings, and demographic characteristics (including males) would be important
to demonstrate generalizability. Second, although enjoyability of the tool was rated by AN participants above
average, this experience could be further improved (particularly in the AN patient group) by incorporating AN
participants’ usability assessment feedback to include other dimensions of the body (e.g., they suggested adding
muscularity). Longitudinal and reliability studies will be necessary to test the tool’s ability to detect changes in
illness severity (e.g., relapse/remission/recovery), and comparing this assessment with other measures of BID
to highlight the incremental validity will be essential. An additional approach could evaluate how changes in
BID relate to (or predict) longitudinal clinical outcomes. A strength of the current approach is that it eectively
captures a broad range of body dissatisfaction self-concepts in relation individual body parts, and illustrates
their association with illness severity. e digital multi-platform nature of this tool could have broad applications
including remote deployment, in multiple clinical settings.
Conclusion. e current study demonstrates the initial feasibility of Somatomap for assessing multidimen-
sional aspects of BID related to ED severity in AN. e novel visual perceptual mapping approach entailed by
this tool can capture implicit responses and estimate perceptual disturbances at the level of individual body
parts, yielding objective markers of BID. is positions Somatomap as a potentially useful tool both for research
and clinical applications.
Received: 28 October 2020; Accepted: 7 May 2021
References
1. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (American Psychiatric Pub,
2013).
2. Arcelus, J., Mitchell, A. J., Wales, J. & Nielsen, S. Mortality rates in patients with anorexia nervosa and other eating disorders. Arch
Gen Psychiatry 68(7), 724. https:// doi. org/ 10. 1001/ archg enpsy chiat ry. 2011. 74 (2011).
3. Channon, S. & DeSilva, W. P. Psychological correlates of weight gain in patients with anorexia nervosa. In Anorexia Nervosa and
Bulimic Disorders (eds Channon, S. & DeSilva, W. P.) 267–271 (Pergamon, 1986).
4. Keel, P. K., Dorer, D. J., Franko, D. L., Jackson, S. C. & Herzog, D. B. Postremission predictors of relapse in women with eating
disorders. Am. J. Psychiatry 162(12), 2263–2268. https:// doi. org/ 10. 1176/ appi. ajp. 162. 12. 2263 (2005).
5. Khalsa, S., Portno, L. C., McCurdy-McKinnon, D. & Feusner, J. What happens aer treatment? A systematic review of relapse,
remission, and recovery in anorexia nervosa. J. Eat. Disord. 5(1), 20. https:// doi. org/ 10. 1186/ s40337- 017- 0145-3 (2017).
6. Gardner, R. M. & Brown, D. L. Body size estimation in anorexia nervosa: A brief review of ndings from 2003 through 2013.
Psychiatry Res. 219(3), 407–410. https:// doi. org/ 10. 1016/j. psych res. 2014. 06. 029 (2014).
7. Madsen, S. K., Bohon, C. & Feusner, J. D. Visual processing in anorexia nervosa and body dysmorphic disorder: Similarities,
dierences, and future research directions. J. Psychiatr. Res. 47(10), 1483–1491. https:// doi. org/ 10. 1016/j. jpsyc hires. 2013. 06. 003
(2013).
8. Metral, M. et al. Painfully thin but locked inside a fatter body: Abnormalities in both anticipation and execution of action in
anorexia nervosa. BMC Res. Notes 7(1), 707 (2014).
9. Spitoni, G. F. et al. e two dimensions of the body representation in women suering from Anorexia Nervosa. Psychiatry Res.
230(2), 181–188. https:// doi. org/ 10. 1016/j. psych res. 2015. 08. 036 (2015).
10. Cash, T., Fleming, E., Alindogan, J., Steadman, L. & Whitehead, A. Beyond body image as a trait: e development and validation
of the body image states scale. Eat. Disord. 10(2), 103–113. https:// doi. org/ 10. 1080/ 10640 26029 00816 78 (2002).
11. Stice, E. Risk and maintenance factors for eating pathology: A meta-analytic review. Psychol. Bull. 128(5), 825. https:// doi. org/ 10.
1037/ 0033- 2909. 128.5. 825 (2002).
12. Vitousek, K. B. & Hollon, S. D. e investigation of schematic content and processing in eating disorders. Cogn. er. Res. 14(2),
191–214 (1990).
13. Gardner, R. M., & Moncrie, C. Body image distortion in anorexics as a non-sensory phenomenon: A signal detection approach.
J. Clin. Psychol. 44,(2), 101–107. https:// doi. org/ 10. 1002/ 1097- 4679(198803) 44: 2< 101:: aid- jclp2 27044 0203>3. 0. co;2-u (1988).
14. Molinari, E. Body-size estimation in anorexia nervosa. Percept. Motor Skills. 81(1), 23–31. https:// doi. org/ 10. 2466/ pms. 1995. 81.1.
23 (1995).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
15. Touyz, S. W., Beumont, P. J., Collins, J. K., McCabe, M. & Jupp, J. Body shape perception and its disturbance in anorexia nervosa.
Br. J. Psychiatry 144(2), 167–171. https:// doi. org/ 10. 1192/ bjp. 144.2. 167 (1984).
16. Feusner, J., Deshpande, R. & Strober, M. A translational neuroscience approach to body image disturbance and its remediation in
anorexia nervosa. Int. J. Eat. Disord. https:// doi. org/ 10. 1002/ eat. 22742 (2017).
17. Li, W. et al. Anorexia nervosa and body dysmorphic disorder are associated with abnormalities in processing visual information.
Psychol. Med. 45(10), 2111–2122. https:// doi. org/ 10. 1017/ S0033 29171 50000 45 (2015).
18. Li, W. et al. Aberrant early visual neural activity and brain-behavior relationships in anorexia nervosa and body dysmorphic
disorder. Front. Hum. Neurosci. 9, 301. https:// doi. org/ 10. 3389/ fnhum. 2015. 00301 (2015).
19. Moody, T. D. et al. Functional connectivity for face processing in individuals with body dysmorphic disorder and anorexia nervosa.
Psychol. Med. 45(16), 3491–3503. https:// doi. org/ 10. 1017/ S0033 29171 50013 97 (2015).
20. Moody, T. D. et al. Abnormal brain activation and connectivity in anorexia nervosa and body dysmorphic disorder when viewing
bodies. Brain Imaging Behav. https:// doi. org/ 10. 1101/ 2020. 02. 12. 934083 (2020).
21. Ralph-Nearman, C. & Filik, R. New body scales reveal body dissatisfaction, thin-ideal, and muscularity-ideal in males. Am. J. Mens
Health 12(4), 740–750. https:// doi. org/ 10. 1177/ 15579 88318 763516 (2018).
22. Ralph-Nearman, C. & Filik, R. Development and validation of new gural scales for female body dissatisfaction assessment on two
dimensions: in-ideal and muscularity. BMC Public Health 20(1114), 2020. https:// doi. org/ 10. 1186/ s12889- 020- 09094-6 (2020).
23. Stunkard, A. J., Sørensen, T. & Schulsinger, F. Use of the danish adoption register for the study of obesity and thinness. Res. Publ.
Assoc. Res. Nerv. Ment. Dis. 60, 115–120 (1983).
24. Swami, V., Salem, N., Furnham, A. & Tovée, M. J. Initial examination of validity and reliability of the female photographic gure
rating scale for body image assessment. Personality Individ. Dier. 44(8), 1752–1761. https:// doi. org/ 10. 1016/j. paid. 2008. 02. 002
(2008).
25. Cornelissen, K. K., McCarty, K., Cornelissen, P. L. & Tovée, M. J. Body size estimation in women with anorexia nervosa and healthy
controls using 3D avatars. Sci. Rep. 7(1), 15773. https:// doi. org/ 10. 1038/ s41598- 017- 15339-z (2017).
26. Dickson-Parnell, B., Jones, M., Braddy, D. & Parnell, C. P. Assessment of body image perceptions using a computer program. Beh av.
Res. Methods 19(3), 353–354 (1987).
27. Letosa-Porta, A., Ferrer-García, M. & Gutiérrez-Maldonado, J. A. Program for assessing body image disturbance using adjustable
partial image distortion. Behav. Res. Methods 37(4), 638–643. https:// doi. org/ 10. 3758/ BF031 92734 (2005).
28. Schlundt, D. G. & Bell, C. Body image testing system: A microcomputer program for assessing body image. J. Psychopathol. Behav.
Assess. 15(3), 267–285 (1993).
29. Tovée, M. J., Benson, P. J., Emery, J. L., Mason, T. & Cohen-Tovée, E. M. Measurement of body size and shape perception in eating-
disordered and control observers using body-shape soware. Br. J. Psychol. 94(4), 501–516. https:// doi. org/ 10. 1348/ 00071 26033
22503 060 (2003).
30. Smeets, M. A., Smit, F., Panhuysen, G. E. & Ingleby, J. D. e inuence of methodological dierences on the outcome of body size
estimation studies in anorexia nervosa. Br. J. Clin. Psychol. 36(2), 263–277. https:// doi. org/ 10. 1111/j. 2044- 8260. 1997. tb014 12.x
(1997).
31. Ralph-Nearman, C. et al. A novel mobile tool (Somatomap) to assess body image perception pilot tested with fashion models and
nonmodels: Cross-sectional study. JMIR Mental Health 6(10), e14115. https:// doi. org/ 10. 2196/ 14115 (2019).
32. Sheehan, D. et al. MINI International Neuropsychiatric Interview-Version 7.0 (MINI 7.0) (Medical Outcomes Systems Inc., Jack-
sonville, 2015).
33. Markis, T. A. & McLennan, C. T. e eect of priming a thin ideal on the subsequent perception of conceptually related body
image words. Body Image 8(4), 423–426. https:// doi. org/ 10. 1016/j. bodyim. 2011. 05. 001 (2011).
34. Hendershot, T. et al. Using the PhenX toolkit to add standard measures to a study. Curr. Protoc. Hum. Genet. 86(1), 1–21. https://
doi. org/ 10. 1002/ 04711 42905. hg012 1s86 (2015).
35. Fairburn, C. G. & Beglin, S. Eating disorder examination questionnaire (6.0). In Cognitive Behavior erapy and Eating Disorders
(ed. Fairburn, C. G.) (Guilford Press, 2008).
36. Carter, J. C., Stewart, D. A. & Fairburn, C. G. Eating disorder examination questionnaire: Norms for young adolescent girls. Behav.
Res. er. 39(5), 625–632. https:// doi. org/ 10. 1016/ S0005- 7967(00) 00033-4 (2001).
37. Luce, K. H., Crowther, J. H. & Pole, M. Eating disorder examination questionnaire (EDE-Q): Norms for undergraduate women.
Int. J. Eat. Disord. 41(3), 273–276. https:// doi. org/ 10. 1002/ eat. 20504 (2008).
38. Arevian, A. C., O’Hora, J., Rosser, J., Mango, J. D. & Wells, K. B. Patient and provider cocreation of mobile texting apps to support
behavioral health: Usability study. JMIR Mhealth Uhealth 8(7), e12655 (2020).
39. Khalsa, S. S. et al. Interoceptive anxiety and body representation in anorexia nervosa. Front. Psychol. 9, 444. https:// doi. org/ 10.
3389/ fpsyt. 2018. 00444 (2018).
40. Khalsa, S. S. et al. e practice of meditation is not associated with improved interoceptive awareness of the heartbeat. Psycho-
physiology 57(2), e13479. https:// doi. org/ 10. 1111/ psyp. 13479 (2020).
41. Ralph-Nearman, C., Achee, M., Lapidus, R., Stewart, J. L. & Filik, R. A systematic and methodological review of attentional biases
in eating disorders: Food, body, and perfectionism. Brain Behav. 9(12), e01458. https:// doi. org/ 10. 1002/ brb3. 1458 (2019).
42. Higgins, T. E. Self-discrepancy: A theory relating self and aect. Psychol. Rev. 94(3), 319–340. https:// doi. or g/ 10. 1037/ 0033- 295X.
94.3. 319 (1987).
43. Mason, T. B. et al. Self-discrepancy theory as a transdiagnostic framework: A meta-analysis of self-discrepancy and psychopathol-
ogy. Psychol. Bull. 145(4), 372–389. https:// doi. org/ 10. 1037/ bul00 00186 (2019).
44. Via, E. et al. Self and other body perception in anorexia nervosa: e role of posterior DMN nodes. World J. Biol. Psychiatry. 19(3),
210–224. https:// doi. org/ 10. 1080/ 15622 975. 2016. 12499 51 (2016).
45. Vocks, S. et al. Dierential neuronal responses to the self and others in the extrastriate body area and the fusiform body area. Cogn.
Aect. Behav. Neurosci. 10(3), 422–429 (2010).
46. Suchan, B. et al. Reduced connectivity between the le fusiform body area and the extrastriate body area in anorexia nervosa is
associated with body image distortion. Behav. Brain Res. 241, 80–85. https:// doi. org/ 10. 1016/j. bbr. 2012. 12. 002 (2013).
Acknowledgements
We would like to thank Rachel Lapidus. M.S., Danielle Deville, M.S., Olivia Shadid, M.D., Alexandra Weindel,
B.S., Rachel Wirginis, D.O., Valerie Upshaw, R.N., B.S.N,, and Abigail Kimball, for assistance with data collection,
Beth Persac, L.M.F.T. and all of the clinicians, therapists and dieticians at the Laureate Eating Disorders Program
for participant coordination, Catherine Wilkerson, Joseph Mango, and Rajay Kumar for technical support, Shane
Nearman for graphic creation, and our human models who provided body scans to generate the 3D avatars.
Author contributions
C.R.N. wrote the main manuscript text. S.K.K worked on subsequent dras with C.R.N. C.R.N. performed
analyses and prepared the gures and tables. J.D.F., S.E.M. and A.C.A. provided edits and comments. M.S. and
S.C. helped with some data organization. All authors reviewed the manuscript.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:12262 | 
www.nature.com/scientificreports/
Funding
This work was supported by The William K. Warren Foundation (SSK). Additional support included
NIMH R01MH093676-02S1 (A.C.A.), NIMH K23MH112949 (SSK), NIGMS P20GM121312 (SSK), NIMH
R01MH105662-03S1 (JDF), and personal development funding (CRN). e content is solely the responsibility
of the authors and does not necessarily represent the ocial views of the National Institutes of Health.
Competing interests
A.C.A. is founder of Insight Health Systems, Arevian Technologies, and Open Science Initiative. A.C.A. devel-
oped the Chorus platform, which is licensed from the University of California Los Angeles to Insight Health
Systems. ere are no other competing interests to report for any authors.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 021- 90739-w.
Correspondence and requests for materials should be addressed to C.R.N.orS.S.K.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2021
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com