ATOM-HP DATA README

This files provides information on the ATOM-HP dataset comprising summaries and per-subject data files in CSV format.

Obtaining the ATOM-HP dataset

The complete dataset is available at https://br-atomhp.usc.edu (includes data from PDMS website - VPN required)

The data collected through the PDMS website is available at https://pdms.usc.edu/export. These are options to get a local copy.

cd data
./run_rsync.sh

Bulk Download using Filezilla

With Filezilla, sftp to pdms.usc.edu as atom-hp then navigate to /var/www/html/export.

Bulk Download using scp

cd data
scp -r atom-hp@pdms.usc.edu:/var/www/html/export/* .

Bulk Download using wget

cd data
wget -r --no-parent -nH --cut-dirs=1 --reject "index.html*"  --user=atom-hp --password=<PASSWORD> https://pdms.usc.edu/export/

Studies

Clinical

Kinect1    Band1      Chemo1  Kinect2    Chemo2         Band2               Followup150

  |--------|------------|-------|----------|-------------|----------------------|

Subject ids for clinical study are 001-060 and 121-150.

CRF (Case Report Form) are stored in RedCap.

Visit1: date of visit 1 corresponding to Kinect1 recordings (date of "Visit 1 CRF")

Visit2: date of visit 2 corresponding to Kinect2 recordings (date of "Visit 2 CRF")

Kinect1: Kinect file recording date and time of first recording

Kinect2: Kinect files recording date and time of second recording

Band1: day band data collection starts (date of "screening CRF")

Band2: day band data collection must stop (on day 60 of study)

Chemo1: date of first chemo

Chemo2: date of second chemo

Followup150: day of follow-up (on day 150 of study)

Band data records made during [Chemo1, Chemo2].

Kinect recordings made during [Visit1, Visit2]. Note that Visit1, Visit2 may differ from Chemo1, Chemo2.

Military

       Day 1        Day 2        Day 3        Day 4       Day 5      
                                                      Exercise
  |------------|------------|--o---------|------------|-*****-o----|
                            Kinect1                        Kinect2

Subject IDs for military study are 061-120.

Band data is collected over 5 days.

Kinect1 recordings are made on day 3, 7:50 - 9:10 am.

Kinect2 recordings are made on day 5, 11:40 am - 12:40 pm, after the exercise.

Data files

Military test file

Some test files for Military tasks are in /u01/MilTests20170615 on PDMS server. - Note: the Talk task is not the fully correct setup since we decided to count backward instead of reading text. However, all other setup such as standing position, camera angle, ... should be correct.

Summary data files

CFT and demographics

Below is the outdated data from the pilot study. There is a new file ATOM-HP_military_data_20170719.xlsx that has not been processed yet.

mil_cases_151216.csv contains military cases measurements corresponding to the data in MilitaryRawData_15DEC16.xlsx. Measuremnts contain demographics and CFT assessment scores.

id
age: in years
service_time_mos: service time military occupational specialty
mos:subjects military occupational specialty (i.e. mortarman, rifleman, radio operator, etc...)
rank:
prior_cft_score: [0, 300]
marksman: 1=expert > 2=sharp shooter > 3=marksman, 4=NA or not reported
deployments:
ethnicity: 1= asian, 2= black non-hispanic, 3= caucasian non-hispanic, 4= hispanic, 5= native american/eskimo, 6= pacific islander/hawaiian, 7= other
height: (in)
weight: (lbs)
cigarette_use: 1= never smoked, 2= experimented with smoking, 3= occasional smoker, 4= daily smoker, 5= former smoker
chewing_tobacco_use: 1= never used smokeless tobacco, 2= experimented with smokeless tobacco, 3= occasional user, 4= daily user, 5= former user
pre_test_a: (s)
pre_test_b: (s)
post_test_a: (s)
post_test_b: (s)
pre_marksmanship_score: [0, 50]
pre_marksmanship_group_size: (cm)
pre_marksmanship_mpiczp: Mean point of impact from the center of the target. Theoretically shots are aimed at the center of the target so it demonstrates how far off the group size is from the center of the target. (cm)
post_marksmanship_score: [0, 50]
post_marksmanship_group_size: (cm)
post_marksmanship_mpiczp: (cm)
cft_run_time: (s)
cft_manuf_time: maneuver under fire time (s)
cft_lift_count: number of hammo can lift performed
cft_runs_score: [0, 100]
cft_manufs_score: maneuver under fire score [0, 100]
cft_lift_score: of hammo can lift score [0, 100]
cft_score: total CFT score (cft_lift_score + cft_runs_score + cft_manufs_score) [0, 300]

Per-subject files

Nomenclature

Per subject files are named as follows:

P###_DATA_TYPE_VISIT_#_TASK_#_REP_#_[XRF_FILENAME].csv

P###: subject id, military indices are 061-120, the other are all clinical.

DATA_TYPE:

VISIT_#: 1 for Visit1, 2 for Visit2

TASK_#: 1 for ChairToTable, 2 for GetUpAndWalk, 3 for WalkAndTalk, 4 for Walk, 5 for Talk

REP_#: repeated measurements ordered by time (starts at 1)

BAND

 email: account email
 start_time: start time of the summary
 end_time: end time of the summary
 period: period of the summary (1: daily, 2: hourly)
 steps: The total of steps taken in the time period 
 active_hours: Number of active hours in the period, used for Daily summary
 total_calories: Total calories burned in the period
 average_heart_rate: Average heart rate during the period
 peek_heart_rate: Peak heart rate during the period
 lowest_heart_rate: Lowest heart rate during the period
 total_distance: Total distance in the period
 total_distance_on_foot: Total distance covered on foot
 actual_distance: Absolute distance, including any paused time
 elevation_gain: Cumulative elevation gain during the period
 elevation_loss: Cumulative elevation loss during the period
 max_elevation: Maximum elevation during the period
 min_elevation: Minimum elevation during the period
 way_point_distance: Distance in cm used to waypoint the GPS data
 speed: Total Period distance divided by period duration
 pace: Period duration divided by total period distance
 overall_pace: Duration of all periods divided by distance of all periods

KINECT_METADATA

visit_number: visit number (eg. visit 1, visit 2.)
filename: Kinect recording raw file name
file_length: Kinect recording raw file size (in Byte)
creation_time: created/accessed time in recording file properties 
email: account email
comment: comments from the coordinator
success: KinectVerifier recording validation code: 1 is Pass, 2 is Fail
task: 1 is ChairToTable (clinical), 2 is GetUpAndWalk (clinical), 3 is WalkAndTalk (military), 4 is Walk (military), 5 is Talk (military)

KINECT_SKELETON

See https://msdn.microsoft.com/en-us/library/dn799273.aspx

unix_timestamp: Unix timestamp of recording replaying/data extraction time
frame_number:   frame index
spine_base_x, spine_base_y, spine_base_z, spine_mid_x, spine_mid_y, spine_mid_z, neck_x, neck_y, neck_z, head_x, head_y, head_z, shoulder_left_x, shoulder_left_y, shoulder_left_z, elbow_left_x, elbow_left_y, elbow_left_z, wrist_left_x, wrist_left_y, wrist_left_z, hand_left_x, hand_left_y, hand_left_z, shoulder_right_x, shoulder_right_y, shoulder_right_z, elbow_right_x, elbow_right_y, elbow_right_z, wrist_right_x, wrist_right_y, wrist_right_z, hand_right_x, hand_right_y, hand_right_z, hip_left_x, hip_left_y, hip_left_z, knee_left_x, knee_left_y, knee_left_z, ankle_left_x, ankle_left_y, ankle_left_z, foot_left_x, foot_left_y, foot_left_z, hip_right_x, hip_right_y, hip_right_z, knee_right_x, knee_right_y, knee_right_z, ankle_right_x, ankle_right_y, ankle_right_z, foot_right_x, foot_right_y, foot_right_z, spine_shoulder_x, spine_shoulder_y, spine_shoulder_z, hand_tip_left_x, hand_tip_left_y, hand_tip_left_z, thumb_left_x, thumb_left_y, thumb_left_z, hand_tip_right_x, hand_tip_right_y, hand_tip_right_z, thumb_right_x, thumb_right_y, thumb_right_z: skeleton joints coordinates

KINECT_FACE_PROPERTIES (military only)

See https://msdn.microsoft.com/en-us/library/dn782034.aspx

unix_timestamp: from replay
frame_number: first frame corresponds to the first frame in which the face is detected in the recording
happy: The user appears to be smiling and is putting forward a happy expression. This will also pick up smiles during laughter. It’s important to note that some users appear happy even when they are not, so this should not be considered an exact translation to emotion.
wearing_glasses: The user is wearing glasses.
left_eye_closed: The user's left eye is closed.
right_eye_closed: The user's right eye is closed.
mouth_open: The user's mouth is open.
mouth_moved: The user's mouth moved. This is the only property that needs frame over frame results to make an accurate determination. This feature works best if the user's head is mostly still.

KINECT_FACE_MESH (military only)

unix_timestamp: from replay
frame_number: first frame corresponds to the first frame in which the face is detected in the recording with its corresponding x,y and z coordinates.
 

KINECT_FACE_ROTATION (military only)

See https://msdn.microsoft.com/en-us/library/microsoft.kinect.face.faceframeresult.facerotationquaternion.aspx For conversion from quaternions to angles, refer : https://social.msdn.microsoft.com/Forums/en-US/245a3a09-a2e4-4e0e-8c12-b8625102376a/kinect-v2-sdk-joint-orientation?forum=kinectv2sdk To understand the orientation details, refer https://msdn.microsoft.com/en-us/library/jj130970.aspx

unix_timestamp: from replay
frame_number: first frame corresponds to the first frame in which the face is detected in the recording with its corresponding w,x,y and z quaternion coordinates.

KINECT_FACE_ANIMATION_UNITS (military only)

See https://msdn.microsoft.com/en-us/library/microsoft.kinect.face.faceshapeanimations.aspx To understand and analyze the output of this file, refer https://msdn.microsoft.com/en-us/library/jj130970.aspx

unix_timestamp: from replay
frame_number: first frame corresponds to the first frame in which the face is detected in the recording.
jaw_open
jaw_slide_right
left_cheek_puff
left_eyebrow_lowerer
left_eye_closed
lip_corner_depressor_left
lip_corner_depressor_right
lip_corner_puller_left
lip_corner_puller_right
lip_pucker
lip_stretcher_left
lip_stretcher_right
lower_lip_depressor_left
lower_lip_depressor_right
right_cheek_puff
right_eyebrow_lowerer
right_eye_closed

KINECT_HIGH_DETAIL_FACE_POINTS (military only)

See https://msdn.microsoft.com/en-us/library/microsoft.kinect.face.highdetailfacepoints.aspx

unix_timestamp: from replay
frame_number: first frame corresponds to the first frame in which the face is detected in the recording.
chin_center: corresponding x,y,z coordinates
forehead_center: corresponding x,y,z coordinates
left_cheek_bone: corresponding x,y,z coordinates
left_cheek_center: corresponding x,y,z coordinates
left_eyebrow_center: corresponding x,y,z coordinates
left_eyebrow_inner: corresponding x,y,z coordinates
left_eyebrow_outer: corresponding x,y,z coordinates
left_eye_inner_corner: corresponding x,y,z coordinates
lower_jaw_left_end: corresponding x,y,z coordinates
lower_jaw_right_end: corresponding x,y,z coordinates
mouth_left_corner: corresponding x,y,z coordinates
mouth_lower_lip_mid_bottom: corresponding x,y,z coordinates
mouth_lower_lip_mid_top: corresponding x,y,z coordinates
mouth_right_corner: corresponding x,y,z coordinates
mouth_upper_lip_mid_bottom: corresponding x,y,z coordinates
mouth_upper_lip_mid_top: corresponding x,y,z coordinates
nose_bottom: corresponding x,y,z coordinates
nose_bottom_left: corresponding x,y,z coordinates
nose_bottom_right: corresponding x,y,z coordinates
nose_tip: corresponding x,y,z coordinates
nose_top: corresponding x,y,z coordinates
nose_top_left: corresponding x,y,z coordinates
nose_top_right: corresponding x,y,z coordinates
right_cheek_bone: corresponding x,y,z coordinates
right_cheek_center: corresponding x,y,z coordinates
right_eyebrow_center: corresponding x,y,z coordinates
right_eyebrow_inner: corresponding x,y,z coordinates
right_eyebrow_outer: corresponding x,y,z coordinates
right_eye_inner_corner: corresponding x,y,z coordinates
right_eye_mid_bottom: corresponding x,y,z coordinates
right_eye_mid_top: corresponding x,y,z coordinates
right_eye_outer_corner: corresponding x,y,z coordinates

Detection Result Description:

Yes: We are very certain that the property is true and you could take a destructive action on this result.
No: We are very certain that the property is false and you could take a destructive action on this result.
Maybe: We are pretty sure that the property is true. You could reward the user or give a positive action for this result. For most properties, you could potentially infer from this result that the corresponding movement was small.
Unknown: We don’t have enough information to make a determination. This typically occurs because the user has orientated their face away from the sensor and we don’t want to give a bad result.

PRO (clinical only)

Patient reported outcomes (PRO) data. The PRO scales are using come from the NIH PROMIS measures: http://www.healthmeasures.net/explore-measurement-systems/promis. See details in ATOM_HP_PRO.doc on how to use the PRO scale. We adapted the fatigue short form for use in daily measurement.

References: Junghaenel, Doerte U., et al. "Identification of distinct fatigue trajectories in patients with breast cancer undergoing adjuvant chemotherapy." Supportive Care in Cancer 23.9 (2015): 2579-2587.

Christodoulou, Christopher, et al. "Measuring daily fatigue using a brief scale adapted from the Patient-Reported Outcomes Measurement Information System (PROMIS®)." Quality of Life Research 23.4 (2014): 1245-1253.

study_id
day
Chemo
FATEXP16,FATEXP18,FATEXP18_w,FATEXP20,FATEXP20_w,FATEXP5,FATEXP5_w,FATEXP6,FATIMP21_w,FATIMP3,FATIMP30,FATIMP30_w,FATIMP33,FATIMP33_w,FATIMP40_w,IsoCaPS1,IsoCaPS2,PFA01,PFA03,PFA05,PFA11,PFA16,PFA55,PFB26,PFC36,PFC37,PFC45,SDS001,Sleep108,Sleep109,Sleep110,Sleep115,Sleep116,Sleep44,Sleep87,Sleep90,UCLA11x2,UCLA13x3,UCLA14x2,UCLA18x2: PRO 
Weight

REDCAP (clinical only)

Does not make sense to exported alphabetically!

study_id,
adv_events_v2
age
ante_weight_loss_10per_v1
ante_weight_loss_5per_v1
antiemetic_plan_v1
antiemetic_plan_v2
baseline_adv_events_v1
blood_press_v1
blood_press_v2
bmi_v1
bmi_v2
cancer_diagnosis
changes_chemo_plan_v2
chemo_day1_date_v2_1
chemo_day1_date_v2_2
chemo_day1_date_v2_3
chemo_day1_date_v2_4
chemo_intent
chemo_plan
clinical_site
concom_meds_v1
concom_meds_v2
coord_ecog_v1
coord_ecog_v2
coord_ecog_v3
coord_karnofsky_v1
coord_karnofsky_v2
datacol_start_date
datacol_stop_date
date_cancer_diag
day_followup_crf_complete
dayhosp_visits_day1_30_v3
dayhosp_visits_day31_60_v3
dayhosp_visits_day61_150_v3
dhospvis_day1_30_n_v3
dhospvis_day31_60_n_v3
dhospvis_day61_150_n_v3
dis_stat_2rounds_v3
dis_stat_study_start_v3
draw_comments
draw_date
draw_time
education
ethnicity
flow_rate_v1
flow_rate_v2
gender
grad_degree
heart_rate_v1
heart_rate_v2
height_units_v1
height_units_v2
height_v1
height_v2
hosp_day1_30_n_v3
hosp_day31_60_n_v3
hosp_day61_150_n_v3
hospital_day1_30_v3
hospital_day31_60_v3
hospital_day61_150_v3
inspired_o2_v1
inspired_o2_v2
marital_status
occupation
oxygen_sat_v1
oxygen_sat_v2
patient_ecog_v1
patient_ecog_v2
patient_ecog_v3
patient_karnofsky_v1
patient_karnofsky_v2
phys_ecog_v1
phys_ecog_v2
phys_ecog_v3
phys_karnofsky_v1
phys_karnofsky_v2
phys_visits_day1_30_v3
phys_visits_day31_60_v3
phys_visits_day61_150_v3
physvis_day1_30_n_v3
physvis_day31_60_n_v3
physvis_day61_150_n_v3
present_stage
presently_employed
rad_therapy_notes_v2
rad_therapy_v2
resp_rate_v1
resp_rate_v2
screening_crf_complete
screening_date
stage_at_diag
visit1_date
visit2_date
visit_1_crf_complete
visit_2_crf_complete
weight_units_v1
weight_units_v2
weight_v1
weight_v2