Procedures

Connected Speech Mastersheets

Maintain and periodically review the Connected Speech Data Quality smartsheets, which contain all updated data of all the connected speech procedures:

• These sheets provide an overview of the full Connected Speech workflow, from raw data to analysis-ready output.

• Supervisors ensure that there is no missing data, filling it out and sending reminders if necessary to the different teams to fill out all necessary information.

Recommendation: create filter per participant to check and fill out missing information

Open

R Dashboards and Data Visualization

• The Connected Speech Dashboard in R provides a real-time visualization of data completeness and progress using the data of the Spanish Connected Speech Data Analysis and Catalan Connected Speech Data Analysis.

  • Track project metrics such as: sample size for a specific study, available visit pairs (e.g., Pre–Follow-up, Pre–Post) and days between visits (for sanity checks) and for longitudinal studies.

• Supervisors use these dashboards to monitor trends, confirm data coverage, and identify outliers or missing entries.

  • When those errors are detected please refer to the REDCap reports and MADR participant smartsheet to check data and correct it in the MADR paticipant smartheet. If there’s clarification needed from clinicians or data entry please adress it in the R01_CS_Data_Processing teams channel tagging the clinicians.

• R project dashboard: https://utexas.box.com/s/96jez19c5hdh2w2rpov4w7addvt95pw1

REDCap Data Monitoring

• Use predefined REDCap reports to track Connected Speech data status and ensure synchronization with SmartSheets. These reports should be reviewed to identify:

  • Missing or incomplete Connected Speech data.

  • Inconsistent visit dates or timepoints.

• Core reports include:

  • BACC

    • Connected Speech Data Control SPANISH

    • Connected Speech Data Control CATALA

    • MID- Connected Speech Data

  • DellMed - Still pending to create

SmartSheet Data Monitoring

• Use the MADR Participant SmartSheet as the master tracking sheet with respect to decisions about participant inclusion in connected speech smartsheets.

  • Clinicians update the sheet after each visit or follow-up to indicate:

    • Participant status (timepoints completed, pending, withdrawn, DNQ (Do Not Qualify), Corrections in timepoints or visit labels

• Supervisors cross-reference this SmartSheet to ensure all data is reflected consistently across systems.

• Check here participant naming convention: https://cloud.wikis.utexas.edu/wiki/x/IrXhB

Quality Control and Review - Connected Speech Supervisor

• Weekly: Review REDCap reports and SmartSheets for missing or inconsistent data.

• Monthly: Audit R dashboard summaries and confirm data integrity across projects.

• Quarterly (or before the start of each cohort): checking the progress of the different processes, use smartsheets reports to visualize progress (clipping, whisper, transcription, reliability)

Procedure When a Connected Speech Sample Is Missing or Data Are Inconsistent

1. Checking Available Data

• Verify expected timepoints using the MADR participant smartsheet and:

  • assessment timeline (in Participants folder)

  • REDCap (Connected Speech instrument).

• Check Connected Speech Smartsheets for existing rows and confirm that each row corresponds to an actual recording. (! if there’s no audio or video or transcription available we will not have a row in the connected speech smartsheet)

• Check in Box for the participants video/audio:

  • the Connected Speech Data Raw folder

  • the ALL IP SESSIONS folder (this is a backup folder for all the files, all the clinicians upload the audios-videos here and then they copy them to Connected Speech Data Raw)

  • the participant’s Box folder

• Identify any mismatches (row without file, file without row, wrong date, etc.).

2. Correct the mismatches: Copying Files if They Exist

• If the recording exists but is misplaced or mislabeled, copy it to the correct timepoint folder and rename using CS conventions.

• If a file exists but there is no Smartsheet row, add the row.

• If the Smartsheet row exists but the file does not, delete the row.

3. Adding Notes and Documenting Inconsistencies

• In Smartsheet: Add comments explaining added/removed rows or missing recordings.

• In REDCap: Add a note in the Connected Speech instrument indicating missing or inconsistent data.

• In Box: Ensure final folder structure and naming reflect the corrected timepoint status.

Data Analysis Tasks

For assigning tasks we use student supervisor chat.

Recurrent tasks:

• Create new filters for new participants to help visualize them in smartsheet report, eg.

Open

Assigned Tasks:

@Jada Li add exact date of administration for all pre-r01 CS samples starting from post, pre, 3m, 6m, 12m. Decision to start with post to make quick progress with inclusion of participants for longitudinal picnic scene ppa study, but we want the exact dates for the 3 tasks - important event, cat rescue and picnic scene.

@Jada Li create 3 columns in English Smartsheet for dates of picnic scene, cat rescue and important event Sheet

Future RA Tasks (not assigned yet):

• High priority:

• Mid priority:

• Low priority:

update Connected Speech Detailed overview and Box folder structure Connected Speech Detailed overview and Box folder structurePreview

create and get familiar with Box participant folder structure, we used to have a tree scheme in box, start from that and create a whiteboard in box

Completed Tasks