PCC: Patient Consultation Corpus

The Patient Consultation Corpus (PCC) is a multimodal dataset of simulated consultations between a patient and multiple healthcare professionals. Developed within the EU-funded Council of Coaches (COUCH) project, it supports research into multiparty dialogue, healthcare communication, verbal and non-verbal interaction, and conversational systems. The corpus contains seven main consultations and two pilot consultations, recorded across three days. Actors portray patients with different interpersonal and engagement styles, interacting with healthcare professionals. The consultations are simulated encounters, rather than recordings of patients receiving clinical care. Contents Each consultation includes: A mixed audio recording in WAV format. A corresponding transcript in Microsoft Word format. Video recordings of the interaction. The main consultations include participant-focused video recordings, typically providing head and full-body views. The two pilot consultations each use a single camera covering all participants. Recordings are supplied in their original formats, including MTS, MPG, MP4 and MOV. Sessions are identified by day and consultation number: D0.C1–D0.C2 for the pilots, D1.C1–D1.C4 for the first main recording day, and D2.C1–D2.C3 for the second. Download organisation The deposit contains approximately 112.6 GB of data. Audio recordings and transcripts are provided in separate archives, allowing researchers to download these modalities without downloading the video. Video archives are organised by consultation and, for the main sessions, by participant or shared camera stream. The archives preserve the original corpus directory structure and filenames. The README provides a session overview, an archive index and guidance on reconstructing the directory structure. A SHA-256 checksum manifest is supplied for verifying downloaded files. Recording notes Video files may include material before or after the transcribed consultation. Some camera recordings are split across multiple files. During D2.C3, a camera power failure interrupted part of Colin’s video recording. The affected files are marked as incomplete, and a supplementary, lower-quality recording from a room-view camera is included. Further details are provided in the accompanying README. Access and reuse Files are available on request through Zenodo. Applicants should briefly describe their intended research and acknowledge the confidentiality conditions stated in the access-request information, reflecting the commitments made to participants. The corpus is distributed under the Creative Commons Attribution 4.0 International licence (CC BY 4.0). Participant consent and confidentiality information should also be considered when using and disseminating material from the corpus. Please cite this dataset and the associated paper in research outputs using the PCC: Snaith, M., Conway, N., Beinema, T., De Franco, D., Pease, A., Kantharaju, R., Janier, M., Huizing, G., Pelachaud, C., & op den Akker, H. (2021). A multimodal corpus of simulated consultations between a patient and multiple healthcare professionals. Language Resources and Evaluation, 55, 1077–1092. https://doi.org/10.1007/s10579-020-09526-0 Funding and provenance The corpus was developed as part of the Council of Coaches project, funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 769553. This deposit provides a new distribution location for the previously released corpus, retaining the original recordings, transcripts and filenames.