You are here

  • Home
  • Archive
  • Volume 49, Issue Suppl 1
  • EP128 One-month pain recovery patterns after total knee arthroplasty: two distinct patient groups identified by an unsupervised learning algorithm

Article Text

Article menu
Download PDFPDF
ePoster session 4 – Station 1
EP128 One-month pain recovery patterns after total knee arthroplasty: two distinct patient groups identified by an unsupervised learning algorithm
  1. Haoyan Zhong1,
  2. Schindler Melanie2,
  3. Park Jiwoo3,
  4. Yendluri Avanish3,
  5. Crispiana Cozowicz4,
  6. Jiabin Liu1,5,
  7. Stavros Memtsoudis5,6 and
  8. Jashvant Poeran7
  1. 1Department of Anesthesiology, Critical Care and Pain Management, Hospital for Special Surgery, New York, USA
  2. 2Department of Orthopedics, University Hospital Regensburg, Regensburg, Bavaria, Germany
  3. 3Icahn School of Medicine at Mount Sinai, New York, USA
  4. 4Department of Anesthesiology, Perioperative Medicine and Intensive Care Medicine, Paracelsus Medical University, Salzburg, Austria
  5. 5Department of Anesthesiology, Weill Cornell Medicine, New York, USA
  6. 6Department of Anesthesiology, Critical Care and Pain Management, Hospital for Special Surgery, New York, USA
  7. 7Institute for Healthcare Delivery Science, Department of Population Health Science and Policy/Department of Orthopedics/Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, USA

Abstract

Please confirm that an ethics committee approval has been applied for or granted: Yes: I’m uploading the Ethics Committee Approval as a PDF file with this abstract submission

Background and Aims While much research exists on patterns of pain scores in the perioperative period, much less data exists on longer term follow-up of pain scores, especially after total knee arthroplasty (TKA). Re-using data from a published prospective study capturing patients‘ pain scores recorded over 29 days post-TKA, we aimed to assess whether unsupervised machine learning can discern distinct postoperative recovery patterns.

Methods This study was approved by an Institutional Review Board as it re-used data from a published study that prospectively enrolled 103 patients undergoing primary TKA (2020-2021) at a single university hospital. Patients recorded daily numeric rating scale pain scores at morning, lunchtime, evening, and nighttime for 29 days. A K-Means clustering algorithm (unsupervised) was applied to identify distinct pain recovery patterns after which the identified recovery groups were compared based on available patient and surgical characteristics.

Results Two clusters of patients with distinct recovery patterns were discovered: patients in Cluster 1 (versus Cluster 2) had higher pain levels throughout the recovery period (figure 1); Cluster 1 also represented patients that were more likely female and with higher Knee Society Scores (KSS) at both week 1 and 4 post-TKA, as well as a higher KSS Functional Score at week 1. (table 1)

Abstract EP128 Figure 1
Abstract EP128 Figure 1

K-Means clustering analysis applied to daily average and maximum NRS pain

View this table:
Abstract EP128 Table 1

Patient and procedural characteristics stratified by clustering

Conclusions Machine learning algorithms applied to longitudinal pain level data have identified two distinct postoperative recovery patterns after TKA. Notably, patients who experienced higher pain levels early postoperatively exhibited consistently higher pain levels later within the first month of recovery.

  • total knee arthroplasty
  • pain
  • recovery pattern.

https://doi.org/10.1136/rapm-2024-ESRA.201

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.